Three bugs from teknium1's successful install + diagnostic chat on Windows:
1. **Start-Process -FilePath npm.cmd fails with "%1 is not a valid Win32
application".** Start-Process bypasses cmd.exe and PATHEXT to call
CreateProcessW directly, which refuses .cmd batch shims. Switched
Install-NodeDeps to use PowerShell's invocation operator (``& $npmExe
install --silent *> $log``) which DOES honour PATHEXT. Extracted a
``_Run-NpmInstall`` helper so the browser + TUI paths share the same
logic. Captures $LASTEXITCODE correctly, still surfaces the real
stderr on failure with a log-file pointer for the full output.
2. **patch tool returns false-negative on Windows due to CRLF round-trip.**
Root cause was upstream of patch: ``subprocess.Popen(..., text=True,
stdin=PIPE)`` on Windows translates ``\\n`` → ``\\r\\n`` when data flows
through the stdin pipe. ``_pipe_stdin()`` was writing the patch's
new_content string through a text-mode pipe, bash then wrote those
CRLF bytes to disk, and patch's post-write verify compared the
on-disk CRLF bytes against the original LF-only string — fail.
Fixed in two places for defense in depth:
- ``_pipe_stdin()`` now writes through ``proc.stdin.buffer`` with
explicit UTF-8 encoding, bypassing Python's newline translation on
every platform. No behaviour change on POSIX (bytes are identical)
but stops the CRLF injection on Windows.
- ``patch_replace``'s post-write verify normalizes CRLF→LF on both
sides before comparing, so even if some future backend still
translates newlines the patch tool won't report a bogus failure.
3. **SOUL.md gets a UTF-8 BOM on Windows PowerShell 5.1.** ``Set-Content
-Encoding UTF8`` on PS5.1 writes UTF-8 WITH a byte-order-mark (changed
in PS7 via ``utf8NoBOM``). Hermes's prompt-injection scanner sees
the BOM (U+FEFF invisible char) and refuses to load the file, so
SOUL.md's persona instructions never get applied.
Fixed by writing the file via ``[System.IO.File]::WriteAllText``
with an explicit ``UTF8Encoding($false)`` — BOM-free on every
PowerShell version.
All POSIX behaviour verified unchanged: 198 tests pass across
test_file_operations, test_local_env_cwd_recovery, test_code_execution,
test_windows_native_support, test_windows_compat.
Three interrelated bugs from teknium1's first interactive chat on Windows:
1. **Snapshot/cwd file paths unquoted in bash command strings.** The session
bootstrap and per-command wrapper interpolated
``self._snapshot_path`` / ``self._cwd_file`` unquoted into bash commands
like ``export -p > C:/Users/ryanc/.../hermes-snap-xxx.sh``. Git Bash's
MSYS2 layer handles ``C:/...`` paths correctly ONLY when quoted; unquoted,
the colon and forward-slash get glob-parsed and the redirect targets a
bogus path. Symptom: every terminal command emitted two
``C:/Users/.../hermes-snap-*.sh (No such file or directory)`` lines that
bled into stdout (``stderr=STDOUT`` on the local backend) and corrupted
file contents when the agent wrote to scratch paths via the terminal
tool. Fix: ``shlex.quote()`` every interpolation of ``_snapshot_path``
and ``_cwd_file`` in base.py — no-op on POSIX (the paths contain no
shell-metachars), critical on Windows.
2. **Stale PATH on first hermes launch after install.** ``install.ps1``
adds the PortableGit ``cmd`` / ``bin`` / ``usr\bin`` directories to the
Windows **User** PATH via ``SetEnvironmentVariable(..., "User")``. That
write propagates to newly *spawned* processes only — already-running
shells (including the one the user types ``hermes`` into immediately
after install) retain their old PATH. So hermes starts with a PATH that
doesn't include bash, rg, grep, ssh — and ``search_files`` reports
"rg/find not available" when the user clearly just installed them.
Fix: new ``_augment_path_with_known_tools()`` helper called from
``configure_windows_stdio()`` on startup. Prepends the Hermes-managed
Git directories + the WinGet Links directory (where ripgrep lands) to
``os.environ['PATH']`` if they exist on disk but aren't already in
PATH. Subsequent subprocess calls (including bash spawns via
``_find_bash()``) inherit the augmented PATH and find everything.
No-op on POSIX and when the directories don't exist.
3. **Root cause of "file content corruption".** #1 was the proximate cause.
Errors like ``C:/Users/.../hermes-snap-xxx.sh: No such file or directory``
were emitted on stderr by the failed redirect, captured into stdout via
``stderr=subprocess.STDOUT``, and if the agent used terminal commands
like ``cat > file`` the leaked error bytes became part of the file.
Fixing #1 eliminates this entirely.
## Tests
All 77 Windows-compat tests still pass on Linux (POSIX path is
shlex.quote('/tmp/foo.sh') → '/tmp/foo.sh' — unchanged).
## Not addressed here (would need a bigger design)
- Python file tools (``write_file``, ``read_file``) and the bash-backed
terminal tool see DIFFERENT views of ``/tmp`` on Windows. Python treats
``/tmp`` as ``C:\tmp`` (drive-relative), Git Bash's MSYS2 treats it as
a virtual mount to the PortableGit install's ``tmp\``. Would need a
translation shim in the Python tools to resolve bash-virtual paths to
their native-Windows equivalents. Workaround for users today: use
absolute native paths (``C:\Users\you\...``) instead of ``/tmp/...``
when crossing between terminal and Python file tools.
Three real bugs from teknium1's first Windows install run:
1. **MinGit has no bash.exe.** MinGit is the minimal-automation Git for Windows
distribution — it ships git.exe but deliberately strips bash and the POSIX
coreutils. Installer logged "Could not locate bash.exe" and Hermes would
fail to run any shell command. Switched to PortableGit — the full Git for
Windows minus the installer UI. PortableGit ships bash.exe at
<root>\bin\bash.exe plus sh, awk, sed, grep, curl, ssh in usr\bin\. ARM64
variant is detected separately (PortableGit-*-arm64.7z.exe). 32-bit falls
back to MinGit-32-bit with a warning (PortableGit is 64-bit only).
PortableGit ships as a 7z self-extractor (56MB vs MinGit's 38MB). We
invoke it with `-o<target> -y` to extract silently — no 7z install needed,
it's self-contained.
Updated tools/environments/local.py::_find_bash candidate order to prefer
the PortableGit layout (<root>\bin\bash.exe) with the MinGit layout
(<root>\usr\bin\bash.exe) as a fallback so existing installs keep working.
2. **os.execvp "Exec format error" on Windows.** Setup wizard's "Launch
hermes chat now? Y" called `os.execvp(["hermes", "chat"])` which on
Windows can only swap to real Win32 .exe files — chokes with OSError(8)
on .cmd batch shims and Python console-script wrappers. Added a
win32 branch in hermes_cli/relaunch.py::relaunch() that uses
subprocess.run + sys.exit — functionally identical (user sees "hermes
exited, then new hermes started") with one extra PID in play. POSIX
path is UNCHANGED — still uses os.execvp for in-place replacement.
Catches OSError in the Windows branch and surfaces a "open a new
terminal so PATH picks up, then re-run hermes" hint instead of a
cryptic traceback.
3. **npm install failures silent on Windows.** The install.ps1 was invoking
`npm install --silent 2>&1 | Out-Null` inside a try/catch. PowerShell's
try/catch does NOT trigger on non-zero process exit codes — only on
unhandled .NET exceptions — so npm failing printed a generic "npm
install failed" with zero information about WHY. The silent pipe ate
the stderr.
Rewrote Install-NodeDeps to:
- Resolve npm.cmd via Get-Command (respects PATHEXT) instead of
relying on bare `npm` name resolution.
- Use Start-Process with -PassThru to capture the actual exit code.
- Redirect stderr to a temp log and surface the first ~800 chars of
the real npm error when install fails, plus the log path for the
full text.
- Fail loudly with the right exit code instead of a misleading success.
- Bail cleanly with a helpful message when npm isn't on PATH at all.
4. **"True" printing to console after Node check.** `Test-Node` returns $true;
installer called it as a bare statement (no assignment, no cast). PowerShell
prints bare return values. Wrapped the call in `[void](Test-Node)`.
## Tests
- Added 3 new tests in tests/hermes_cli/test_relaunch.py covering the
Windows branch: subprocess is called (not execvp), child exit code
propagates, OSError surfaces a helpful message. All 23 tests pass
(20 existing + 3 new).
- 77 Windows-compat tests still pass, POSIX behaviour unchanged.
Second pass on native Windows support, driven by a systematic audit across
five areas: POSIX-only primitives (signal.SIGKILL/SIGHUP/SIGPIPE, os.WNOHANG,
os.setsid), path translation bugs (/c/Users → C:\Users), subprocess patterns
(npm.cmd batch shims, start_new_session no-op on Windows), subsystem health
(cron, gateway daemon, update flow), and module-level import guards.
Every change is platform-gated — POSIX (Linux/macOS) behaviour is preserved
bit-identical. Explicit "do no harm" test: test_posix_path_preserved_on_linux,
test_posix_noop, test_windows_detach_popen_kwargs_is_posix_equivalent_on_posix.
## New module
- hermes_cli/_subprocess_compat.py — shared helpers (resolve_node_command,
windows_detach_flags, windows_hide_flags, windows_detach_popen_kwargs).
All no-ops on non-Windows.
## CRITICAL fixes (would crash or silently break on Windows)
- tui_gateway/entry.py: SIGPIPE/SIGHUP referenced at module top level would
AttributeError on import on Windows, breaking `hermes --tui` entirely (it
spawns this module as a subprocess). Guard each signal.signal() call with
hasattr() and add SIGBREAK as Windows' SIGHUP equivalent.
- hermes_cli/kanban_db.py: os.waitpid(-1, os.WNOHANG) in dispatcher tick was
unguarded. os.WNOHANG doesn't exist on Windows. Gate the whole reap loop
behind `os.name != "nt"` — Windows has no zombies anyway.
- tools/code_execution_tool.py: AF_UNIX socket for execute_code RPC fails on
most Windows builds. Fall back to loopback TCP (AF_INET on 127.0.0.1:0
ephemeral port) when _IS_WINDOWS. HERMES_RPC_SOCKET env var now accepts
either a filesystem path (POSIX) or `tcp://127.0.0.1:<port>` (Windows).
Generated sandbox client parses both.
- cron/scheduler.py: `argv = ["/bin/bash", str(path)]` hardcoded. Use
shutil.which("bash") so Windows (Git Bash via MinGit) works, with a
readable error when bash is genuinely absent.
- 6 bare npm/npx spawn sites: tools_config.py x2, doctor.py, whatsapp.py
(npm install + node version probe), browser_tool.py x2. On Windows npm
is npm.cmd / npx is npx.cmd (batch shims); subprocess.Popen(["npm", ...])
fails with WinError 193. shutil.which(...) returns the absolute .cmd
path which CreateProcessW accepts because the extension routes through
cmd.exe /c. POSIX behaviour unchanged (shutil.which still returns the
same path subprocess would resolve itself).
## HIGH fixes (silent misbehaviour on Windows)
- tools/environments/local.py get_temp_dir: hardcoded /tmp returned on
Windows meant `_cwd_file = "/tmp/hermes-cwd-*.txt"`, which bash wrote
via MSYS2's virtual /tmp but native Python couldn't open. Result: cwd
tracking silently broken — `cd` in terminal tool did nothing. Windows
branch now returns `%HERMES_HOME%/cache/terminal` with forward slashes
(works in both bash and Python, guaranteed no spaces).
- tools/environments/local.py _make_run_env PATH injection: `/usr/bin not
in split(":")` heuristic mangles Windows PATH (";" separator). Gate
the injection behind `not _IS_WINDOWS`.
- hermes_cli/gateway.py launch_detached_profile_gateway_restart: outer
Popen + watcher-script Popen both used start_new_session=True, which
Windows silently ignores. Watcher stayed attached to CLI's console,
died when user closed terminal after `hermes update`, left gateway
stale. Now branches through windows_detach_popen_kwargs() helper
(CREATE_NEW_PROCESS_GROUP | DETACHED_PROCESS | CREATE_NO_WINDOW on
Windows, start_new_session=True on POSIX — identical to main).
## MEDIUM fixes
- gateway/run.py /restart and /update handlers: hardcoded bash/setsid
chain crashes on Windows when user triggers /update in-gateway. Now
has sys.platform=="win32" branch using sys.executable + a tiny
Python watcher with proper detach flags. POSIX path is unchanged.
- cli.py _git_repo_root: Git on Windows sometimes returns /c/Users/...
style paths that break subprocess.Popen(cwd=...) and Path().resolve().
Added _normalize_git_bash_path() helper that translates /c/Users,
/cygdrive/c, /mnt/c variants to native C:\Users form. POSIX no-op.
_git_repo_root() now routes every result through it.
- cli.py worktree .worktreeinclude: os.symlink on directories failed
hard on Windows (requires admin or Developer Mode). Falls back to
shutil.copytree with a warning log.
## Tests
- 29 new tests in tests/tools/test_windows_native_support.py covering:
subprocess_compat helpers, TUI entry signal guards, kanban waitpid
guard, code_execution TCP fallback source-level invariants, cron bash
resolution, npm/npx bare-spawn lint per-file, local env Windows temp
dir, PATH injection gating, git bash path normalization, symlink
fallback, gateway detached watcher flags.
- One existing test assertion adjusted in test_browser_homebrew_paths:
it compared captured Popen argv to the BARE `"npx"` literal; after the
shutil.which() change argv[0] is the absolute path. New assertion
checks the shape (two items, second is `agent-browser`) rather than
the exact first-item string. Behaviour unchanged; test was too strict.
All 56 tests pass on Linux (30 from previous commits + 26 new).
267 tests from the affected files/dirs (browser, code_exec, local_env,
process_registry, kanban_db, windows_compat) all pass — zero regressions.
tests/hermes_cli/ (3909 pass) and tests/gateway/ (5021 pass) unchanged;
all pre-existing test failures confirmed unrelated via `git stash` re-run.
## What's still deferred (LOW priority)
- Visible cmd-window flashes on short-lived console apps (~14 sites) —
cosmetic, needs a follow-up pass once we have user reports.
- agent/file_safety.py POSIX-only security deny patterns — separate
hardening task.
- tools/process_registry.py returning "/tmp" as fallback — theoretical;
reachable only when all env-var candidates fail.
User hit a real failure case: their system Git was in a half-installed state
(can neither uninstall nor reinstall) and winget refused to work around it.
We were one step away from shipping an installer that would have left users
with exactly the problem he already had.
What other agents do (reality check):
- Claude Code: requires pre-installed Git; breaks if user doesn't have it.
- OpenCode, Codex: don't need bash at all — PowerShell-first design.
- Cline: uses whatever shell VSCode is configured with; installs nothing.
None of them solve the "broken system Git" problem. We need to own our Git.
Changes:
- scripts/install.ps1::Install-Git: dropped winget path entirely. Now:
(1) use existing git if present; (2) download portable MinGit from the
official git-for-windows GitHub release to %LOCALAPPDATA%\hermes\git.
No winget, no admin, no Windows installer registry, no system impact.
- Added %LOCALAPPDATA%\hermes\git\{cmd,usr\bin} to User PATH so git + bash
+ POSIX coreutils (which, env, grep, …) resolve in fresh shells.
- tools/environments/local.py::_find_bash: reorder so Hermes' portable
MinGit install is checked BEFORE falling through to shutil.which("bash")
or system install locations. This way a broken system Git can't
hijack the bash lookup.
- README + installation docs reworded to reflect the new story: "portable
Git Bash, isolated from any system install, recoverable via rm -rf if it
ever breaks."
Recoverability: if Hermes' Git install ever breaks, ``Remove-Item %LOCALAPPDATA%\hermes\git``
and re-run the installer — no system impact, no uninstall drama, no winget
to fight with.
Native Windows (with Git for Windows installed) can now run the Hermes CLI
and gateway end-to-end without crashing. install.ps1 already existed and
the Git Bash terminal backend was already wired up — this PR fills the
remaining gaps discovered by auditing every Windows-unsafe primitive
(`signal.SIGKILL`, `os.kill(pid, 0)` probes, bare `fcntl`/`termios`
imports) and by comparing hermes against how Claude Code, OpenCode, Codex,
and Cline handle native Windows.
## What changed
### UTF-8 stdio (new module)
- `hermes_cli/stdio.py` — single `configure_windows_stdio()` entry point.
Flips the console code page to CP_UTF8 (65001), reconfigures
`sys.stdout`/`stderr`/`stdin` to UTF-8, sets `PYTHONIOENCODING` + `PYTHONUTF8`
for subprocesses. No-op on non-Windows. Opt out via `HERMES_DISABLE_WINDOWS_UTF8=1`.
- Called early in `cli.py::main`, `hermes_cli/main.py::main`, and
`gateway/run.py::main` so Unicode banners (box-drawing, geometric
symbols, non-Latin chat text) don't `UnicodeEncodeError` on cp1252
consoles.
### Crash sites fixed
- `hermes_cli/main.py:7970` (hermes update → stuck gateway sweep): raw
`os.kill(pid, _signal.SIGKILL)` → `gateway.status.terminate_pid(pid, force=True)`
which routes through `taskkill /T /F` on Windows.
- `hermes_cli/profiles.py::_stop_gateway_process`: same fix — also
converted SIGTERM path to `terminate_pid()` and widened OSError catch
on the intermediate `os.kill(pid, 0)` probe.
- `hermes_cli/kanban_db.py:2914, 3041`: raw `signal.SIGKILL` →
`getattr(signal, "SIGKILL", signal.SIGTERM)` fallback (matches the
pattern already used in `gateway/status.py`).
### OSError widening on `os.kill(pid, 0)` probes
Windows raises `OSError` (WinError 87) for a gone PID instead of
`ProcessLookupError`. Widened the catch at:
- `gateway/run.py:15101` (`--replace` wait-for-exit loop — without this,
the loop busy-spins the full 10s every Windows gateway start)
- `hermes_cli/gateway.py:228, 460, 940`
- `hermes_cli/profiles.py:777`
- `tools/process_registry.py::_is_host_pid_alive`
- `tools/browser_tool.py:1170, 1206`
### Dashboard PTY graceful degradation
`hermes_cli/pty_bridge.py` depends on `fcntl`/`termios`/`ptyprocess`,
none of which exist on native Windows. Previously a Windows dashboard
would crash on `import hermes_cli.web_server` because of a top-level
import. Now:
- `hermes_cli/web_server.py` wraps the pty_bridge import in
`try/except ImportError` and sets `_PTY_BRIDGE_AVAILABLE=False`.
- The `/api/pty` WebSocket handler returns a friendly "use WSL2 for
this tab" message instead of exploding.
- Every other dashboard feature (sessions, jobs, metrics, config
editor) runs natively on Windows.
### Dependency
- `pyproject.toml`: add `tzdata>=2023.3; sys_platform == 'win32'` so
Python's `zoneinfo` works on Windows (which has no IANA tzdata
shipped with the OS). Credits @sprmn24 (PR #13182).
### Docs
- README.md: removed "Native Windows is not supported"; added
PowerShell one-liner and Git-for-Windows prerequisite note.
- `website/docs/getting-started/installation.md`: new Windows section
with capability matrix (everything native except the dashboard
`/chat` PTY tab, which is WSL2-only).
- `website/docs/user-guide/windows-wsl-quickstart.md`: reframed as
"WSL2 as an alternative to native" rather than "the only way".
- `website/docs/developer-guide/contributing.md`: updated
cross-platform guidance with the `signal.SIGKILL` / `OSError`
rules we enforce now.
- `website/docs/user-guide/features/web-dashboard.md`: acknowledged
native Windows works for everything except the embedded PTY pane.
## Why this shape
Pulled from a survey of how other agent codebases handle native
Windows (Claude Code, OpenCode, Codex, Cline):
- All four treat Git Bash as the canonical shell on Windows, same as
hermes already does in `tools/environments/local.py::_find_bash()`.
- None of them force `SetConsoleOutputCP` — but they don't have to,
Node/Rust write UTF-16 to the Win32 console API. Python does not get
that for free, so we flip CP_UTF8 via ctypes.
- None of them ship PowerShell-as-primary-shell (Claude Code exposes
PS as a secondary tool; scope creep for this PR).
- All of them use `taskkill /T /F` for force-kill on Windows, which
is exactly what `gateway.status.terminate_pid(force=True)` does.
## Non-goals (deliberate scope limits)
- No PowerShell-as-a-second-shell tool — worth designing separately.
- No terminal routing rewrite (#12317, #15461, #19800 cluster) — that's
the hardest design call and needs a separate doc.
- No wholesale `open()` → `open(..., encoding="utf-8")` sweep (Tianworld
cluster) — will do as follow-up if users hit actual breakage; most
modern code already specifies it.
## Validation
- 28 new tests in `tests/tools/test_windows_native_support.py` — all
platform-mocked, pass on Linux CI. Cover:
- `configure_windows_stdio` idempotency, opt-out, env-preservation
- `terminate_pid` taskkill routing, failure → OSError, FileNotFoundError fallback
- `getattr(signal, "SIGKILL", …)` fallback shape
- `_is_host_pid_alive` OSError widening (Windows-gone-PID behavior)
- Source-level checks that all entry points call `configure_windows_stdio`
- pty_bridge import-guard present in `web_server.py`
- README no longer says "not supported"
- 12 pre-existing tests in `tests/tools/test_windows_compat.py` still pass.
- `tests/hermes_cli/` ran fully (3909 passed, 9 failures — all confirmed
pre-existing on main by stash-test).
- `tests/gateway/` ran fully (5021 passed, 1 pre-existing failure).
- `tests/tools/test_process_registry.py` + `test_browser_*` pass.
- Manual smoke: `import hermes_cli.stdio; import gateway.run;
import hermes_cli.web_server` — all clean, `_PTY_BRIDGE_AVAILABLE=True`
on Linux (as expected).
## Files
- New: `hermes_cli/stdio.py`, `tests/tools/test_windows_native_support.py`
- Modified: `cli.py`, `gateway/run.py`, `hermes_cli/main.py`,
`hermes_cli/profiles.py`, `hermes_cli/gateway.py`,
`hermes_cli/kanban_db.py`, `hermes_cli/pty_bridge.py`,
`hermes_cli/web_server.py`, `tools/browser_tool.py`,
`tools/process_registry.py`, `pyproject.toml`, `README.md`, and 4
docs pages.
Credits to everyone whose prior PR work informed these fixes — see
the co-author trailers. All of the PRs listed in
`~/.hermes/plans/windows-support-prs.md` fixing `os.kill` / `signal.SIGKILL`
/ UTF-8 stdio / tzdata / README patterns found the same issues; this PR
consolidates them.
Co-authored-by: Philip D'Souza <9472774+PhilipAD@users.noreply.github.com>
Co-authored-by: Arecanon <42595053+ArecaNon@users.noreply.github.com>
Co-authored-by: XiaoXiao0221 <263113677+XiaoXiao0221@users.noreply.github.com>
Co-authored-by: Lars Hagen <1360677+lars-hagen@users.noreply.github.com>
Co-authored-by: Luan Dias <65574834+luandiasrj@users.noreply.github.com>
Co-authored-by: Ruzzgar <ruzzgarcn@gmail.com>
Co-authored-by: sprmn24 <oncuevtv@gmail.com>
Co-authored-by: adybag14-cyber <252811164+adybag14-cyber@users.noreply.github.com>
Co-authored-by: Prasanna28Devadiga <54196612+Prasanna28Devadiga@users.noreply.github.com>
Extends the cua-driver computer-use backend to drive backgrounded macOS
windows without stealing keyboard or mouse focus from the foreground app.
All changes target the cua-driver MCP backend and the shared dispatcher.
## cua_backend.py
**Window-aware capture**: capture() now calls list_windows + get_window_state
instead of the removed capture tool. Prefers structuredContent.windows
(MCP 2024-11-05+ cua-driver) for zero-parse window enumeration; falls back
to regex-parsed text for older builds. Stores the selected (pid, window_id)
as sticky context so subsequent action calls do not need a redundant round-trip.
**Action routing**: click/scroll/type_text/key all carry the sticky pid
(and window_id for element-indexed clicks). type_text routes through
type_text_chars (individual key events) rather than AX attribute write --
WebKit AXTextFields reject attribute writes from backgrounded processes.
**Key parsing**: _parse_key_combo splits cmd+s-style strings into
(key, [modifiers]) and routes to hotkey (modifier present) or
press_key (bare key) -- cua-driver actual tool names.
**set_value method**: new set_value(value, element) calls the cua-driver
set_value MCP tool. For AXPopUpButton / HTML select in a backgrounded Safari,
AXPress opens the native macOS popup which closes immediately when the app is
non-frontmost; set_value AX-presses the matching child option directly
(no menu required, no focus steal).
**focus_app**: reimplemented as a pure window-selector (enumerates
list_windows, sets sticky pid/window_id) without ever raising the window
or stealing focus.
**list_apps**: fixed tool name from listApps to list_apps; handles plain-text
response via regex when structured data is absent.
**Structured-content extraction**: _extract_tool_result now surfaces
structuredContent from MCP results, enabling the list_windows window array
without text parsing.
**Helpers**: _parse_windows_from_text, _parse_elements_from_tree,
_split_tree_text, _parse_key_combo extracted as module-level functions.
## schema.py
Added set_value to the action enum with a description explaining when to
prefer it over click (select/popup elements, sliders, no focus steal).
Added value field for set_value payloads.
## tool.py
Routed set_value action through _dispatch to backend.set_value.
Added set_value to _DESTRUCTIVE_ACTIONS (approval-gated).
Fixed MIME-type detection in _capture_response: cua-driver may return
JPEG; detect from base64 magic bytes (/9j/ -> image/jpeg, else image/png)
rather than hardcoding image/png.
## agent/display.py + run_agent.py
Guard _detect_tool_failure and result-preview logic against non-string
function_result values: multimodal tool results (dicts with _multimodal=True)
are not string-sliceable; treat them as successes and fall back to str()
for length/preview.
Background macOS desktop control via cua-driver MCP — does NOT steal the
user's cursor or keyboard focus, works with any tool-capable model.
Replaces the Anthropic-native `computer_20251124` approach from the
abandoned #4562 with a generic OpenAI function-calling schema plus SOM
(set-of-mark) captures so Claude, GPT, Gemini, and open models can all
drive the desktop via numbered element indices.
- `tools/computer_use/` package — swappable ComputerUseBackend ABC +
CuaDriverBackend (stdio MCP client to trycua/cua's cua-driver binary).
- Universal `computer_use` tool with one schema for all providers.
Actions: capture (som/vision/ax), click, double_click, right_click,
middle_click, drag, scroll, type, key, wait, list_apps, focus_app.
- Multimodal tool-result envelope (`_multimodal=True`, OpenAI-style
`content: [text, image_url]` parts) that flows through
handle_function_call into the tool message. Anthropic adapter converts
into native `tool_result` image blocks; OpenAI-compatible providers
get the parts list directly.
- Image eviction in convert_messages_to_anthropic: only the 3 most
recent screenshots carry real image data; older ones become text
placeholders to cap per-turn token cost.
- Context compressor image pruning: old multimodal tool results have
their image parts stripped instead of being skipped.
- Image-aware token estimation: each image counts as a flat 1500 tokens
instead of its base64 char length (~1MB would have registered as
~250K tokens before).
- COMPUTER_USE_GUIDANCE system-prompt block — injected when the toolset
is active.
- Session DB persistence strips base64 from multimodal tool messages.
- Trajectory saver normalises multimodal messages to text-only.
- `hermes tools` post-setup installs cua-driver via the upstream script
and prints permission-grant instructions.
- CLI approval callback wired so destructive computer_use actions go
through the same prompt_toolkit approval dialog as terminal commands.
- Hard safety guards at the tool level: blocked type patterns
(curl|bash, sudo rm -rf, fork bomb), blocked key combos (empty trash,
force delete, lock screen, log out).
- Skill `apple/macos-computer-use/SKILL.md` — universal (model-agnostic)
workflow guide.
- Docs: `user-guide/features/computer-use.md` plus reference catalog
entries.
44 new tests in tests/tools/test_computer_use.py covering schema
shape (universal, not Anthropic-native), dispatch routing, safety
guards, multimodal envelope, Anthropic adapter conversion, screenshot
eviction, context compressor pruning, image-aware token estimation,
run_agent helpers, and universality guarantees.
469/469 pass across tests/tools/test_computer_use.py + the affected
agent/ test suites.
- `model_tools.py` provider-gating: the tool is available to every
provider. Providers without multi-part tool message support will see
text-only tool results (graceful degradation via `text_summary`).
- Anthropic server-side `clear_tool_uses_20250919` — deferred;
client-side eviction + compressor pruning cover the same cost ceiling
without a beta header.
- macOS only. cua-driver uses private SkyLight SPIs
(SLEventPostToPid, SLPSPostEventRecordTo,
_AXObserverAddNotificationAndCheckRemote) that can break on any macOS
update. Pin with HERMES_CUA_DRIVER_VERSION.
- Requires Accessibility + Screen Recording permissions — the post-setup
prints the Settings path.
Supersedes PR #4562 (pyautogui/Quartz foreground backend, Anthropic-
native schema). Credit @0xbyt4 for the original #3816 groundwork whose
context/eviction/token design is preserved here in generic form.
The prior implementation routed download_to_file through the shared
_request() path, which uses httpx.AsyncClient.request() inside a
context manager that closes before aiter_bytes() iterates. The body
was read into memory first and the chunked write loop replayed it
from buffer. On small test payloads this was invisible; on real
Teams meeting recordings (hundreds of MB) it would force the full
artifact into RAM per download.
Rewrites download_to_file to open its own AsyncClient and use
client.stream(), keeping the context open across the aiter_bytes
iteration so the body is actually streamed chunk-by-chunk to disk.
Retry/token-refresh/Retry-After semantics are preserved by handling
them inline on the stream path. Partial .part files are cleaned up
on transport errors and on exhausted retries.
Adds three tests: large-payload streaming verifies the chunk loop
runs multiple times (discriminator: 512 KiB at chunk_size=65536
yields 8 chunks under streaming, 1 under buffering), transient-5xx
retry recovers after a single retry, and exhausted-retry cleans up
the partial file.
The new _is_gateway_approval_context() widened the gateway classification
to any call with HERMES_SESSION_PLATFORM bound via contextvars. But
cron/scheduler.py binds that same contextvar for delivery routing on
cron jobs that originate from a gateway platform (telegram/discord/etc.),
so those jobs were getting routed through submit_pending with no
listener — blocking indefinitely instead of honoring approvals.cron_mode.
Short-circuit on HERMES_CRON_SESSION before any gateway check. Cron is
always governed by cron_mode config, regardless of where the job was
scheduled from.
Adds regression coverage in TestCronWithGatewayOrigin and records the
contributor email mapping for scripts/release.py.
Adds one reserved token to the cron `deliver` field:
- `all` — expand to every platform with a configured home channel
Resolves at fire time, not create time, so a job created before Telegram
was wired up picks it up once `TELEGRAM_HOME_CHANNEL` is set. Composes
with existing targets: `origin,all`, `all,telegram:-100:17`.
Inspired by Vellum Assistant's reminder routing-intent system.
## Changes
- cron/scheduler.py: _expand_routing_tokens + integrate into _resolve_delivery_targets
- tools/cronjob_tools.py: schema description updated
- tests/cron/test_scheduler.py: TestRoutingIntents (5 cases)
- website/docs/user-guide/features/cron.md: docs + table rows
## Validation
- tests/cron/test_scheduler.py -k 'Routing or Deliver' → 57 passed
Both implement WebSearchProvider via tools/web_providers/ — matching the
existing SearXNG pattern (PR #5c906d702). Search-only; pair with any
extract provider via web.extract_backend.
- tools/web_providers/brave_free.py — Brave Search API (free tier, 2k
queries/mo). Uses BRAVE_SEARCH_API_KEY as X-Subscription-Token.
- tools/web_providers/ddgs.py — DuckDuckGo via the ddgs Python package.
No API key; gated on package importability.
- tools/web_tools.py: both backends added to _get_backend() config list
and auto-detect chain (trails paid providers), _is_backend_available,
web_search_tool dispatch, web_extract_tool + web_crawl_tool search-only
refusals, check_web_api_key, and the __main__ diagnostic. Introduces
_ddgs_package_importable() helper so tests can monkeypatch a single
symbol for the ddgs availability check.
- hermes_cli/tools_config.py: picker entries for both providers; ddgs
gets a post_setup handler that runs `pip install ddgs`.
- hermes_cli/config.py: BRAVE_SEARCH_API_KEY in OPTIONAL_ENV_VARS.
- scripts/release.py: AUTHOR_MAP entry for @Abd0r.
- tests: 14 new tests (brave-free) + 15 new tests (ddgs) covering
provider unit behavior, backend wiring, and search-only refusals.
Salvages the brave-free + ddgs portion of PR #19796. Not included: the
in-line helpers in web_tools.py (replaced with provider modules to match
the shipped architecture), the lynx-based extract path (these backends
should refuse extract with a clear error — users pair with a real
extract provider), and scripts/start-llama-server.sh (unrelated).
Co-authored-by: Abd0r <223003280+Abd0r@users.noreply.github.com>
For every connected MCP server we register four "utility" tool schemas
(mcp_<server>_list_resources, read_resource, list_prompts, get_prompt).
The existing gate was `hasattr(server.session, method)` — but
`mcp.ClientSession` defines all four methods on the class regardless of
what the remote server supports, so the gate never filtered anything.
Tools-only servers (e.g. @upstash/context7-mcp which advertises only
`tools`) ended up with 4 dead stubs; every model call to them returned
JSON-RPC -32601 Method not found, which made the model conclude the
server was broken even when the real tools worked.
Capture the `InitializeResult` returned by `await session.initialize()`
on the `MCPServerTask`, then gate each utility schema on the
corresponding `capabilities` sub-object (resources / prompts). A
legacy `hasattr` fallback runs when `initialize_result` is missing
(older test fixtures / not-yet-captured code paths) so pre-existing
behavior is preserved.
Verified against real `mcp.types.InitializeResult` pydantic models:
- Context7 shape (tools only) → 0 utility stubs registered (was 4)
- Resources-only server → 2 stubs (list_resources, read_resource)
- Prompts-only server → 2 stubs (list_prompts, get_prompt)
- Fully capable server → all 4 stubs
Closes#18051.
Co-authored-by: nikolay-bratanov <nikolay-bratanov@users.noreply.github.com>
MCP tool results can include ImageContent blocks (screenshots from
Playwright/Blockbench/Puppeteer etc). The tool result handler only
extracted block.text, so image blocks were silently dropped and the
agent saw an empty or text-only response — losing the actual payload.
Add _cache_mcp_image_block() that base64-decodes the block, validates
the bytes via gateway.platforms.base.cache_image_from_bytes (which
sniffs for PNG/JPEG/WebP signatures and rejects non-images), writes to
the shared `~/.hermes/cache/images/` dir, and returns a MEDIA:<path>
tag. The handler appends that tag to the result parts so downstream
gateway adapters render the image inline.
Logs and drops on malformed base64 / non-image payload rather than
raising — a single bad block shouldn't kill the tool call.
Distilled from #17915 (c3115644151) and #10848 (gnanirahulnutakki), both
too stale to cherry-pick (branches diverged enough to revert dozens of
unrelated fixes). Went with #10848's approach of plumbing through
Hermes' existing MEDIA tag / cache_image_from_bytes infrastructure
rather than #17915's raw tempfile path, because it integrates with the
remote-backend mount system and messaging adapters that already handle
MEDIA tags natively.
Co-authored-by: c3115644151 <c3115644151@users.noreply.github.com>
Co-authored-by: gnanirahulnutakki <gnanirahulnutakki@users.noreply.github.com>
* fix(mcp): re-raise CancelledError explicitly in MCPServerTask.run
On Python 3.11+, `asyncio.CancelledError` inherits from `BaseException`
(not `Exception`), so the broad `except Exception as exc:` in
`MCPServerTask.run`'s transport loop did NOT catch it. Task cancellation
from gateway restart / explicit `task.cancel()` silently escaped past
the reconnect logic — the MCP server task died without going through
the shutdown/reconnect code paths that check `_shutdown_event`.
Add an explicit `except asyncio.CancelledError: raise` before the broad
catch so cancellation propagation is self-documenting rather than an
accident of exception hierarchy, and future sibling-site work (e.g.
distinguishing shutdown-cancel from transport-cancel) has an obvious
hook. Behavior on pre-3.8 Pythons where CancelledError WAS an Exception
subclass is also corrected: the old path would have caught it and
treated it as a connection failure worth retrying.
Closes#9930.
* fix(mcp): forward OAuth auth and bump sse_read_timeout on SSE transport
Two surgical correctness bugs in the SSE branch of MCPServerTask._run_http,
distilled from @amiller's PR #5981 that couldn't be cherry-picked wholesale
(branch too stale).
1. sse_read_timeout was set to the tool timeout (default 60s). That's the
wrong dimension — it governs how long sse_client will wait between
events on the SSE stream, not per-call latency. SSE servers routinely
hold the stream idle for minutes between events; a 60s read timeout
drops the connection after the first slow stretch (Router Teamwork,
Supermemory on Cloudflare Workers idle-disconnect at ~60s). Bump to
300s to match the Streamable HTTP path's httpx read timeout.
2. OAuth auth was built via get_manager().get_or_build_provider() but
never forwarded to sse_client. SSE MCP servers behind OAuth 2.1 PKCE
would silently fail with 401s on every request.
Keepalive (the other half of #5981) intentionally left for a follow-up —
it's a real improvement but a bigger change, and these two are obvious
corrections to ship now. Credits to @amiller.
Co-authored-by: Andrew Miller <socrates1024@gmail.com>
---------
Co-authored-by: Andrew Miller <socrates1024@gmail.com>
On Python 3.11+, `asyncio.CancelledError` inherits from `BaseException`
(not `Exception`), so the broad `except Exception as exc:` in
`MCPServerTask.run`'s transport loop did NOT catch it. Task cancellation
from gateway restart / explicit `task.cancel()` silently escaped past
the reconnect logic — the MCP server task died without going through
the shutdown/reconnect code paths that check `_shutdown_event`.
Add an explicit `except asyncio.CancelledError: raise` before the broad
catch so cancellation propagation is self-documenting rather than an
accident of exception hierarchy, and future sibling-site work (e.g.
distinguishing shutdown-cancel from transport-cancel) has an obvious
hook. Behavior on pre-3.8 Pythons where CancelledError WAS an Exception
subclass is also corrected: the old path would have caught it and
treated it as a connection failure worth retrying.
Closes#9930.
PR #21238 introduced top-level `allOf: [{if/then/required}]` blocks in the
built-in memory tool's parameters schema as conditional-required hints.
Two problems:
1. OpenAI's Codex backend (chatgpt.com/backend-api/codex, gpt-5.x) rejects
top-level `allOf`/`anyOf`/`oneOf`/`enum`/`not` outright with a
non-retryable 400 — affected every user on openai-codex/gpt-5.x.
2. The `if/then` hints were silently ignored by every other provider
(Chat Completions doesn't honour them on function schemas), so they
never actually enforced anything anywhere.
The runtime handler in `memory_tool()` already validates the per-action
required fields and returns actionable error messages, so removing the
block changes nothing behaviourally.
Paired with the defense-in-depth sanitizer in the previous commit, this
closes the bug both at the source (schema no longer emits the forbidden
form) and at the wire boundary (sanitizer strips it if anything else
re-introduces it).
- Rewrites `tests/tools/test_memory_tool_schema.py` to guard against
regressing the forbidden-combinator shape instead of asserting it.
- Adds AUTHOR_MAP entry for @hrkzogw (author of the sanitizer fix).
When the parent agent uses a composite toolset like hermes-cli, calling
delegate_task with individual toolsets (e.g. web, terminal) resulted in
zero tools because the name-based intersection failed: 'web' != 'hermes-cli'.
Add _expand_parent_toolsets() which collects all tool names from parent
toolsets, then recognises any individual toolset whose tools are a subset
of the parent's available tools. This allows delegate_task(toolsets=['web'])
to work correctly when the parent has hermes-cli enabled.
Fixes#19447
Some exception classes (e.g. anyio.ClosedResourceError) are raised without
a message argument, so str(exc) returns an empty string. The existing error
format f'{type(exc).__name__}: {exc}' would produce messages like
'MCP call failed: ClosedResourceError: ' with nothing after the colon.
Add _exc_str() helper that falls back to repr(exc) when str(exc) is empty,
and apply it to all 6 MCP error formatting sites (5 tool/prompt/resource
handlers + 1 sampling handler).
Fixes#19417
Treat closed-resource, closed-transport, broken-pipe, and EOF MCP failures as stale session equivalents so the existing reconnect/retry-once path can recover. Add regression coverage for the stale-pipe marker variants.\n\nChecks:\n- python -m py_compile tools/mcp_tool.py tests/tools/test_mcp_tool_session_expired.py\n- python -m pytest tests/tools/test_mcp_tool_session_expired.py -q -o addopts=\n- selected secret scan over touched files
Track elapsed wall time in _run_on_mcp_loop, cancel the in-flight future when a timeout expires, and raise a descriptive TimeoutError that includes the elapsed and configured timeout. Add regression coverage for the new timeout diagnostics.
Fixes#9930
When an agent session is interrupted (Ctrl+C or gateway timeout), the
current thread's interrupt flag is set in _interrupted_threads. asyncio
executor threads are pooled and reused across sessions, so a thread that
carried an interrupt flag from a prior session will immediately cancel
any new asyncio work dispatched to it — including MCP server discovery.
Fix: in register_mcp_servers(), temporarily clear the interrupt flag on
the current thread before running _discover_all(), then restore it
afterward in a finally block so the original interrupt state is not lost.
Image generation plugins were dispatched without a model name, leaving
the plugin to pick its default. Users on OpenRouter, ComfyUI, or custom
backends had no way to select a specific model through config — they
had to fork the plugin or patch the tool.
Add _read_configured_image_model() that reads image_gen.model from the
active profile's config.yaml and forwards it into
_dispatch_to_plugin_provider(). When model is set, the plugin call
gains a 'model' kwarg; when unset, the plugin falls back to its own
default, so single-model users see no behavior change.
Example config:
image_gen:
provider: openrouter
model: flux-pro
Tests: all 170 image tool tests pass. The new code path is opt-in via
config and no existing test exercises it, so the change is strictly
additive.
Two fixes for the local terminal backend on Windows (Git Bash):
1. `_drain()` in base.py: `select.select()` only works on sockets on
Windows, not pipe file descriptors. On Windows, use blocking
`os.read()` in the daemon thread instead. EOF arrives promptly
when bash exits, so this is safe.
2. `_run_bash()` in local.py: When `self.cwd` is updated from `pwd`
output, it contains Git Bash-style paths (`/c/Users/...`).
`subprocess.Popen(cwd=...)` needs a native Windows path
(`C:\Users\...`). Added a conversion before Popen.
Without these fixes, all terminal() calls on Windows return empty
output (exit code 126), and cwd tracking breaks.
Tested on Windows 11 with Git for Windows + Python 3.13.
Fixes#14638
Lists the skills sitting in ~/.hermes/skills/.archive/ so users have
something to pass to `hermes curator restore`. `curator status` already
shows counts; this fills the name-discovery gap.
Archive layout is flat (`archive_skill` writes to `.archive/<skill>/`),
so the directory name IS the skill name — no frontmatter parsing
needed. Timestamped collision directories (`<skill>-<ts>`) are listed
literally; user can still pass them to `restore`.
Reshape of @EvilDrag0n's #20651, simplified: drop the frontmatter
rglob + preamble/trailer output + duplicate subcommand registration.
Co-authored-by: EvilDrag0n <lxl694522264@gmail.com>
Cloud metadata endpoints (169.254.169.254 etc.) are now always blocked
by browser_navigate regardless of hybrid routing, allow_private_urls,
or backend.
Bug: commit 42c076d3 (#16136) added hybrid routing that flips
auto_local_this_nav=True for private URLs and short-circuits
_is_safe_url(). IMDS endpoints are technically private (169.254/16
link-local), so the sidecar happily routed them to a local Chromium,
and the agent could read IAM credentials via browser_snapshot. On
EC2/GCP/Azure this is a full SSRF-to-credential-theft.
Fix: new is_always_blocked_url() in url_safety.py — a narrow floor
that checks _BLOCKED_HOSTNAMES, _ALWAYS_BLOCKED_IPS,
_ALWAYS_BLOCKED_NETWORKS only. Applied as an independent gate in
browser_navigate's pre-nav and post-redirect checks, BEFORE
auto_local_this_nav gets a chance to short-circuit. Ordinary private
URLs (localhost, 192.168.x, 10.x, .local, CGNAT) still route to the
local sidecar as the #16136 feature intends.
Secondary fix (reporter's finding): _url_is_private() now explicitly
checks 172.16.0.0/12. ipaddress.is_private only covers that range on
Python ≥3.11 (bpo-40791), so on 3.10 runtimes those URLs were routed
to cloud instead of the local sidecar. No security impact — just a
correctness fix for the hybrid-routing feature.
Closes#16234.
Add support for MCP servers using the SSE transport protocol
(SseServerTransport) alongside the existing Streamable HTTP and stdio
transports. Many MCP servers use SSE (GET /sse + POST /messages/)
which was previously unsupported -- the client silently fell back to
Streamable HTTP, causing 10s connection timeouts.
Changes:
- Import mcp.client.sse.sse_client with graceful fallback
- Check config.get('transport') == 'sse' in _run_http() to select
the SSE transport path with proper timeout handling
- Read transport type from config in get_mcp_status() instead of
hardcoding 'http' for URL-based servers
- Update docstring, example config, and feature list
The MCP SDK discovers OAuth server metadata (token_endpoint, etc.) on
demand and keeps it in memory only. Without disk persistence, a restart
with valid cached refresh tokens forces the SDK to fall back to the
guessed '{server_url}/token' path — which returns 404 on most real
providers (Notion, Atlassian, GitHub remote MCP, etc.) and triggers a
full browser re-authorization even though the refresh token is fine.
Add a .meta.json file next to the existing tokens/client_info files:
HERMES_HOME/mcp-tokens/<server>.json -- tokens (existing)
HERMES_HOME/mcp-tokens/<server>.client.json -- client info (existing)
HERMES_HOME/mcp-tokens/<server>.meta.json -- oauth metadata (new)
Changes:
- HermesTokenStorage.save_oauth_metadata / load_oauth_metadata / _meta_path
— disk layer for the discovered OAuthMetadata.
- HermesTokenStorage.remove() now also clears .meta.json so
'hermes mcp remove <name>' and the manager's remove() path clean up fully.
- HermesMCPOAuthProvider._initialize cold-restores from disk before the
existing pre-flight discovery runs. If disk has metadata we skip the
discovery HTTP round-trips entirely.
- HermesMCPOAuthProvider._prefetch_oauth_metadata now persists ASM as
soon as it's discovered, so even the first pre-flight run seeds disk.
- HermesMCPOAuthProvider._persist_oauth_metadata_if_changed() is called
at the end of async_auth_flow so metadata discovered via the SDK's
lazy 401-branch (not pre-flight) is also saved for next time.
Tests cover the storage roundtrip (save/load/missing/corrupt/remove) and
the manager provider path (cold-load restore, skip-when-in-memory,
persist-on-discover, noop-when-unchanged, end-to-end async_auth_flow).
Co-authored-by: nocturnum91 <50326054+nocturnum91@users.noreply.github.com>
Skills that produce large/lossless images (e.g. info-graph, where a
rendered JPG is 1-2 MB) currently lose quality in Telegram delivery
because `_IMAGE_EXTS` membership routes the file through
`send_multiple_images` → `sendMediaGroup`, which Telegram's server
re-encodes to JPEG @ 1280px max edge. The original bytes only survive
when the file goes through `send_document`, which the dispatch tables
in three places (`_process_message_background`, `_deliver_media_from_response`,
and the `send_message` tool's telegram path) only reach for files
whose extension is NOT in `_IMAGE_EXTS`.
This commit adds an `[[as_document]]` directive that mirrors the
existing `[[audio_as_voice]]` shape: a skill emits the directive once
in its response, and every image-extension MEDIA: file in that response
is delivered via `send_document` instead of `send_multiple_images` /
`sendPhoto`. The directive is detected at the dispatch sites (which see
the raw response) and the directive string is stripped from the
user-visible cleaned text in `extract_media` so it never leaks.
Granularity is intentionally all-or-nothing per response, matching
[[audio_as_voice]]'s scope. Skills that need fine control can split into
two responses.
Verified the targeted use case: info-graph emits
信息图已生成（...）
[[as_document]]
MEDIA:/tmp/info-graph-x/infographic.jpg
→ Telegram receives `infographic.jpg` via sendDocument, original 1MB
JPEG bytes preserved, no recompression. Forwarding and download
filenames stay clean (`infographic.jpg`).
Tests: +3 cases in TestExtractMedia covering directive strip, isolation
from voice flag, and coexistence with [[audio_as_voice]]. All
113 pre-existing media/extract/send tests pass.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The delegate_task tool schema descriptions referenced 'claude --acp --stdio'
as an example, but Claude Code CLI does not support --acp or --stdio flags.
The ACP subprocess transport (agent/copilot_acp_client.py) is specifically
built for GitHub Copilot CLI ('copilot --acp --stdio').
Changes:
- Per-task acp_command example: 'claude' → 'copilot'
- Top-level acp_command description: remove 'Claude Code' reference,
clarify requirement for ACP-compatible CLI (currently Copilot only)
- acp_args description: remove misleading claude-opus-4-6 example
Fixes#19055
The kanban_heartbeat tool called heartbeat_worker but never
heartbeat_claim, so a worker that loops the tool while a single tool
call blocks the agent for >DEFAULT_CLAIM_TTL_SECONDS still got
reclaimed by release_stale_claims. The function name and
heartbeat_claim's own docstring imply otherwise:
"Workers that know they'll exceed 15 minutes should call this
every few minutes to keep ownership."
But there was no caller in the worker tool path. Workers couldn't
invoke heartbeat_claim themselves either — it isn't exposed as a tool.
Fix: _handle_heartbeat now calls heartbeat_claim first, reading
HERMES_KANBAN_CLAIM_LOCK from the worker env (the dispatcher pins
this in _default_spawn). Falls back to _claimer_id() for locally-
driven workers that didn't go through dispatcher spawn.
Test: tests/tools/test_kanban_tools.py::test_heartbeat_extends_claim_expires
rewinds claim_expires into the past, calls the tool, and asserts the
new value is at least now + DEFAULT_CLAIM_TTL_SECONDS // 2. Verified to
fail against the unfixed code (claim_expires stays at the rewound
value).
Closes the root cause underlying the symptom in #21141 (15-min
respawns of long-running workers). #21141 separately addresses
post-reclaim cleanup; this fixes the upstream "shouldn't have been
reclaimed in the first place" half.
When terminal.backend is docker, inbound documents uploaded via messaging
platforms (Telegram, Slack, Discord, Feishu, Email, etc.) are cached at a host
path under ~/.hermes/cache/documents, but the container sandbox only sees them
at the auto-mounted /root/.hermes/cache/documents path.
This PR adds to_agent_visible_cache_path() in tools/credential_files.py (the
natural sibling to get_cache_directory_mounts()) and calls it at the
document-context-injection site in gateway/run.py so the agent always receives
a path it can open directly, matching the mount layout already established
by get_cache_directory_mounts() (#4846).
Scope: only Docker backend for now; other backends use different mount
semantics and are left unchanged until verified.
Fixes#18787
_write_json (the persistence helper used by HermesTokenStorage for both
tokens and client_info) created the temp file via Path.write_text and
only chmod'd it to 0o600 afterward. Between create and chmod the file
existed on disk at the process umask (commonly 0o644 = world-readable),
briefly exposing MCP OAuth access/refresh tokens to other local users.
Use os.open with O_WRONLY|O_CREAT|O_EXCL and an explicit S_IRUSR|S_IWUSR
mode so the file is created atomically at 0o600, plus tighten the parent
dir to 0o700 so siblings can't traverse to the creds file. The temp name
also gains a per-process random suffix to avoid collisions between
concurrent writers and stale leftovers from a crashed prior write.
Mirrors the fix shipped for agent/google_oauth.py in #19673.
Adds a regression test asserting the resulting file mode is 0o600 and
the parent directory is 0o700 (skipped on Windows where POSIX mode bits
aren't enforced).
Adds SearXNG as a free, self-hosted web search provider. SearXNG is a
privacy-respecting metasearch engine that requires no API key — just a
running instance and SEARXNG_URL pointing at it.
## What this adds
- `tools/web_providers/searxng.py` — `SearXNGSearchProvider` implementing
`WebSearchProvider` (search only; no extract capability)
- `_is_backend_available("searxng")` — gates on SEARXNG_URL
- `_get_backend()` — accepts "searxng" as a configured value; adds it to
auto-detect candidates (lower priority than paid services)
- `web_search_tool` — dispatches to SearXNG when it is the active backend
- `check_web_api_key()` — includes SearXNG in availability check
- `OPTIONAL_ENV_VARS["SEARXNG_URL"]` — registered with tools=["web_search"]
- `tools_config.py` — SearXNG appears in the `hermes tools` provider picker
- `nous_subscription.py` — `direct_searxng` detection, web_active / web_available
- `setup.py` — SEARXNG_URL listed in the missing-credential hint
- 23 tests covering: is_configured, happy-path search, score sorting, limit,
HTTP/request errors, _is_backend_available, _get_backend, check_web_api_key
## Config
```yaml
# Use SearXNG for search, any paid provider for extract
web:
search_backend: "searxng"
extract_backend: "firecrawl"
# Or: SearXNG as the sole backend (web_extract will use the next available)
web:
backend: "searxng"
```
SearXNG is search-only — it does not implement WebExtractProvider. Users
who only configure SEARXNG_URL get web_search available; web_extract falls
back to the next available extract provider (or is unavailable if none).
Closes#19198 (Phase 2 Task 4 — SearXNG provider)
Ref: #11562 (original SearXNG PR)
Introduce the foundation for independently selecting web search and
extract backends — enabling future combinations like SearXNG for
search + Firecrawl for extract.
Architecture:
- tools/web_providers/base.py: WebSearchProvider and WebExtractProvider
ABCs with normalized result contracts (mirrors CloudBrowserProvider)
- tools/web_tools.py: _get_search_backend() and _get_extract_backend()
read per-capability config keys, fall through to shared web.backend
- hermes_cli/config.py: web.search_backend and web.extract_backend in
DEFAULT_CONFIG (empty = inherit from web.backend)
Behavioral change:
- web_search_tool() now dispatches via _get_search_backend()
- web_extract_tool() now dispatches via _get_extract_backend()
- When per-capability keys are empty (default), behavior is identical
to before — _get_search_backend() falls through to _get_backend()
This is purely structural — no new backends are added. SearXNG and
other search-only/extract-only providers can now be added as simple
drop-in modules in follow-up PRs.
12 new tests, 49 existing tests pass with zero regressions.
Ref: #19198
Replaces the per-directory shadow-repo design with a single shared shadow
git store at ~/.hermes/checkpoints/store/. Object DB is now deduplicated
across every working directory the agent has ever touched; a dozen
worktrees of the same project cost near-zero in additional disk.
Why
---
Pre-v2 design had three compounding problems that let ~/.hermes/checkpoints/
grow to multi-GB on active machines:
1. Each working directory got its own full shadow git repo — no object
dedup across projects or across worktrees of the same project.
2. _prune() was a documented no-op: max_snapshots only limited the
/rollback listing. Loose objects accumulated forever.
3. Defaults: enabled=True, auto_prune=False — users paid the disk cost
without ever asking for /rollback.
Field report on a single workstation: 847 MB across 47 shadow repos,
mostly redundant clones of the hermes-agent source tree.
Changes
-------
- tools/checkpoint_manager.py: full rewrite. Single bare store, per-project
refs (refs/hermes/<hash>), per-project indexes (store/indexes/<hash>),
per-project metadata (store/projects/<hash>.json with workdir +
created_at + last_touch). On first v2 init, any pre-v2 per-directory
shadow repos are auto-migrated into legacy-<timestamp>/ so the new
store starts clean. _prune() now actually rewrites the per-project ref
to the last max_snapshots commits and runs git gc --prune=now. New
_enforce_size_cap() drops oldest commits round-robin across projects
when the store exceeds max_total_size_mb. _drop_oversize_from_index()
filters any single file larger than max_file_size_mb out of the snapshot.
- hermes_cli/checkpoints.py: new 'hermes checkpoints' CLI
(status / list / prune / clear / clear-legacy) for managing the store
outside a session.
- hermes_cli/config.py: flipped defaults — enabled=False, max_snapshots=20,
auto_prune=True. Added max_total_size_mb=500, max_file_size_mb=10.
Tightened DEFAULT_EXCLUDES (added target/, *.so/*.dylib/*.dll,
*.mp4/*.mov, *.zip/*.tar.gz, .worktrees/, .mypy_cache/, etc.).
- run_agent.py / cli.py / gateway/run.py: thread the new kwargs through
AIAgent and the startup auto_prune hooks.
- Tests rewritten to match v2 storage while keeping backwards-compat
coverage for the pre-v2 prune path (per-directory shadow repos under
base/ are still swept correctly for anyone mid-migration).
- Docs updated: user-guide/checkpoints-and-rollback.md explains the
shared store, new defaults, migration, and the new CLI;
reference/cli-commands.md documents 'hermes checkpoints'.
E2E validated
-------------
- Legacy migration: pre-v2 shadow repos auto-archived into legacy-<ts>/.
- Object dedup: two projects with an identical shared.py blob resolve to
7 total objects in the store (v1 would have stored the blob twice).
- max_snapshots=3 actually enforced: after 6 commits, list shows 3.
- Orphan prune: deleting a project's workdir + 'hermes checkpoints prune
--retention-days 0' removes its ref, index, and metadata; GC reclaims
the objects.
- max_file_size_mb=1 excludes a 2 MB weights.bin while keeping the
tracked source code files.
- hermes checkpoints {status,prune,clear,clear-legacy} all work from the
CLI without an agent running.
Breaking / migration
--------------------
No in-place data migration — legacy per-directory shadow repos are moved
into legacy-<timestamp>/ on first run. Old /rollback history is still
accessible by inspecting the archive with git; run
'hermes checkpoints clear-legacy' to reclaim the space when ready. Users
relying on /rollback must now set checkpoints.enabled=true (or pass
--checkpoints) explicitly.
Add Lightpanda as an optional browser engine for local mode.
Lightpanda is a headless browser built from scratch in Zig -- faster
navigation than Chrome with significantly less memory.
One config line to enable:
browser:
engine: lightpanda
New functions in browser_tool.py:
- _get_browser_engine() -- config/env reader with validation + caching
- _should_inject_engine() -- only inject in local non-cloud mode
- _needs_lightpanda_fallback() -- detect empty/failed LP results
- _chrome_fallback_screenshot() -- temporary Chrome session for screenshots
- Engine injection in _run_browser_command (--engine flag)
- browser_vision pre-routes screenshots to Chrome when engine=lightpanda
Config:
- browser.engine in DEFAULT_CONFIG (auto/lightpanda/chrome)
- AGENT_BROWSER_ENGINE in OPTIONAL_ENV_VARS
- /browser status shows engine info in local mode
Rebased from PR #7144 onto current main. All existing code preserved --
pure additions only (+520/-2).
25 new tests + 81 total browser tests pass (0 failures).
README:24 claimed "Six terminal backends" while tools/environments/ exposes
seven top-level backend choices through TERMINAL_ENV: local, docker, ssh,
singularity, modal, daytona, vercel_sandbox. Modal additionally has direct
and Nous-managed modes selected via terminal.modal_mode (the
ManagedModalEnvironment class is a Modal sub-mode, not a separate top-level
backend).
The same drift appeared in five other doc and code-comment sites with
inconsistent counts (six, seven, or implicit) and varying lists. Updated
all sites to a consistent seven-backend list in canonical order. The
configuration guide also clarifies how Modal's two modes are selected so
operators do not search for a non-existent backend: managed_modal value.
CONTRIBUTING.md:160 lists six backend filenames in a code tree but does
not carry the "Six terminal" prose; left out of scope per cohesion sweep
guidance to bundle only identical wording.
Files updated:
- README.md (line 24, marketing copy)
- website/docs/index.md (line 49, landing page)
- website/docs/user-guide/configuration.md (line 86, config guide)
- tools/environments/__init__.py (lines 3-6, package docstring)
- tools/file_operations.py (line 6, module docstring)
- environments/README.md (line 43, RL training docs — TERMINAL_ENV list)
The comment at tools/web_tools.py:700-702 stated the runtime default for
auxiliary.web_extract.timeout is 360s. The actual runtime default is 30s
(_DEFAULT_AUX_TIMEOUT in agent/auxiliary_client.py:3140), used by
_get_task_timeout when no auxiliary.web_extract.timeout key is present in
config.yaml.
The 360s figure is the config template default written by
hermes_cli/config.py:697 into freshly-generated config.yaml files. It only
takes effect when that key exists in the user's config — not as a fallback.
Users on configs that predate commit 20b4060d (Apr 5, 2026), or who removed
the key, fall through to the 30s _DEFAULT_AUX_TIMEOUT runtime default.
The comment was introduced in 20b4060d alongside the template-default bump
from 30 to 360. The runtime default in auxiliary_client.py was not changed
in that commit and has remained 30s since 839d9d74 (Mar 28, 2026).
Workers completing a kanban task can now claim the ids of cards they
created via an optional ``created_cards`` field on ``kanban_complete``.
The kernel verifies each id exists and was created by the completing
worker's profile; any phantom id blocks the completion with a
``HallucinatedCardsError`` and records a
``completion_blocked_hallucination`` event on the task so the rejected
attempt is auditable. Successful completions also get a non-blocking
prose-scan pass over their ``summary`` + ``result`` that emits a
``suspected_hallucinated_references`` event for any ``t_<hex>``
reference that doesn't resolve.
Closes#20017.
Recovery UX (kernel + CLI + dashboard)
--------------------------------------
A structural gate alone isn't enough — operators also need to see and
act on stuck workers, especially when a profile's model is the root
cause. This PR ships the full loop:
* ``kanban_db.reclaim_task(task_id)`` — operator-driven reclaim that
releases an active worker claim immediately (unlike
``release_stale_claims`` which only acts after claim_expires has
passed). Emits a ``reclaimed`` event with ``manual: True`` payload.
* ``kanban_db.reassign_task(task_id, profile, reclaim_first=...)`` —
switch a task to a different profile, optionally reclaiming a stuck
running worker in the same call.
* ``hermes kanban reclaim <id> [--reason ...]`` and
``hermes kanban reassign <id> <profile> [--reclaim] [--reason ...]``
CLI subcommands wired through to the same helpers.
* ``POST /api/plugins/kanban/tasks/{id}/reclaim`` and
``POST /api/plugins/kanban/tasks/{id}/reassign`` endpoints on the
dashboard plugin.
Dashboard surfacing
-------------------
* ⚠ **warning badge** on cards with active hallucination events.
* **attention strip** at the top of the board listing all flagged
tasks; dismissible per session.
* **events callout** in the task drawer — hallucination events render
with a red left border, amber icon, and phantom ids as styled chips.
* **recovery section** in the task drawer with three actions: Reclaim,
Reassign (with profile picker + reclaim-first checkbox), and a
copy-to-clipboard hint for ``hermes -p <profile> model`` since
profile config lives on disk and can't be edited from the browser.
Auto-opens when the task has warnings, collapsed otherwise.
Keyed by task id so state doesn't leak between drawers.
Active-vs-stale rule: warnings clear when a clean ``completed`` or
``edited`` event supersedes the hallucination, so recovery is never
permanently stigmatising — the audit events persist for debugging but
the badge goes away once the worker succeeds.
Skill updates
-------------
* ``skills/devops/kanban-worker/SKILL.md`` documents the
``created_cards`` contract with good/bad examples.
* ``skills/devops/kanban-orchestrator/SKILL.md`` gains a "Recovering
stuck workers" section with the three actions and when to use each.
Tests
-----
* Kernel gate: verified-cards manifest, phantom rejection + audit
event, cross-worker rejection, prose scan positive + negative.
* Recovery helpers: reclaim on running task, reclaim on non-running
returns False, reassign refuses running without reclaim_first,
reassign with reclaim_first succeeds on running.
* API endpoints: warnings field present on /board and /tasks/:id,
warnings cleared after clean completion, reclaim 200 + 409 paths,
reassign 200 + 409 + reclaim_first paths.
* CLI smoke: reclaim + reassign subcommands.
Live-verified end-to-end on a dashboard with seeded scenarios:
attention strip renders, badges land on the right cards, drawer
callout shows phantom chips, Reclaim on a running task flips status to
ready + emits manual reclaimed event + refreshes the drawer,
Reassign swaps the assignee and triggers board refresh.
359/359 kanban-suite tests pass
(test_kanban_{db,cli,boards,core_functionality} + dashboard + tools).
* revert(gateway): remove stale-code self-check and auto-restart
Removes the _detect_stale_code / _trigger_stale_code_restart mechanism
introduced in #17648 and iterated in #19740. On every incoming message
the gateway compared the boot-time git HEAD SHA to the current SHA on
disk, and if they differed it would reply with
Gateway code was updated in the background --
restarting this gateway so your next message runs
on the new code. Please retry in a moment.
and then kick off a graceful restart. This is unwanted behaviour:
users who run a long-lived gateway and do their own ad-hoc git
operations on the checkout end up with their chat interrupted and
the current message dropped every time HEAD moves, with no way to
opt out.
If an operator really needs the old protection against stale
sys.modules after "hermes update", the SIGKILL-survivor sweep in
hermes update (hermes_cli/main.py, also tagged #17648) already
handles the supervisor-respawn case on its own.
Removed:
gateway/run.py:
- _STALE_CODE_SENTINELS, _GIT_SHA_CACHE_TTL_SECS
- _read_git_head_sha(), _compute_repo_mtime() module helpers
- class-level _boot_wall_time / _boot_repo_mtime / _boot_git_sha /
_stale_code_restart_triggered defaults
- __init__ boot-snapshot block (_boot_*, _cached_current_sha*,
_repo_root_for_staleness, _stale_code_notified)
- _current_git_sha_cached(), _detect_stale_code(),
_trigger_stale_code_restart() methods
- stale-code check + user-facing restart notice at the top of
_handle_message()
tests/gateway/test_stale_code_self_check.py (deleted, 412 lines)
No new logic added. Zero remaining references to any removed
symbol. Gateway test suite passes the same 4589 tests it passed
before; the 3 pre-existing unrelated failures (discord free-channel,
feishu bot admission, teams typing) are unchanged by this commit.
* feat(i18n): add display.language for static message translation (zh/ja/de/es)
Adds a thin-slice i18n layer covering the highest-impact static user-facing
messages: the CLI dangerous-command approval prompt and a handful of gateway
slash-command replies (restart-drain, goal cleared, approval expired, config
read/save errors).
Out of scope (stays English): agent responses, log lines, tool outputs,
slash-command descriptions, error tracebacks.
Infrastructure:
- agent/i18n.py: catalog loader, t() helper, language resolution
(HERMES_LANGUAGE env var > display.language config > en)
- locales/{en,zh,ja,de,es}.yaml: ~19 translated strings per language
- display.language in DEFAULT_CONFIG (hermes_cli/config.py)
Tests:
- tests/agent/test_i18n.py: 21 tests covering catalog parity, placeholder
parity across locales, fallback behavior, env-var override, alias
normalization, missing-key graceful degradation.
Docs:
- website/docs/user-guide/configuration.md: display.language entry plus a
short section explaining scope so users don't expect agent responses to
translate via this knob.
Sends a lightweight list_tools() probe every 3 minutes during idle
periods to prevent TCP connections from going stale behind LB / NAT
idle timeouts (commonly 300-600s). When the keepalive fails, the
reconnect event fires so the transport rebuilds the session cleanly.
Salvages the keepalive portion of @vominh1919's PR #17016. The
circuit-breaker half-open recovery from the same PR was independently
landed on main via #benbarclay's commit 8cc3cebca ("fix(mcp): add
half-open state to circuit breaker", Apr 21); only the keepalive is
salvaged here.
Fixes#17003.
Previously, pinning a skill blocked every skill_manage write action
(edit, patch, delete, write_file, remove_file). The 'hard fence'
design conflated two concerns:
1. Pin as deletion protection — don't let the curator archive
or the agent delete a stable skill.
2. Pin as content freeze — don't let the agent rewrite it mid-conversation.
In practice (1) is what users pin for: they want a skill to survive
curator passes. (2) created friction — agents finding a new pitfall
in a pinned skill had to ask the user to unpin, then the agent
patches, then the user re-pins. The dance discouraged skill
maintenance and pinned skills went stale.
This narrows the _pinned_guard to skill_manage(action='delete') only.
Patches, edits, and supporting-file writes go through on pinned
skills so the agent can keep improving them. The curator's own
pinned-skip behavior (agent/curator.py:271 for auto-archive,
line 349 for the LLM review prompt) is unchanged — curator still
never touches pinned skills.
Changes:
- tools/skill_manager_tool.py: remove _pinned_guard calls from
_edit_skill, _patch_skill, _write_file, _remove_file; keep on
_delete_skill. Updated _pinned_guard docstring and error message.
- tools/skill_manager_tool.py: updated skill_manage model-facing tool
description to reflect the new semantic.
- website/docs/user-guide/features/curator.md: updated pinning
section.
- tests/tools/test_skill_manager_tool.py: flipped refuses-pinned
tests for edit/patch/write_file/remove_file into allowed-when-pinned;
kept test_delete_refuses_pinned (strengthened assertion to check the
'cannot be deleted' wording).
Closes#18354
Closes the gap where write_file skipped the post-edit syntax check that
patch already ran, so silent file corruption (bad quote escaping,
truncated writes, etc.) would persist on disk until a later read.
## Changes
tools/file_operations.py:
- Add in-process linters for .py, .json, .yaml, .toml (LINTERS_INPROC).
Python uses ast.parse, JSON/YAML/TOML use stdlib/PyYAML parsers.
Zero subprocess overhead; preferred over shell linters when both apply.
- _check_lint() now accepts optional content and routes to in-process
linter first. Shell linter (py_compile, node --check, tsc, go vet,
rustfmt) remains the fallback for languages without an in-process
equivalent.
- New _check_lint_delta() implements the post-first/pre-lazy pattern
borrowed from Cline and OpenCode: lint post-write state first; only
if errors are found AND pre-content was captured does it lint the
pre-state and diff. If the pre-existing file had the SAME errors the
edit didn't introduce anything new, so the file is reported as 'still
broken, pre-existing' with success=False but a message explaining the
errors were pre-existing. If the edit introduced genuinely new errors,
those are surfaced and pre-existing ones are filtered out.
- WriteResult gains a lint field.
- write_file() captures pre-content for in-process-lintable extensions
and calls _check_lint_delta after a successful write.
- patch_replace() switches from _check_lint to _check_lint_delta,
reusing the pre-edit content it already has in scope.
tools/file_tools.py:
- Update write_file schema description to mention the post-write lint.
tests/tools/test_file_operations_edge_cases.py:
- Update existing brace-path tests to use .js (shell linter) now that
.py is in-process.
- Add TestCheckLintInproc (9 tests) covering Python/JSON/YAML/TOML
in-process linters.
- Add TestCheckLintDelta (5 tests) covering the post-first/pre-lazy
short-circuit, new-file path, and the single-error-parser caveat.
## Performance
In-process linters are microseconds per call (ast.parse, json.loads).
The hot path (clean write) runs exactly one lint — matches main's cost
for patch. Pre-state capture is skipped when the file has no applicable
linter. Measured 4.89ms/write average over 100 .py writes including lint.
## Inspiration
- Cline's DiffViewProvider.getNewDiagnosticProblems() — filters pre-write
diagnostics from post-write diagnostics (src/integrations/editor/DiffViewProvider.ts).
- OpenCode's WriteTool — runs lsp.diagnostics() after write and appends
errors to tool output (packages/opencode/src/tool/write.ts).
- Claude Code's DiagnosticTrackingService — captures baseline via
beforeFileEdited() and returns new-diagnostics-only from
getNewDiagnostics() (src/services/diagnosticTracking.ts).
## Validation
- tests/tools/test_file_operations.py + test_file_operations_edge_cases.py
+ test_file_tools.py + test_file_tools_live.py + test_file_write_safety.py
+ test_write_deny.py + test_patch_parser.py + test_file_ops_cwd_tracking.py:
228 passed locally.
- Live E2E reproduction of the tips.py corruption incident: broken
content written; lint field surfaces 'SyntaxError: invalid syntax.
Perhaps you forgot a comma? (line 6, column 5)' — the exact error
that would have self-corrected the bug on the next turn.
MCP servers commonly emit JSON Schema `pattern` (e.g. `\\d{4}-\\d{2}-\\d{2}`
for date-time params) and `format` keywords. llama.cpp's
`json-schema-to-grammar` converter rejects regex escape classes
(\\d/\\w/\\s) and most format values, returning HTTP 400
"parse: error parsing grammar: unknown escape at \\d" — the whole request
fails.
Cloud providers (OpenAI, Anthropic, OpenRouter, Gemini) accept these
keywords fine and use them as prompting hints. Stripping unconditionally
loses useful hints for every cloud user to fix a llama.cpp-only bug.
Approach: classify the llama.cpp grammar-parse 400 in the error
classifier, and on match do a one-shot in-place strip of pattern/format
from `self.tools`, then retry. Follows the existing
`thinking_signature` recovery pattern. Cloud users hit zero overhead;
llama.cpp users pay one failed request per session.
Changes
- agent/error_classifier.py: new `FailoverReason.llama_cpp_grammar_pattern`
+ narrow HTTP-400 branch matching "error parsing grammar",
"json-schema-to-grammar", or "unable to generate parser ... template".
- tools/schema_sanitizer.py: new `strip_pattern_and_format()` helper —
reactive, walks schema nodes, skips property names (search_files.pattern
survives). Returns strip count for logging.
- run_agent.py: new one-shot recovery block in the retry loop. Strips,
logs, continues. Falls through to normal retry if nothing to strip.
- tests: 4 classifier tests (3 variants + 1 non-400 negative), 7 strip
tests including the property-name preservation and idempotency checks.
Co-authored-by: Chris Danis <cdanis@gmail.com>
Follow-up to #19928 which fixed the foreground path in _run_bash.
The background process spawn in process_registry.py had the same
vulnerability: Popen(cwd=session.cwd) and PtyProcess.spawn(cwd=...)
would raise FileNotFoundError if the directory was deleted.
Apply _resolve_safe_cwd() at session creation time so both the PTY
and pipe-mode Popen paths receive a validated cwd.
Address Copilot review on PR #17569:
1. _resolve_safe_cwd never tested the filesystem root because the loop
exited when `os.path.dirname(parent) == parent`, which is true once
`parent == '/'`. Restructure so the root is checked before the
self-equal exit. Adds `test_returns_root_when_only_root_exists` —
regression-guarded by reverting the loop and watching it fail.
2. The fake `Popen.stdout` was a `MagicMock`; `BaseEnvironment._wait_for_process`
calls `proc.stdout.fileno()` then `select.select`/`os.read` against it,
which raised `TypeError: fileno() returned a non-integer` (visible as a
thread exception in test output) and could in theory read from an
unrelated real fd. Hand `fake_popen` a real `os.pipe()` with the write
end pre-closed so the drain loop sees EOF immediately. Helper records
each fd so the test cleans up after itself.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
When a tool call deletes its own working directory (`cd /tmp/foo &&
rm -rf /tmp/foo`), the next `subprocess.Popen(args, cwd=self.cwd)` raised
`FileNotFoundError: [Errno 2]` before bash even started — every subsequent
terminal/file-tool call hit the same wedge until the gateway restarted.
Fix in `LocalEnvironment._run_bash`: before handing `self.cwd` to Popen,
resolve a safe alternative when the path is gone (walk up to the nearest
existing ancestor, falling back to `tempfile.gettempdir()` only as a last
resort). Log a warning so the recovery is visible — not silent — and
update `self.cwd` so the next call doesn't repeat the message.
Defense in depth in `LocalEnvironment._update_cwd`: only adopt the new
cwd when it still exists as a directory. `pwd -P` from a deleted cwd can
leave a stale value in the marker file; refusing to store a missing path
keeps `self.cwd` valid by construction.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
MiniMax deprecated the old v1/t2a_v2 endpoint (api.minimax.io) and
moved to v1/text_to_speech (api.minimax.chat). The new API:
- Uses a flat payload: {model, text, voice_id} instead of nested
voice_setting / audio_setting objects
- Returns raw audio bytes (Content-Type: audio/mpeg) instead of
JSON with hex-encoded audio
- Uses model 'speech-01' instead of 'speech-2.8-hd'
- Updated default voice_id to 'female-shaonv' for Chinese TTS
The implementation detects Content-Type to handle both old and new
API responses, maintaining backward compatibility for any users who
manually configured the legacy base_url.
* feat(cron): add no_agent mode for script-only cron jobs (watchdog pattern)
Adds a no_agent=True option to the cronjob system. When enabled, the
scheduler runs the attached script on schedule and delivers its stdout
directly to the job's target — no LLM, no agent loop, no token spend.
This is the classic bash-watchdog pattern (memory alert every 5 min,
disk alert every 15 min, CI ping) reimplemented as a first-class Hermes
primitive instead of a systemd timer + curl + bot token triplet living
outside the system.
## What
hermes cron create "every 5m" \
--no-agent \
--script memory-watchdog.sh \
--deliver telegram \
--name memory-watchdog
Agent tool:
cronjob(action='create',
schedule='every 5m',
script='memory-watchdog.sh',
no_agent=True,
deliver='telegram')
Semantics:
- Script stdout (trimmed) → delivered verbatim as the message
- Empty stdout → silent tick (no delivery; watchdog pattern)
- wakeAgent=false gate → silent tick (same gate LLM jobs use)
- Non-zero exit/timeout → delivered as an error alert
(broken watchdogs shouldn't fail silently)
- No LLM ever invoked; no tokens spent; no provider fallback applied
## Implementation
cron/jobs.py
* create_job gains no_agent: bool = False
* prompt becomes Optional (no_agent jobs don't need one)
* Validation: no_agent=True requires a script at create time
* Field roundtrips via load_jobs / save_jobs / update_job
cron/scheduler.py
* run_job: new short-circuit branch at the top that runs the script,
wraps its output into the (success, doc, final_response, error)
tuple downstream delivery already expects, and returns before any
AIAgent import or construction
* _run_job_script: picks interpreter by extension — .sh/.bash run
under /bin/bash, anything else under sys.executable (Python).
Shell support unlocks the bash-watchdog pattern without wrapping
scripts in Python. Extension is explicit; we deliberately do NOT
trust the file's own shebang. Path-containment guard (scripts dir)
unchanged.
tools/cronjob_tools.py
* Schema: new no_agent boolean property with clear trigger guidance
* cronjob() accepts no_agent and validates mode-specific shape:
- no_agent=True requires script; prompt/skills optional
- no_agent=False keeps the existing 'prompt or skill required' rule
* update path rejects flipping no_agent=True on a job without a script
* _format_job surfaces no_agent in list output
* Handler lambda forwards no_agent from tool args
hermes_cli/main.py, hermes_cli/cron.py
* 'hermes cron create --no-agent' and edit's --no-agent / --agent
pair for toggling at CLI parity with the agent tool
* Existing --script help text updated to describe both modes
* List / create / edit output now shows 'Mode: no-agent (...)' when set
## Tests
tests/cron/test_cron_no_agent.py — 18 tests covering:
* create_job: no_agent shape, validation, field persistence
* update_job: flag roundtrip across reload
* cronjob tool: schema validation, update toggling, mode-specific
requirements, prompt-relaxation rule
* run_job short-circuit:
- success path delivers stdout verbatim
- empty stdout → SILENT_MARKER (no delivery downstream)
- wakeAgent=false gate → silent
- script failure → error alert
- run_job does NOT import AIAgent (verified via mock)
* _run_job_script:
- .sh executes via bash (no shebang required)
- .bash executes via bash
- .py still runs via sys.executable (regression)
- path-traversal still blocked (security regression)
All 18 new tests pass. 341/342 pre-existing cron tests still pass; the
one failure (test_script_empty_output_noted) was already broken on main
and is unrelated to this change.
## Docs
website/docs/guides/cron-script-only.md — new dedicated guide covering
the watchdog pattern, interpreter rules, delivery mapping, worked
examples (memory / disk alerts), and the comparison table vs hermes send,
regular LLM cron jobs, and OS-level cron.
website/docs/user-guide/features/cron.md — new 'No-agent mode' section
in the cron feature reference, cross-linked to the guide.
website/docs/guides/automate-with-cron.md — new tip box pointing users
to no-agent mode when they don't need LLM reasoning.
## Compatibility
- Existing jobs: unchanged. no_agent defaults to False, existing code
paths untouched until the flag is set.
- Schema additive only; older jobs.json without the field load fine
via .get() with False default.
- New CLI flags are opt-in and don't alter existing flag behavior.
* fix(cron): lazy-import AIAgent + SessionDB so no_agent ticks pay zero
The unconditional `from run_agent import AIAgent` + SessionDB() init at
the top of run_job() meant every no_agent tick still paid the full agent
module load cost (~300ms + transitive imports + DB open) even though it
never touched any of that machinery.
Move both to live under the default (LLM) path, after the no_agent
short-circuit has returned. Now a no_agent tick's sys.modules stays
clean — verified end-to-end:
assert 'run_agent' not in sys.modules # before
run_job(no_agent_job)
assert 'run_agent' not in sys.modules # after
The existing mock-based unit test (test_run_job_no_agent_never_invokes_aiagent)
kept passing because patch() replaces the class AFTER import; the leak
was only visible via real subprocess-style verification. End-to-end
demo confirmed: agent calls cronjob(no_agent=True) → script runs →
stdout delivered → no LLM machinery loaded.
* docs(cron): tighten no_agent tool schema — defaults, silent semantics, pick rule
Previous description buried the important bits in one long sentence.
Agents could plausibly miss three things an LLM-facing schema should
make unmissable:
1. What the default is — now first sentence + JSON Schema `default: false`
2. What 'silent run' actually means for the user — now spelled out:
'nothing is sent to the user and they won't see anything happened'
3. When to pick True vs False — now a concrete decision rule with
examples on both sides (watchdogs/metrics/pollers → True;
summarize/draft/pick/rephrase → False)
Also adds explicit 'prompt and skills are ignored when True' since the
agent could otherwise still pass them out of habit.
No behavior change — schema text only.
On VPS/Docker and some Ubuntu 23.10+ hosts, Chromium refuses to start
without --no-sandbox:
- uid=0 (root): hard requirement (VPS/Docker deployments)
- AppArmor apparmor_restrict_unprivileged_userns=1 (Ubuntu 23.10+):
non-root too, under systemd or unprivileged containers
Detect both conditions and inject AGENT_BROWSER_CHROME_FLAGS with
--no-sandbox --disable-dev-shm-usage when the user hasn't already
set the flags themselves.
Salvage of #15771 — only the browser_tool.py fix is cherry-picked.
The PR's accompanying MCP preset addition (new feature surface)
was dropped so the bug fix can land independently.
Co-authored-by: ygd58 <buraysandro9@gmail.com>
Clear inherited provider preference filters when delegation.provider is set so delegated children do not route back to the parent provider. Add a regression test for cross-provider delegation with parent OpenRouter filters.
Closes#10653
When a delegation child session (e.g. source='telegram') contains the
FTS5 hit but _resolve_to_parent() maps it to a different root session
(source='api_server'), the result entry was still reporting the child's
source because the loop discarded session_meta as `_` and fell back to
match_info.get('source'), which carries the child session's value.
Use the resolved parent's session_meta for source, model, and started_at
with match_info as a fallback, so the output accurately reflects the
session the user actually interacted with.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Reverts ff3d2773e2. Teknium reviewed the merged PR and decided this
behavior isn't wanted — tool-driven kanban_create should not mirror
the slash-command path's auto-subscribe. Orchestrators that want
their originating chat notified can call kanban_notify-subscribe
explicitly; we're not going to make it implicit.
Closes#19479.
When an orchestrator agent calls kanban_create from a gateway session
(e.g. a Telegram user delegating to an orchestrator profile), auto-
subscribe the originating (platform, chat, thread, user) to the new
task's terminal events. Mirrors the behavior of the /kanban create
slash command in gateway/run.py so tool-driven creation is at parity
with human-driven creation.
Without this, a user who interacts with an orchestrator exclusively
via the gateway never receives blocked / completed / gave_up
notifications for tasks the orchestrator created on their behalf —
silently breaking the gateway-first multi-agent flow the reporter
describes.
Reads the context-local HERMES_SESSION_* vars via get_session_env()
(not os.environ — those are contextvars for asyncio concurrency
safety). Falls through cleanly in CLI / cron contexts with no
session active (subscribed=False in the response). Best-effort: if
the gateway module isn't importable (test rigs stubbing gateway.*),
the task still creates, we just skip the subscription.
Response gains a 'subscribed' bool so the orchestrator knows whether
terminal events will land back in the originating chat or whether it
needs to poll / unblock manually.
Tests: 4 new in tests/tools/test_kanban_tools.py covering
CLI/no-subscribe, telegram/gateway-auto-subscribe, discord-DM/no-
thread subscribe, and partial-ctx/no-chat_id no-subscribe. 40/40
kanban tool tests pass.
ENV-assignment and JSON-field regex patterns in redact_sensitive_text()
cause false positives when reading source code files:
- MAX_TOKENS=*** triggers the ENV assignment pattern
- "apiKey": "test" in test fixtures triggers the JSON field pattern
Add code_file=False parameter. When code_file=True, skip only the
ENV-assignment and JSON-field regex passes; all other patterns (prefixes,
auth headers, private keys, DB connstrings, JWTs, URL secrets) are
still applied.
Update file_tools.py (read_file and search_files) to pass code_file=True
so agent code analysis is not polluted by false-positive redactions.
Closes#15934
Closes#19534 (security).
A worker spawned by the kanban dispatcher has HERMES_KANBAN_TASK set
to its own task id. The destructive tools (kanban_complete,
kanban_block, kanban_heartbeat) resolved task_id via
_default_task_id() which preferred an explicit arg over the env var,
with no ownership check — so a buggy or prompt-injected worker could
complete / block / heartbeat any OTHER task (sibling, cross-tenant,
anything) by supplying its id. Reporter's repro: worker for t_A
passed task_id=t_B to kanban_complete and got {"ok": true}.
Fix: add _enforce_worker_task_ownership(tid). If HERMES_KANBAN_TASK
is set and tid doesn't match, return a structured tool error with
guidance to use kanban_comment (for information handoff across tasks)
or kanban_create (for follow-up work). Orchestrator profiles (no env
var, but kanban toolset enabled per #18968) are exempt — their job
is routing and sometimes includes closing out child tasks.
Kept unrestricted (deliberately):
- kanban_show — workers legitimately read parent/sibling handoff context
- kanban_comment — cross-task comments are the handoff mechanism
- kanban_create — orchestrator fan-out, worker follow-up spawning
- kanban_link — parent/child linking
Tests: 5 new regression tests in tests/tools/test_kanban_tools.py
covering the grid (worker-attacks-foreign ×3 tools, worker-own-task
preserved, orchestrator-unrestricted). 36/36 pass.
AnyUrl was imported inside the same try block as mcp.client.auth, so
when the mcp package was not installed, AnyUrl was undefined and
_build_client_metadata raised NameError at runtime.
Moved the AnyUrl import to its own try/except block so it's available
whenever pydantic is installed (which is a core dependency), regardless
of whether the mcp SDK is present.
Also added pytest.importorskip('mcp') to the three
test_build_client_metadata tests that exercise _build_client_metadata,
since that function depends on OAuthClientMetadata from the mcp package.
Set max_result_size_chars=100_000 on the read_file registry entry (was
float('inf')), closing the Layer 2 defense-in-depth gap in
tool_result_storage.py. The existing Layer 1 guard inside
_handle_read_file already returns a JSON error for oversized reads;
this aligns the registry cap with every other tool.
Update test_read_file_never_persisted → test_read_file_result_size_cap
to assert 100_000, and add test_read_file_registry_cap_is_100k as an
explicit regression guard against re-introducing float('inf').
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Treat explicit CDP override mode as a valid browser backend even when agent-browser is absent, and add a regression test to prevent false-negative availability gating.
`_resolve_model_override` treated any non-empty `provider` string from
the LLM as user-specified and skipped the pin-to-current-provider
fallback. When the LLM wrote bare `'custom'` (instead of the canonical
`'custom:<name>'` referring to a custom_providers entry), the value
serialized into jobs.json as `"provider": "custom"` and the scheduler
could never resolve a provider from it — the cron job failed silently
at run time.
Treat bare `'custom'` as "no provider supplied" so the current main
provider gets pinned instead, matching behaviour for the omitted case.
Defence-in-depth complement to a schema-description fix (#15477) that
discourages the LLM from emitting bare `'custom'` in the first place.
YAML parses `delegation: null` as Python None. `dict.get(key, {})`
only uses the default when the key is *missing*, not when it exists with
a None value, so `cfg.get("max_concurrent_children")` crashes with
`'NoneType' object has no attribute 'get'`.
Same pattern as fd9b692d (fix(tui): tolerate null top-level sections).
Use `dict.get(key) or {}` to handle both missing and None-valued keys.
Closes: delegation null config crash (same class as #7215, #7346)
The `provider` field in CRONJOB_SCHEMA only showed examples like
'openrouter' and 'anthropic', with no mention of the canonical
'custom:<name>' form required for custom_providers entries. When the
user has custom providers configured, LLMs tend to write the bare type
name ('custom') because the schema does not advertise the ':<name>'
suffix. The bare value then serializes into jobs.json and causes the
cron job to fail silently at run time — `_resolve_model_override`
treats it as a user-specified provider and skips the pin-to-current
fallback, but no provider ever resolves from the bare 'custom' string.
Clarifying the schema so the canonical form is discoverable addresses
the root cause at the tool-definition boundary.
Adds RFC 5322 Date header to the _send_email tool path in tools/send_message_tool.py.
Issue #15160 noted that both gateway/platforms/email.py and tools/send_message_tool.py
construct MIMEMultipart/MIMEText messages without setting a Date header. RFC 5322
requires the Date header; mail filters reject messages that lack it.
PR #15207 fixed the gateway/platforms/email.py path but did not cover
tools/send_message_tool._send_email, which is used by the send_message tool
for cross-channel messaging.
This change adds msg["Date"] = formatdate(localtime=True) to _send_email,
mirroring the fix applied to the gateway email adapter.
Closes#15160
Tighten the provenance semantics added in #19618: skills a user asks a
foreground agent to write via skill_manage(create) now stay invisible to
the curator. Only skills the background self-improvement review fork
sediments through skill_manage get the created_by=agent marker.
- tools/skill_provenance.py — new ContextVar module mirroring the
_approval_session_key pattern: set_current_write_origin / reset /
get / is_background_review. Default origin is 'foreground'; the
review fork sets 'background_review'.
- run_agent.py — run_conversation() binds the ContextVar from
self._memory_write_origin at the top of each call. The review fork
runs on its own thread (fresh context), so foreground and review
contexts never cross-contaminate.
- tools/skill_manager_tool.py — skill_manage(action='create') now
only calls mark_agent_created() when is_background_review(). All
other cases (foreground create, patch, edit, write_file, delete)
continue as before.
- tests: test_skill_provenance.py (6 tests covering the ContextVar
surface), split test_full_create_via_dispatcher into foreground
vs. review-fork variants, curator status tests now mark-first.
Why: the agent routinely edits existing user skills on the user's
behalf; those writes must never flip provenance. And when a user
explicitly asks the foreground agent to create a skill, that skill
belongs to the user. The curator should only be cleaning up after
its own autonomous sediment from the review nudge loop.
When delegation.model differs from model.default and the provider is
opencode-go or opencode-zen, the wrong api_mode is computed because
resolve_runtime_provider falls back to model_cfg.get('default') — the
main model — instead of the configured delegation model.
For example, with model.default=minimax-m2.7 (anthropic_messages) and
delegation.model=glm-5.1 (chat_completions), subagents get
anthropic_messages, which strips /v1 from the base URL and causes a 404.
resolve_runtime_provider already accepts target_model for exactly this
purpose; _resolve_delegation_credentials just wasn't passing it.
Fixes#15319
Related: #13678
The _check_kanban_mode() gating function only checked for
HERMES_KANBAN_TASK env var, which is only set by the dispatcher
when spawning workers. This prevented orchestrator profiles (like
techlead) from using kanban_create, kanban_link, etc. even when
they had 'kanban' explicitly in their toolsets config.
Now uses load_config() from hermes_cli.config (which has mtime-based
caching) to check if 'kanban' is in the profile's toolsets list.
This enables orchestrators to route work via Kanban while workers
continue using the dispatcher env var.
Fixes#18968
_build_child_agent constructed child AIAgents without passing
fallback_model, leaving _fallback_chain=[] for every subagent.
When a subagent hit a rate-limit or credential exhaustion the
runtime fallback check (run_agent.py:7486 / 12267) found an empty
chain and failed immediately — even though the parent agent was
configured with fallback_providers and would have recovered.
The cron scheduler already propagates fallback_model correctly
(scheduler.py:1038). Fix closes the parity gap by reading the
parent's _fallback_chain (the normalised list form accepted by
AIAgent's fallback_model parameter) and threading it through.
Empty chains coerce to None so AIAgent initialises _fallback_chain=[]
as usual rather than iterating an empty list.
The _send_feishu() function already supports media_files (images, video,
audio, documents) via the adapter's send_image_file/send_video/send_voice
/send_document methods, but _send_to_platform() never routed Feishu into
the early media-handling branch — media attachments were silently dropped
with a "not supported" warning.
Add a Feishu-specific media branch (matching the existing Yuanbao/Signal
pattern) so that MEDIA:<path> tags in send_message calls are correctly
delivered as native Feishu attachments. Also update the two error/warning
message strings to include feishu in the supported platform list.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Before this fix, _chromium_installed() only searched Playwright-style
chromium-* / chromium_headless_shell-* directories, which meant users
with system Chrome or AGENT_BROWSER_EXECUTABLE_PATH configured still
had all browser_* tools gated.
Now checks three sources in priority order:
1. AGENT_BROWSER_EXECUTABLE_PATH env var (if set and points to a real binary)
2. System Chrome/Chromium via shutil.which() (google-chrome, chromium-browser, chrome)
3. Playwright browser cache (existing logic, kept as fallback)
Closes#19294
The _send_qqbot function was hardcoded to use the guild channel
endpoint (/channels/{id}/messages), which fails for C2C private
chats and QQ groups with 'channel does not exist' (code 11263).
This change tries the appropriate endpoints in order:
1. /channels/{id}/messages (guild channels)
2. /v2/users/{id}/messages (C2C private chats)
3. /v2/groups/{id}/messages (QQ groups)
Fixes active sending to QQBot C2C and group recipients.
* feat: add video_analyze tool for native video understanding
Adds a video_analyze tool that sends video files to multimodal LLMs
(e.g. Gemini) for analysis via the OpenRouter-compatible video_url
content type. Mirrors vision_analyze in structure, error handling,
and registration pattern.
Key design:
- Base64 encodes entire video (no frame extraction, no ffmpeg dep)
- Uses 'video_url' content block type (OpenRouter standard)
- Supports mp4, webm, mov, avi, mkv, mpeg formats
- 50 MB hard cap, 20 MB warning threshold
- 180s minimum timeout (videos take longer than images)
- AUXILIARY_VIDEO_MODEL env override, falls back to AUXILIARY_VISION_MODEL
- Same SSRF protection, retry logic, and cleanup as vision_analyze
Default disabled: registered in 'video' toolset (not in _HERMES_CORE_TOOLS).
Users opt in via: hermes tools enable video, or enabled_toolsets=['video'].
* feat(video): add models.dev capability pre-check + CONFIGURABLE_TOOLSETS entry
- Pre-checks model video capability via models.dev modalities.input
before expensive base64 encoding. Fails early with helpful message
suggesting video-capable alternatives (gemini, mimo-v2.5-pro).
- Passes optimistically if model unknown or lookup fails.
- Adds ModelInfo.supports_video_input() helper.
- Adds 'video' to CONFIGURABLE_TOOLSETS and _DEFAULT_OFF_TOOLSETS
so 'hermes tools enable video' works from CLI.
- 8 new tests for the capability check (37 total).
* refactor(video): remove models.dev capability pre-check
Removes _check_video_model_capability and ModelInfo.supports_video_input.
The vision_analyze tool doesn't pre-check image capability either — both
tools rely on the same pattern: send request, handle API errors gracefully
with categorized user-facing messages. The pre-check was inconsistent
(only worked for some providers/models) so drop it for parity.
* cleanup: compress comments, fix fragile timeout coupling
- Replace _VISION_DOWNLOAD_TIMEOUT * 2 with hardcoded 60s (no silent
breakage if vision timeout changes independently)
- Strip verbose comments and redundant log lines throughout
- No behavioral changes
Under context pressure, frontier models sometimes emit tool calls with
required fields dropped. Previously _handle_write_file() used
args.get('content', '') which substituted an empty string for the missing
key, returned success with bytes_written=0, and created a zero-byte file
on disk. The model had no way to detect the failure.
Changes:
- Reject calls where 'path' is absent or not a non-empty string
- Reject calls where 'content' key is entirely absent (key-presence check,
not truthiness) — distinguishing a legitimately empty file from a dropped arg
- Reject calls where 'content' is a non-string type
- All error messages include guidance to re-emit the tool call or switch
to execute_code with hermes_tools.write_file() for large payloads
- Explicit empty string content (file truncation) continues to work
Regression tests added for all four cases: missing path, missing content,
explicit-empty content, and wrong content type.
Fixes#19096
Terminal commands can write to shell RC files (~/.bashrc, ~/.zshrc,
~/.profile) and credential files (~/.netrc, ~/.pgpass, ~/.npmrc,
~/.pypirc) via redirection or tee without triggering approval, even
though write_file already blocks these paths in file_safety.py.
This creates an inconsistency: write_file protects these paths but
terminal shell redirections bypass the same protection. An agent
prompted via indirect injection could install persistent backdoors
(e.g. PATH manipulation, alias overrides) or write credential entries
without user approval.
Extend _SENSITIVE_WRITE_TARGET with two new regex groups matching the
same paths that file_safety.py's WRITE_DENIED_PATHS already covers:
_SHELL_RC_FILES — ~/.bashrc, ~/.zshrc, ~/.profile, ~/.bash_profile,
~/.zprofile
_CREDENTIAL_FILES — ~/.netrc, ~/.pgpass, ~/.npmrc, ~/.pypirc
All 130 existing tests pass.
* fix(curator): authoritative absorbed_into declarations on skill delete
Closes#18671. The classification pipeline that feeds cron-ref rewriting
used to infer consolidation vs pruning from two brittle signals: the
curator model's post-hoc YAML summary block, and a substring heuristic
scanning other tool calls for the removed skill's name. Both miss in
real consolidations — the model forgets the YAML under reasoning
pressure, and the heuristic misses when the umbrella's patch content
describes the absorbed behavior abstractly instead of naming the old
slug. When both miss, the skill falls through to 'no-evidence fallback'
pruned, and #18253's cron rewriter drops the cron ref entirely instead
of mapping it to the umbrella. Same observable symptom as pre-#18253:
'Skill(s) not found and skipped' at the next cron run.
The fix makes the model declare intent at the moment of deletion.
skill_manage(action='delete') now accepts absorbed_into:
- absorbed_into='<umbrella>' -> consolidated, target must exist on disk
- absorbed_into='' -> explicit prune, no forwarding target
- missing -> legacy path, falls through to heuristic/YAML
The curator reconciler reads these declarations off llm_meta.tool_calls
BEFORE either the YAML block or the substring heuristic. Declaration
wins. Fallback logic stays intact for backward compat with any caller
(human or older curator conversation) that doesn't populate the arg.
Changes
- tools/skill_manager_tool.py: add absorbed_into param to skill_manage
+ _delete_skill. Validate target exists when non-empty. Reject
absorbed_into=<self>. Wire through dispatcher + registry + schema.
- agent/curator.py: new _extract_absorbed_into_declarations() walks
tool calls for skill_manage(delete) with the arg. _reconcile_classification
accepts absorbed_declarations= and treats them as authoritative. Curator
prompt updated to require the arg on every delete.
- Tests: 7 new skill_manager tests covering the tool contract (valid
target, empty string, nonexistent target, self-reference, whitespace,
backward compat, dispatcher plumbing). 11 new curator tests covering
the extractor + authoritative reconciler path + mixed-legacy-and-
declared runs.
Validation
- 307/307 targeted tests pass (curator + cron + skill_manager suites).
- E2E #18671 repro: 3 narrow skills, 1 umbrella, cron job referencing
all 3. Model emits NO YAML block. Heuristic misses (patch prose
doesn't name old slugs). Delete calls carry absorbed_into. Result:
both PR skills correctly classified 'consolidated' + cron rewritten
['pr-review-format', 'pr-review-checklist', 'stale-junk'] ->
['hermes-agent-dev']; stale-junk pruned via absorbed_into=''.
- E2E backward-compat: delete without absorbed_into, model emits YAML
-> routed via existing 'model' source, cron still rewritten correctly.
* feat(curator): capture + restore cron skill links across snapshot/rollback
Before this, rolling back a curator run restored the skills tree but cron
jobs still pointed at the umbrella skills the curator had rewritten them
to. The user would see their old narrow skills back on disk but their
cron jobs still configured with the merged umbrella — not actually 'back
to how it was'.
Snapshot side: snapshot_skills() now captures ~/.hermes/cron/jobs.json
alongside the skills tarball, as cron-jobs.json. The manifest gets a new
'cron_jobs' block with {backed_up, jobs_count} so rollback (and the CLI
confirm dialog) can surface what's in the snapshot. If jobs.json is
missing/unreadable/malformed, snapshot proceeds without cron data — the
skills backup is the core guarantee; cron is additive.
Rollback side: after the skills extract succeeds, the new
_restore_cron_skill_links() reconciles the backed-up jobs into the live
jobs.json SURGICALLY. Only 'skills' and 'skill' fields are restored, and
only on jobs matched by id. Everything else about a cron job — schedule,
last_run_at, next_run_at, enabled, prompt, workdir, hooks — is live
state the user or scheduler has modified since the snapshot; overwriting
it would regress unrelated activity.
Reconciliation rules:
- Job in backup AND live, skills differ → skills restored.
- Job in backup AND live, skills match → no-op.
- Job in backup, NOT in live → skipped (user deleted it
after snapshot; their choice
is later than the snapshot).
- Job in live, NOT in backup → untouched (user created it
after snapshot).
- Snapshot missing cron-jobs.json at all → rollback still succeeds,
reports 'not captured'
(older pre-feature snapshots
keep working).
Writes go through cron.jobs.save_jobs under the same _jobs_file_lock the
scheduler uses, so rollback doesn't race tick().
Also:
- hermes_cli/curator.py: rollback confirm dialog now shows
'cron jobs: N (will be restored for skill-link fields only)' when the
snapshot has cron data, or 'not in snapshot (<reason>)' otherwise.
- rollback()'s message string includes a 'cron links: ...' clause
summarizing the reconciliation outcome.
Tests
- 9 new cases: snapshot-with-cron, snapshot-without-cron, malformed-json
captured-as-raw, full rollback-restores-skills-and-cron, rollback
touches only skill fields, rollback skips user-deleted jobs, rollback
leaves user-created jobs untouched, rollback still works with
pre-feature snapshot that has no cron-jobs.json, standalone unit test
on _restore_cron_skill_links exercising the full report shape.
Validation
- 484/484 targeted tests pass (curator + cron + skill_manager suites).
- E2E: real snapshot_skills, real cron rewrite, real rollback. Before:
['pr-review-format', 'pr-review-checklist', 'pr-triage-salvage'].
After curator: ['hermes-agent-dev']. After rollback: ['pr-review-format',
'pr-review-checklist', 'pr-triage-salvage']. Non-skill fields (id,
name, prompt) preserved across the round trip.
Widens #16528 to two sibling sites that had the same quoted-boolean
bug: a YAML string "false" (or "0", "no", "off") silently evaluated
truthy under bool() / if-check.
- gateway/run.py _load_show_reasoning: is_truthy_value wrap
- tools/skill_manager_tool.py _guard_agent_created_enabled: is_truthy_value wrap
- regression tests for both
When running on a host with sudoers NOPASSWD configured for the current
user, interactive Hermes sessions were unnecessarily entering the
password prompt path before executing sudo commands. Outside Hermes,
`sudo -n true` exits 0 for that user.
Add `_sudo_nopasswd_works()` that probes `sudo -n true` and, when it
succeeds, lets `_transform_sudo_command()` return the command unchanged
with no stdin password. The probe:
- Is scoped to the `local` terminal backend only, so Docker/SSH/Modal
and other remote backends do not inherit host sudo state.
- Re-probes every call (no process-lifetime cache) so an expired sudo
timestamp cannot silently make a later command block waiting for a
password that Hermes never prompts for.
- Is bypassed entirely when `SUDO_PASSWORD` is configured or a cached
password already exists, preserving existing explicit-password flows.
Co-authored-by: Junting Wu <juntingpublic@gmail.com>
_capability_cache was a single module-level dict shared across all
tokens. If the bot token rotates or multiple tokens are used in one
process, capabilities detected for token A would be returned for
token B, causing wrong schema gating and incorrect runtime behavior.
Replace the single Optional cache with a Dict keyed by token so each
token gets its own isolated capability entry.
_SupervisorRegistry.get_or_start() returned an existing supervisor
whenever the cdp_url matched, without checking if the supervisor's
thread or event loop was still alive. A crashed supervisor would be
silently reused, causing missed dialog/frame updates.
Now checks both _thread.is_alive() and _loop.is_running() before
returning the cached instance. An unhealthy supervisor is torn down
and recreated, matching the existing URL-changed code path.
- order session_search recent-mode results by last activity instead of session start time
- add an opt-in `order_by_last_active` path to `SessionDB.list_sessions_rich`
- add regression coverage for both the database ordering and recent-mode call path
Treat skill views and edits as activity when curator reports and applies lifecycle transitions, so recently loaded or patched skills are not displayed or transitioned as never used.\n\nAdds regression tests for activity derivation, automatic transitions, and CLI status output.
restore_skill() in tools/skill_usage.py used archive_root.iterdir(), which
only walked the top level of .archive/. Skills archived under nested layouts
(e.g. .archive/openclaw-imports/<skill>/ from older archive paths or
external imports) were invisible to both the exact-match and prefix-match
candidate scans, surfacing as a misleading "skill '<name>' not found in
archive" error even though the directory existed on disk.
Switch both candidate scans to archive_root.rglob('*') so the lookup
descends into category subdirectories.
Fixes#17942
Widen #17818 to cover the dominant 'agent actively used this skill' path:
when the model calls the skill_view tool, bump use_count alongside view_count.
The slash-command and --skill preload paths (covered by the cherry-picked
commit) only catch user-initiated invocation; most skill activation happens
via the agent calling skill_view to consume an indexed skill.
Curator's stale-timer keys off last_used_at (agent/curator.py:233), so
without this wire-up agent-created skills would transition to stale
simultaneously regardless of actual use.
Widen #17639 to the fourth sibling site (tools/skills_tool.py _EXCLUDED_SKILL_DIRS)
and register leoneparise in scripts/release.py AUTHOR_MAP so CI release script
resolves the contributor.
Adds a new `send_multiple_images` method to the ``BasePlatformAdapter``
that implements the default "One image per message" loop and allows for
platform-specific overriding.
Implements such an override for the Signal adapter, batching images
and trying (best-effort) to work around rate-limits for voluminous
batches using a specific scheduler.
Also implements batching + rate-limit handling in the `send_message`
tool.
New tests added for the Signal adapter, its rate-limit scheduler and the
`send_message` tool
The sandbox-side `_call()` in both the UDS and file-based transports was
not thread-safe, so scripts that call tools from multiple threads (e.g.
`ThreadPoolExecutor` over `terminal()`) inside a single `execute_code`
run could silently receive each other's responses.
Root cause:
* UDS transport — a single module-level `_sock` was shared across all
threads; the newline-framed protocol has no request-id; and the
server-side RPC loop handles one connection serially. With concurrent
callers, each thread would `sendall()` then race to `recv()` the next
newline-terminated response from the shared buffer, so responses got
delivered to the wrong caller.
* File transport — `_seq += 1` is a non-atomic read-modify-write, so
two threads could allocate the same sequence number and clobber each
other's request/response files.
Fix: guard `_call()` with a `threading.Lock` in the UDS case (covering
send+recv), and guard `_seq` allocation with a lock in the file case.
No protocol change.
Regression tests cover both the generated-source level (lock is present
and used) and an end-to-end concurrency test: running a sandboxed
ThreadPoolExecutor of 10 `terminal()` calls against a slow mock
dispatcher, asserting every caller sees its own tagged response. The
test fails without the fix (10/10 mismatched, matching real-world
repro) and passes with it.
tar xf - -C / extracts the staging directory tree to the remote root.
GNU tar default behavior overwrites metadata (including mode) of existing
directories. When the local umask is 002 (Ubuntu default), the staging
dirs are 0775, and tar chmod's /home/<user> to 0775 — breaking sshd
StrictModes which requires 0755 or stricter for home dirs.
Add --no-overwrite-dir to the remote tar command so existing directory
metadata is preserved.
Fixes#17767
Piper (OHF-Voice/piper1-gpl) is a fast, local neural TTS engine from the
Home Assistant project that supports 44 languages with zero API keys.
Adds it as a native built-in provider alongside edge/neutts/kittentts,
installable via 'hermes tools' with one keystroke.
What ships:
- New 'piper' built-in provider in tools/tts_tool.py
- Lazy import via _import_piper()
- Module-level voice cache keyed on (model_path, use_cuda) so switching
voices doesn't invalidate older cached voices
- _resolve_piper_voice_path() accepts either an absolute .onnx path or a
voice name (auto-downloaded on first use via 'python -m
piper.download_voices --download-dir <cache>')
- Voice cache at ~/.hermes/cache/piper-voices/ (profile-aware via
get_hermes_dir)
- Optional SynthesisConfig knobs: length_scale, noise_scale,
noise_w_scale, volume, normalize_audio, use_cuda — passed through
only when configured, so older piper-tts versions aren't broken
- WAV output then ffmpeg conversion path (same as neutts/kittentts) so
Telegram voice bubbles work when ffmpeg is present
- Piper added to BUILTIN_TTS_PROVIDERS so a user's
tts.providers.piper.command cannot shadow the native provider
(regression test included)
- 'hermes tools' wizard entry
- Piper appears under Voice and TTS as local free, with
'pip install piper-tts' auto-install via post_setup handler
- Prints voice-catalog URL and default-voice info after install
- config.yaml defaults
- tts.piper.voice defaults to en_US-lessac-medium
- Commented advanced knobs for discoverability
- Docs
- New 'Piper (local, 44 languages)' section in features/tts.md
explaining install path, voice switching, pre-downloaded voices,
and advanced knobs
- Piper listed in the ten-provider table and ffmpeg table
- Custom-command-providers section updated to drop the Piper example
(now native) and add a piper-custom example for users with their own
trained .onnx models
- overview.md bumps provider count to ten
- Tests (tests/tools/test_tts_piper.py, 16 tests)
- Registration (BUILTIN_TTS_PROVIDERS, PROVIDER_MAX_TEXT_LENGTH)
- _resolve_piper_voice_path across every branch: direct .onnx path,
cached voice name, fresh download with correct CLI args, download
failure, successful-exit-but-missing-files, empty voice to default
- _generate_piper_tts: loads voice once, reuses cache, voice-name
download wiring, advanced knobs flow through SynthesisConfig
- text_to_speech_tool end-to-end dispatch and missing-package error
- check_tts_requirements: piper availability toggles the return value
- Regression guard: piper cannot be shadowed by a command provider
with the same name
- Pre-existing test_tts_mistral test broadened to mock the new
piper/kittentts/command-provider checks (otherwise it false-passes
when piper is installed in the test venv)
E2E verification (live):
Actual pip install piper-tts, config piper + en_US-lessac-low,
text_to_speech_tool call, voice auto-downloaded from HuggingFace,
WAV synthesized, ffmpeg-converted to Ogg/Opus. Second call hits the
cache (~60ms). Cache dir populated with .onnx and .onnx.json.
This caught a real bug during development: the first pass used '-d' as
the download-dir flag; the actual piper.download_voices CLI wants
'--download-dir'. Fixed before PR opened.
Reshape of PR #17211 (@versun). Lets users wire any local or external
TTS CLI into Hermes without adding engine-specific Python code. Users
declare any number of named providers in config.yaml and switch between
them with tts.provider: <name>, alongside the built-ins (edge, openai,
elevenlabs, …).
Config shape:
tts:
provider: piper-en
providers:
piper-en:
type: command
command: 'piper -m ~/model.onnx -f {output_path} < {input_path}'
output_format: wav
Placeholders: {input_path}, {text_path}, {output_path}, {format},
{voice}, {model}, {speed}. Use {{ / }} for literal braces.
Key behavior:
- Built-in provider names always win — a tts.providers.openai entry
cannot shadow the native OpenAI provider.
- type: command is the default when command: is set.
- Placeholder values are shell-quote-aware (bare / single / double
context), so paths with spaces and shell metacharacters are safe.
- Default delivery is a regular audio attachment. voice_compatible: true
opts in to Telegram voice-bubble delivery via ffmpeg Opus conversion.
- Command failures (non-zero exit, timeout, empty output) surface to
the agent with stderr/stdout included so you can debug from chat.
- Process-tree kill on timeout (Unix killpg, Windows taskkill /T).
- max_text_length defaults to 5000 for command providers; override
under tts.providers.<name>.max_text_length.
Tests: tests/tools/test_tts_command_providers.py — 42 new tests cover
provider resolution, shell-quote context, placeholder rendering with
injection payloads, timeout, non-zero exit, empty output, voice_compatible
opt-in, and end-to-end dispatch through text_to_speech_tool. All 88
pre-existing TTS tests still pass.
Docs: new "Custom command providers" section in
website/docs/user-guide/features/tts.md with three worked examples
(Piper, VoxCPM, MLX-Kokoro), placeholder reference, optional keys,
behavior notes, and security caveat.
E2E-verified live: isolated HERMES_HOME, command provider declared in
config.yaml, text_to_speech_tool dispatches through the registered
shell command and the output file is produced as expected.
Co-authored-by: Versun <me+github7604@versun.org>
Extracted from PR #17211 (@versun) so it can land independently of the
local_command TTS provider redesign.
- Add should_send_media_as_audio(platform, ext, is_voice) in
gateway/platforms/base.py; single source of truth for audio routing.
- Add .flac to recognized audio extensions (MEDIA regex, weixin audio
set, send_message audio set).
- Telegram send_voice() now falls back to send_document for formats
Telegram's Bot API can't play natively (.wav, .flac, ...) instead of
raising; MP3/M4A still go to sendAudio, Opus/OGG still go to sendVoice.
- Route _send_telegram() in send_message_tool through a narrower
_TELEGRAM_SEND_AUDIO_EXTS = {.mp3, .m4a} set.
- cron.scheduler._send_media_via_adapter now delegates the audio
decision to should_send_media_as_audio so it matches the gateway.
- Update the cron live-adapter ogg test to flag [[audio_as_voice]] so
it still routes to sendVoice under the new Telegram-specific policy.
- Tests: unit coverage for should_send_media_as_audio across platforms,
end-to-end MEDIA routing via _process_message_background and
GatewayRunner._deliver_media_from_response, TelegramAdapter.send_voice
fallback for FLAC/WAV.
Co-authored-by: Versun <me+github7604@versun.org>
PR #17660 landed a sweep of CI fixes but left three loose ends:
1. tests/cli/test_cli_loading_indicator.py::test_reload_mcp_sets_busy_state_
and_prints_status — /reload-mcp gained a prompt-cache-invalidation
confirmation (commit 4d7fc0f37) that was never wired into this test.
The test exercises the loading-indicator path, so pre-approve via
config and go straight into _reload_mcp().
2. tools/mcp_tool.py _make_tool_handler — the added
getattr(server, '_rpc_lock', None) + 'skip the lock if missing'
branch is inconsistent with four sibling call sites that still
direct-access server._rpc_lock. The lock is guaranteed by
MCPServerTask.__init__; falling through to an unlocked
session.call_tool would silently serialize-strip RPCs if the guard
ever triggered. Restore direct access.
3. tui_gateway/server.py _messages_as_conversation — the helper
existed only to catch 'TypeError: include_ancestors unexpected'
from mocked SessionDBs that don't actually exist. The real
SessionDB.get_messages_as_conversation has accepted
include_ancestors since introduction, and every test FakeDB in
the repo already declares the kwarg. Remove the shim, inline the
two call sites.
feat(gateway): refine Platform._missing_ and platform-connected dispatch
Restricts plugin-name acceptance to bundled plugin scan + registry
(no arbitrary string -> enum-pollution), pulls per-platform connectivity
checks into a _PLATFORM_CONNECTED_CHECKERS lambda map with a clean
_is_platform_connected method, and adds tests covering the checker map,
plugin platform interface, and IRC setup wizard.
Extends the platform plugin interface from Phase 1 to cover every
touchpoint where built-in platforms have hardcoded behavior.
- allowed_users_env / allow_all_env: per-platform auth env vars
- max_message_length: smart-chunking for send_message tool
- pii_safe: session PII redaction flag
- emoji: CLI/gateway display
- allow_update_command: /update access control
send_message tool (tools/send_message_tool.py):
- Replaced hardcoded platform_map dict with Platform() call
- Added _send_via_adapter() for plugin platforms — routes through
live gateway adapter when available
- Registry-aware max message length for smart chunking
Cron delivery (cron/scheduler.py):
- Replaced hardcoded 15-entry platform_map with Platform() call
- Plugin platforms now work as cron delivery targets
User authorization (gateway/run.py _is_user_authorized):
- Registry fallback: checks PlatformEntry.allowed_users_env and
allow_all_env when platform not in hardcoded maps
- Plugin platforms get per-platform auth support
_UPDATE_ALLOWED_PLATFORMS: checks registry allow_update_command flag
Channel directory: includes plugin platforms in session enumeration
Orphaned config warning: descriptive message when plugin platform is
in config but no plugin registered it
Gateway weakref: _gateway_runner_ref for cross-module adapter access
hermes status: shows plugin platforms with (plugin) tag
hermes gateway setup: plugin platforms appear in menu with setup hints
hermes_cli/platforms.py: get_all_platforms() merges with registry,
platform_label() falls back to registry for plugin names
- 8 new tests (extended fields, cron resolution, platforms merge)
- Updated 3 tests for new Platform() based resolution
- 2829 passed, 24 pre-existing failures, zero new failures
Reloading MCP servers rebuilds the tool set for the active session, which
invalidates the provider prompt cache (tool schemas are baked into the
system prompt). The next message re-sends full input tokens — can be
expensive on long-context or high-reasoning models.
To surface that cost, /reload-mcp now routes through a new slash-confirm
primitive with three options: Approve Once / Always Approve / Cancel.
'Always Approve' persists approvals.mcp_reload_confirm: false so future
reloads run silently.
Coverage:
* Classic CLI (cli.py) — interactive numbered prompt.
* TUI (tui_gateway + Ink ops.ts) — text warning on first call; `now` /
`always` args skip the gate; `always` also persists the opt-out.
* Messenger gateway — button UI on Telegram (inline keyboard), Discord
(discord.ui.View), Slack (Block Kit actions); text fallback on every
other platform via /approve /always /cancel replies intercepted in
gateway/run.py _handle_message.
* Config key: approvals.mcp_reload_confirm (default true).
* Auto-reload paths (CLI file watcher, TUI config-sync mtime poll) pass
confirm=true so they do NOT prompt.
Implementation:
* tools/slash_confirm.py — module-level pending-state store used by all
adapters and by the CLI prompt. Thread-safe register/resolve/clear.
* gateway/platforms/base.py — send_slash_confirm hook (default 'Not
supported' → text fallback).
* gateway/run.py — _request_slash_confirm helper + text intercept in
_handle_message (yields to in-progress tool-exec approvals so
dangerous-command /approve still unblocks the tool thread first).
Tests:
* tests/tools/test_slash_confirm.py — primitive lifecycle + async
resolution + double-click atomicity (16 tests).
* tests/hermes_cli/test_mcp_reload_confirm_gate.py — default-config
shape + deep-merge preserves user opt-out (5 tests).
Targeted runs (hermetic): 89 passed (slash-confirm, config gate,
existing agent cache, existing telegram approval buttons).
Salvage-follow-up to @shannonsands's /reload-skills PR. Trims the feature to
match the design: user-initiated rescan, no prompt-cache reset, no new
schema surface, no phantom user turn, and the next-turn note carries each
added/removed skill's 60-char description (not just its name).
Changes vs the original PR:
* Drop the in-process skills prompt-cache clear in reload_skills(). Skills
are invoked at runtime via /skill-name, skills_list, or skill_view —
they don't need to live in the system prompt for the model to use them.
Keeping the cache intact preserves prefix caching across the reload so
/reload-skills pays no cache-reset cost. (MCP has to break the cache
because tool schemas must be known at conversation start; skills do not.)
* Drop the skills_reload agent tool and SKILLS_RELOAD_SCHEMA from
tools/skills_tool.py, plus the four skills_reload enumerations in
toolsets.py. No new schema surface — agents can already see a freshly-
installed skill via skill_view / skills_list the moment it's on disk.
* Replace the phantom 'role: user' turn injection with a one-shot queued
note. CLI uses self._pending_skills_reload_note (same pattern as
_pending_model_switch_note, prepended to the next API call and cleared).
Gateway uses self._pending_skills_reload_notes[session_key]. The note
is prepended to the NEXT real user message in this session, so message
alternation stays intact and nothing out-of-band is persisted to the
transcript.
* reload_skills() now returns added/removed as
[{'name': str, 'description': str}, ...] (description truncated to 60
chars — matches the curator / gateway adapter budget). The injected
next-turn note formats each entry as 'name — description' so the model
can actually reason about which new skills to call without running
skills_list first.
* Only emit the note when the diff is non-empty. On empty diff, print
'No new skills detected' and do nothing else.
* Tests rewritten to cover the queue semantics, the description payload,
and a regression guard that the prompt-cache snapshot is preserved.
Adds a public reload path for the in-process skill caches so newly
installed (or removed) skills become visible mid-session without a
gateway restart. Mirrors the shape of /reload-mcp.
Three surfaces:
* /reload-skills slash command — CLI (cli.py) and gateway (gateway/run.py),
with /reload_skills alias for Telegram autocomplete and an explicit
Discord registration.
* skills_reload agent tool (tools/skills_tool.py) — lets agents/subagents
pick up freshly-installed skills via tool call.
* agent.skill_commands.reload_skills() — shared helper that clears
_skill_commands, _SKILLS_PROMPT_CACHE (in-process LRU), and the
on-disk .skills_prompt_snapshot.json, then returns an added/removed
diff plus the new total count.
Tested:
* tests/agent/test_skill_commands_reload.py (9 cases)
* tests/cli/test_cli_reload_skills.py (3 cases)
* tests/gateway/test_reload_skills_command.py (4 cases)
Use case: NemoClaw / OpenShell-style sandboxed orchestrators that drop
skills into ~/.hermes/skills mid-session, plus agentic flows where the
agent itself installs a skill via the shell tool and needs it bound
without a gateway restart. The Python helper
clear_skills_system_prompt_cache(clear_snapshot=True) already exists
internally — this PR just exposes it via slash command and tool.
vision_analyze used Path('./temp_vision_images') — a relative path that
resolved against cwd. Under Docker the image's WORKDIR is /opt/hermes,
which is root-owned and only chmoded a+rX (read + traversal). Since
#5811 landed (run as non-root hermes UID 10000, Apr 12), remote-URL
vision calls fail with PermissionError on mkdir.
Switch to get_hermes_dir('cache/vision', 'temp_vision_images'): resolves
to $HERMES_HOME/cache/vision/ (= /opt/data/cache/vision/ in Docker —
the user-owned volume mount). Existing installs with the old dir keep
using it via the get_hermes_dir back-compat path; no migration needed.
Only site in the codebase that stored runtime files via Path('./...').
Reported via Discord: https://juick.com/i/p/3089079.jpg → Telegram →
gateway → [Errno 13] Permission denied: 'temp_vision_images'.
Extend curator's pin flag from 'skip auto-transitions' to 'no agent
edits at all'. All five skill_manage mutation actions (edit, patch,
delete, write_file, remove_file) now refuse pinned skills with a
message pointing the user at `hermes curator unpin <name>`.
Motivation: pin used to only stop the curator's own maintenance pass
from touching a skill. Nothing prevented the main agent from editing
or deleting a pinned skill via skill_manage in-session. This gives
users a hard fence against unwanted agent edits — same semantics as
curator pinning, extended to the write tool.
Create is unaffected (you can't pin a name that doesn't exist yet,
and name collisions already error out). Broken sidecars fail open
rather than lock the agent out.
The schema description advertises the new refusal so models know
not to route around it with rename/recreate tricks.
Closes#4759, closes#4381.
Mutating actions (patch, edit, write_file, remove_file, delete) used to
refuse skills that lived under `skills.external_dirs` with 'Skill X is in
an external directory and cannot be modified. Copy it to your local skills
directory first.' Faced with that error, the agent would fall back to
action='create', which always writes under ~/.hermes/skills/ — producing
a silent duplicate of the external skill in the local store.
Fix: drop the read-only gate. `skills.external_dirs` is configured by the
user; if they pointed it at a directory, they already said 'these are my
skills, treat them the same.' Filesystem permissions handle the genuine
read-only case (write fails, agent sees the error).
- New _containing_skills_root() resolves whichever dir actually contains
the skill; _delete_skill uses it to bound empty-category cleanup so an
external root is never rmdir'd.
- _create_skill behavior is unchanged: new skills still land in local
SKILLS_DIR only. Fewer moving parts.
- Seven new TestExternalSkillMutations tests covering patch/edit/write_file/
remove_file/delete/create against a mocked two-root layout + a category
rmdir-safety check.
Adds Vercel Sandbox as a supported Hermes terminal backend alongside
existing providers (Local, Docker, Modal, SSH, Daytona, Singularity).
Uses the Vercel Python SDK to create/manage cloud microVMs, supports
snapshot-based filesystem persistence keyed by task_id, and integrates
with the existing BaseEnvironment shell contract and FileSyncManager
for credential/skill syncing.
Based on #17127 by @scotttrinh, cherry-picked onto current main.
The cron schema contracts deliver as a string ("local", "origin",
"telegram", "telegram:chat_id[:thread_id]", or comma-separated combos),
but MCP clients and scripts sometimes pass an array like ['telegram'].
Before this change, the list was written to jobs.json verbatim, and
the scheduler's str(deliver).split(',') then tried to resolve the
literal string "['telegram']" as a platform — returning None and
logging 'no delivery target resolved for deliver=[\'telegram\']'.
Fix on both ends:
- tools/cronjob_tools.py: normalize deliver at the API boundary on
create and update, so storage is always a string.
- cron/scheduler.py: normalize deliver in _resolve_delivery_targets,
so existing jobs.json entries with list-form deliver are handled
gracefully without requiring users to edit the file.
Closes#17139
Widen #17163 to the sibling file tools/transcription_tools.py, which had
the same class of bug. STT provider call sites and the _get_provider
selection gate called os.getenv(...) directly and missed keys that only
lived in ~/.hermes/.env.
Same pattern as tts_tool.py: one guarded top-level import of
get_env_value (falls back to os.getenv on ImportError), then every
API-key and paired-base-URL lookup swapped over.
Call sites migrated:
- _transcribe_groq — GROQ_API_KEY
- _transcribe_mistral — MISTRAL_API_KEY
- _transcribe_xai — XAI_API_KEY, XAI_STT_BASE_URL
- _get_provider — GROQ/MISTRAL/XAI_API_KEY in explicit + auto branches
Module-level defaults (DEFAULT_STT_MODEL, GROQ_BASE_URL, etc.) stay on
os.getenv — they're import-time constants, not runtime config, and the
dotenv fallback would add no value there.
New regression tests in tests/tools/test_transcription_dotenv_fallback.py
(8 cases) mirror briandevans' TTS tests: per-provider dotenv-key
forwarding, selection-gate dotenv visibility, and an end-to-end probe
that patches hermes_cli.config.load_env to simulate ~/.hermes/.env
carrying the key while os.environ does not.
Wrap the new top-level `from hermes_cli.config import get_env_value`
in try/except ImportError and fall back to a thin os.getenv shim, so
importing tools.tts_tool keeps working in environments where
hermes_cli.config is unavailable. This matches the existing tolerance
in `_load_tts_config()` (tools/tts_tool.py) and the same
import-fallback pattern in tools/tool_backend_helpers.py::fal_key_is_configured.
Also update the TestDotenvFallbackPerProvider docstring to accurately
describe the mocking strategy: per-provider tests patch
`tools.tts_tool.get_env_value` directly, while the regression-guard
tests cover the lower-level `hermes_cli.config.load_env` integration.
Addresses Copilot review on #17163.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
TTS provider tools (elevenlabs, xai, minimax, mistral, gemini) called
os.getenv("X_API_KEY") directly, which bypassed Hermes's dotenv bridge in
hermes_cli.config. Users who keep their TTS keys only in ~/.hermes/.env saw
"X_API_KEY not set" errors even though the rest of the stack
(agent/credential_pool, hermes_cli/auth) already resolves keys through
get_env_value() — same class of bug as #15914 fixed for those modules.
Switch every TTS env-var lookup (API keys, base URLs, and
check_tts_requirements gates) to get_env_value, which checks os.environ
first and then ~/.hermes/.env. Behaviour for users with keys exported in
the shell is unchanged; users with dotenv-only keys now succeed. The two
diagnostics prints in __main__ are migrated for consistency.
Regression test (tests/tools/test_tts_dotenv_fallback.py):
- per-provider: each backend reads the dotenv key when only
~/.hermes/.env carries it (5 providers).
- end-to-end: with hermes_cli.config.load_env returning the key and
os.environ empty, _generate_minimax_tts and check_tts_requirements
both succeed; reverting tools/tts_tool.py back to os.getenv makes all
7 tests fail with "MINIMAX_API_KEY not set" / similar.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
_update_cwd() uses a bare open(self._cwd_file).read() that never
closes the file descriptor. This method runs on every terminal
command execution, so the fd leaks accumulate in long sessions.
Use a with statement so the fd is released promptly.
Fixes#15552 (standalone resubmission)
When a background terminal process spawns a descendant daemon that
inherits the stdout pipe (e.g. 'hermes update' triggering a gateway
systemctl restart), the reader thread's stdout.read() never returns EOF
and its finally: block never runs. session.exited stays False forever,
so process(action='poll') returns 'running' indefinitely even though
the direct child exited long ago.
Issue #17327: Feishu user polled 74 times over 7 minutes before killing
the gateway manually.
Fix: add _reconcile_local_exit() that checks the direct Popen.poll()
before trusting session.exited. If the direct child has exited, drain
any immediately-readable bytes non-blocking and flip session.exited.
Called from poll() and wait(). The stuck reader thread remains blocked
but is a daemon thread and gets reaped with the process.
Safe no-op for env/PTY sessions, already-exited sessions, and live
children (returns None from Popen.poll()).
_send_yuanbao() already supported media_files= and the user-facing
error strings already advertised yuanbao support, but there was no
dispatch branch in _send_to_platform() actually routing to it. Target
yuanbao in send_message previously fell through to
"Direct sending not yet implemented".
- Add yuanbao media-chunk branch (mirrors Signal/Matrix: media on
final chunk only).
- Add yuanbao elif in the non-media loop.
Salvage of #17411; SKILL.md description change and redundant
sidebars.ts entry dropped, indentation/trailing-whitespace cleaned up.
The "cfg.get('X', {}).get('Y', default)" pattern appears 50+ times
across tools/, gateway/, and plugins/. Each call site manually handles
the same three gotchas:
1. Missing intermediate key → empty dict → chain works
2. Non-dict value at intermediate position → AttributeError
(uncaught in most sites, so a misconfigured YAML crashes the tool)
3. cfg is None → AttributeError
Introduces cfg_get(cfg, *keys, default=None) in hermes_cli/config.py
as the canonical helper. Handles all three uniformly, returns default
only when the final key is *absent* (matches dict.get semantics —
explicit None values are preserved, falsy values like 0 / False / ''
are preserved).
Named cfg_get rather than cfg_path to avoid shadowing the existing
'cfg_path = _hermes_home / "config.yaml"' local variable that appears
in gateway/run.py, cron/scheduler.py, hermes_cli/main.py, etc.
Migrated 20 call sites as the first-batch proof-of-value:
gateway/run.py 10 sites (agent/display subtrees)
tools/browser_tool.py 3 sites
tools/vision_tools.py 2 sites
tools/browser_camofox.py 1 site
tools/approval.py 1 site
tools/skills_tool.py 1 site
tools/skill_manager_tool.py 1 site
tools/credential_files.py 1 site
tools/env_passthrough.py 1 site
The remaining ~30 sites across plugins/ and smaller tool files can be
migrated opportunistically — the helper is now available and the
pattern is established.
Fixed a latent bug along the way: tools/vision_tools.py had its
cfg_get usage at line 560 inside a function that locally re-imports
'from hermes_cli.config import load_config', but the AST-based
migration script wrote the top-level cfg_get import to a different
function scope, leaving line 560's cfg_get as a NameError silently
swallowed by the surrounding try/except. Test
test_vision_uses_configured_temperature_and_timeout caught it. Fixed
by including cfg_get in the function-local import.
Verified:
- 7880/7893 tests/tools/ + tests/gateway/ + tests/hermes_cli/test_config
tests pass; all 13 failures pre-existing on main (MCP, delegate,
session_split_brain — verified earlier in the sweep).
- All 20 migrated sites AST-verified to have cfg_get in scope (either
module-level or function-local).
- Live 'hermes chat' smoke: 2 turns + /model switch + tool calls +
/quit, zero errors. Agent correctly counted 20 cfg_get hits across
8 tool files — matching the migration.
Semantic parity verified against the original pattern across 8 edge
cases (missing keys, None values, falsy values, empty strings, string
instead of dict, None cfg, nested levels).
Add opt-in terminal.docker_run_as_host_user config flag that passes
--user $(id -u):$(id -g) to the Docker backend so files written into
bind-mounted directories (/workspace, /root, docker_volumes entries) are
owned by the host user instead of root.
When enabled on POSIX platforms, also drops SETUID/SETGID caps since the
container no longer needs gosu/su to switch users. Falls back cleanly on
platforms without os.getuid (e.g. native Windows Docker) with a warning.
Wired through all three config.yaml -> TERMINAL_* env-var bridges:
- cli.py env_mappings (CLI + TUI startup)
- gateway/run.py _terminal_env_map (gateway / messaging platforms)
- hermes_cli/config.py _config_to_env_sync (`hermes config set`)
Also fixes docker_mount_cwd_to_workspace silently failing in gateway
mode -- it was missing from gateway/run.py's _terminal_env_map.
Adds tests/tools/test_terminal_config_env_sync.py to guard against
future drift between the three bridges (same bug class shipped twice
in one month).
Bundled Hermes image won't work with this flag since its entrypoint
expects to start as root for the usermod/gosu hermes flow; works with
the default nikolaik/python-nodejs image and plain Debian/Ubuntu.
Previous invariants only gated the primary entry points
(apply_automatic_transitions, archive_skill, CLI pin). Several paths
were unprotected:
- bump_view / bump_use / bump_patch / set_state / set_pinned wrote
usage records unconditionally, which is confusing noise in
.usage.json even though the review list filtered them out
- restore_skill did not check whether a bundled skill now shadows
the archived name
- CLI unpin was asymmetric with CLI pin — it had no gate
Fixes:
- _mutate() (the shared counter / state writer) now drops silently
when the skill is not agent-created. .usage.json never gains a
record for a bundled or hub-installed skill.
- restore_skill() refuses to restore under a name that is now
bundled or hub-installed (would shadow upstream).
- CLI unpin gate matches CLI pin.
New tests:
- 5 provenance-guard tests on skill_usage (one per mutator)
- 1 end-to-end test that hammers every mutator at a bundled skill
and a hub skill, asserts both are untouched on disk, and asserts
the sidecar stays clean
- 2 CLI tests proving pin/unpin refuse bundled skills symmetrically
64/64 tests passing (29 skill_usage + 27 curator + 8 new guards).
Adds the Curator — an auxiliary-model background task that periodically
reviews AGENT-CREATED skills and keeps the collection tidy: tracks usage,
transitions unused skills through active → stale → archived, and spawns
a forked AIAgent to consolidate overlaps and patch drift.
Default: enabled, inactivity-triggered (no cron daemon). Runs on CLI
startup and gateway boot when the last run is older than interval_hours
(default 24) AND the agent has been idle for min_idle_hours (default 2).
Invariants (all load-bearing):
- Never touches bundled or hub-installed skills (.bundled_manifest +
.hub/lock.json double-filter)
- Never auto-deletes — archive only. Archives are recoverable
via `hermes curator restore <skill>`
- Pinned skills bypass all auto-transitions
- Uses the aux client; never touches the main session's prompt cache
New files:
- tools/skill_usage.py — sidecar .usage.json telemetry, atomic writes,
provenance filter
- agent/curator.py — orchestrator: config, idle gating, state-machine
transitions (pure, no LLM), forked-agent review prompt
- hermes_cli/curator.py — `hermes curator {status,run,pause,resume,
pin,unpin,restore}` subcommand
- tests/tools/test_skill_usage.py — 29 tests
- tests/agent/test_curator.py — 25 tests
Modified files (surgical patches):
- tools/skills_tool.py — bump view_count on successful skill_view
- tools/skill_manager_tool.py — bump patch_count on skill_manage
patch/edit/write_file/remove_file; forget record on delete
- hermes_cli/config.py — add curator: section to DEFAULT_CONFIG
- hermes_cli/commands.py — add /curator CommandDef with subcommands
- hermes_cli/main.py — register `hermes curator` subparser via
register_cli() from hermes_cli.curator
- cli.py — /curator slash-command dispatch + startup hook
- gateway/run.py — gateway-boot hook (mirrors CLI)
Validation:
- 54 new tests across skill_usage + curator, all passing in 3s
- 346 tests across all touched files' neighbors green
- 2783 tests across hermes_cli/ + gateway/test_run_progress_topics.py green
- CLI smoke: `hermes curator status/pause/resume` work end-to-end
Companion to PR #16026 (class-first skill review prompt) — together
they form a loop: the review prompt stops near-duplicate skill creation
at the source, and the curator prunes/consolidates what still accumulates.
Refs #7816.
Commit 3c42064e made config.yaml the single source of truth for
TERMINAL_CWD, but the config bridge passes cwd values verbatim to
os.environ. When a user sets terminal.cwd: ~/ in config.yaml, the
literal string '~/'' reaches subprocess.Popen, which the kernel
rejects because it does not expand shell tilde syntax.
This patch adds three defensive layers:
1. gateway/run.py — expanduser at config bridge time so TERMINAL_CWD
is always an absolute path.
2. tools/terminal_tool.py — expanduser when reading TERMINAL_CWD in
_get_env_config(), guarding against stale or manually-set env vars.
3. tools/environments/local.py — expanduser in LocalEnvironment before
passing cwd to subprocess.Popen, the final safety net.
Includes regression tests in test_config_cwd_bridge.py for nested
terminal.cwd, top-level cwd alias, and precedence ordering.
Refs: 3c42064e
init_session() runs a login shell bootstrap that sources profile scripts
(.bashrc, .bash_profile, etc.) before capturing pwd. If any profile
script changes the working directory, the captured cwd overwrites the
configured terminal.cwd value — so terminal commands run in the wrong
directory despite the TUI banner showing the configured path.
Add an explicit 'builtin cd' to the configured cwd in the bootstrap
script, after profile sourcing but before pwd capture, ensuring the
configured terminal.cwd is always what gets recorded.
Fixes#14044
TUI session readiness was still laggy after the gateway-ready fixes. Profiling
session.create -> session.info showed the slow phase is background AIAgent
construction (~1.1s). A cProfile run of tui_gateway.server::_make_agent showed
model_tools/tool discovery importing tools.code_execution_tool, whose
module-level EXECUTE_CODE_SCHEMA calls _get_execution_mode(), which imported
cli.CLI_CONFIG.
That pulled the classic interactive CLI stack (prompt_toolkit/Rich and REPL
setup) into every agent startup path, including hermes --tui where it is not
used. Replace that with hermes_cli.config.read_raw_config(), which is cached and
reads only the raw code_execution section. Existing defaults still apply when
the key is absent.
Measurements on macOS Terminal.app:
- import run_agent: ~466ms -> ~347ms
- model_tools import: ~418ms -> ~272ms
- _make_agent: ~1452ms -> ~1239ms
- session.create -> session.info: ~1167ms -> ~999ms
- full hermes --tui ready p50: ~1655ms -> ~1537ms
Tests:
- scripts/run_tests.sh tests/tools/test_code_execution_modes.py tests/tools/test_code_execution.py
detect_dangerous_command() and detect_hardline_command() were calling
re.search(pattern, text, re.IGNORECASE | re.DOTALL) inline — Python's
re._cache (512 patterns) amortizes compile cost on the warm path, but:
1. The first terminal() call per process pays the full compile fan-out
for all 59 patterns (12 HARDLINE + 47 DANGEROUS). Measured at
~2.6 ms per detect_dangerous_command() call after re.purge().
2. The re._cache is LRU — unrelated regex work elsewhere in the agent
(response parsing, text normalization, etc.) can evict our patterns
and silently re-compile them on the next terminal() call.
Precompiling at module load eliminates both costs:
detect_dangerous_command:
cold 2.613 ms → 0.298 ms (-88%)
warm 0.042 ms → 0.004 ms (-90%)
detect_hardline_command:
cold ~0.6 ms → 0.006 ms
warm 0.011 ms → 0.002 ms
Savings are per terminal() call. Agents with heavy terminal use see
compound savings; the bigger value is the stability guarantee (no
re._cache eviction can silently re-introduce the 2.6 ms cold cost
mid-session).
Implementation:
- HARDLINE_PATTERNS_COMPILED and DANGEROUS_PATTERNS_COMPILED built at
module load from the existing (pattern, description) tuples, using
shared _RE_FLAGS = re.IGNORECASE | re.DOTALL.
- detect_* functions now iterate the compiled list and call pattern_re.search(text).
- Original HARDLINE_PATTERNS and DANGEROUS_PATTERNS lists kept as-is
(other code in the file uses them for key derivation /
_PATTERN_KEY_ALIASES).
Verified:
- 160/161 tests/tools/test_approval*.py pass (1 pre-existing heartbeat
test flake on main).
- 349/349 tests/tools/ 'approval or terminal or dangerous' pass.
- Live hermes chat smoke: 3 benign terminal commands + 1 rm -rf /tmp/
(clarify prompt fired — approval path still works) + 1 sudo (sudo
password prompt fired — DANGEROUS pattern match still works). 23
log lines in the smoke window, zero errors.
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
Two amplifying optimizations to per-turn overhead in the gateway:
1. get_tool_definitions() memoization (model_tools.py)
Keyed on (frozenset(enabled), frozenset(disabled),
registry._generation, config.yaml mtime+size). Only active when
quiet_mode=True (which is every hot-path caller — gateway,
AIAgent.__init__); quiet_mode=False keeps the existing print side
effects. Cached path returns a shallow-copy list sharing read-only
schema dicts.
Measured: 7.5 ms → 0.01 ms per call (~750× speedup). Gateway
constructs fresh AIAgent per message, so this saves ~7 ms/turn before
any LLM work.
2. check_fn() TTL cache (tools/registry.py)
check_fn callables like check_terminal_requirements probe external
state (Docker daemon, Modal SDK, playwright binary). For a long-lived
process, hitting them on every get_definitions() pass was pure waste
— external state changes on human timescales. 30 s TTL so env-var
flips (hermes tools enable X) propagate within a turn or two without
explicit invalidation.
Measured: first call 7.5ms → 1.6ms (check_fn probes now dominate);
subsequent calls ~0.01ms via the upstream memoization.
Invalidation surface:
- registry._generation bumps on register/deregister/register_toolset_alias,
invalidating the memoized definitions automatically.
- config.yaml mtime in the cache key captures user-visible config edits
affecting dynamic schemas (execute_code mode, discord allowlist).
- invalidate_check_fn_cache() exposed for explicit flushes (e.g. after
hermes tools enable/disable).
- tests/conftest.py autouse fixture clears both caches before every test
so env-var monkeypatches don't see stale results.
Also fixes a regression from PR #17046 that I missed:
- tools/web_tools.py — Firecrawl was removed from module scope by the
lazy import, breaking 8 tests that patch 'tools.web_tools.Firecrawl'.
Applied the same _FirecrawlProxy pattern used in auxiliary_client/
run_agent for OpenAI (module-level proxy that looks like the class
but imports the SDK on first call/isinstance; patch() replaces the
attribute as usual).
Verified:
- 49/49 tests/tools/test_web_tools_config.py pass (was 8 failing on main)
- 68/68 tests/tools/test_homeassistant_tool.py pass (was 1 failing in
the full suite due to check_fn TTL cross-test pollution; fixed by
the autouse fixture)
- 3887/3895 tests/tools/ (8 pre-existing fails: 2 delegate, 1 mcp
dynamic discovery, 5 mcp structured content — all confirmed on main)
- 2973/2976 tests/agent/ + tests/run_agent/ (3 pre-existing fails)
- 868/868 tests/run_agent/ (excluding test_run_agent.py which has
pre-existing suite-level issues)
- Live smoke: 2 turns + /model switch + tool calls, zero errors in
agent.log session window.
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
* perf(startup): lazy-import OpenAI, Anthropic, Firecrawl, account_usage
Four heavy SDK/module imports are now deferred off the hot startup path.
Net savings on cold module imports:
cli 1200 → 958 ms (-242)
run_agent 1220 → 901 ms (-319)
tools.web_tools 711 → 423 ms (-288)
agent.anthropic_adapter 230 → 15 ms (-215)
agent.auxiliary_client 253 → 68 ms (-185)
Four independent changes in one PR since they all use the same pattern
and share the same risk profile (heavy SDK import → lazy proxy or
function-local import):
1. tools/web_tools.py:
'from firecrawl import Firecrawl' moved into _get_firecrawl_client(),
which is only called when backend='firecrawl'. Users on Exa/Tavily/
Parallel pay zero firecrawl cost.
2. cli.py + gateway/run.py:
'from agent.account_usage import ...' moved into the /limits handlers.
account_usage transitively pulls the OpenAI SDK chain; only needed
when the user runs /limits.
3. agent/anthropic_adapter.py:
'try: import anthropic as _anthropic_sdk' replaced with a cached
'_get_anthropic_sdk()' accessor. The three usage sites
(build_anthropic_client, build_anthropic_bedrock_client,
read_claude_code_credentials_from_keychain) now resolve via the
accessor. All pre-existing test patches of
'agent.anthropic_adapter._anthropic_sdk' keep working because the
accessor respects any value already in module globals.
4. agent/auxiliary_client.py AND run_agent.py:
'from openai import OpenAI' replaced with an '_OpenAIProxy()' module-
level object that looks like the OpenAI class but imports the SDK on
first call/isinstance check. This preserves:
- 15+ in-module OpenAI(...) construction sites in auxiliary_client
and the single site in run_agent's _create_openai_client (Python's
function-scope name lookup finds the proxy, forwards the call);
- 'patch("agent.auxiliary_client.OpenAI", ...)' and
'patch("run_agent.OpenAI", ...)' test patterns used by 28+ test
files (patch replaces the module attribute as usual).
Tried two alternatives first:
- 'from openai._client import OpenAI' — doesn't skip openai/__init__.py
(the audit's hypothesis here was wrong).
- Module-level __getattr__ — works for external access but Python
function-scope name resolution skips __getattr__, so in-module
OpenAI(...) calls NameError.
Note: 'openai' still loads on 'import cli' because
cli.py -> neuter_async_httpx_del() -> openai._base_client, and
run_agent.py -> code_execution_tool.py (module-level
build_execute_code_schema) -> _load_config() -> 'from cli import
CLI_CONFIG'. Deferring those is a separate, larger change — out of scope
for this PR. The savings above all come from avoiding the openai/*,
anthropic/*, and firecrawl/* top-level type-tree imports on paths that
don't need them.
Verified:
- 302/302 tests in tests/agent/{test_anthropic_adapter,
test_bedrock_1m_context, test_minimax_provider, test_anthropic_keychain}
pass. Two pre-existing failures on main unchanged.
- 106/106 tests/agent/test_auxiliary_client.py pass (1 pre-existing fail).
- 97/97 tests/run_agent/test_create_openai_client_kwargs_isolation.py,
test_plugin_context_engine_init.py, test_invalid_context_length_warning.py,
test_api_max_retries_config.py,
tests/hermes_cli/test_gemini_provider.py, test_ollama_cloud_provider.py
pass (1 pre-existing fail).
- Live hermes chat smoke: 2 turns + /model switch + tool calls, zero
errors in the 57-line agent.log window.
- Module-level import of run_agent + auxiliary_client + anthropic_adapter
no longer pulls 'anthropic' or 'firecrawl' at all.
* fix(gateway): restore top-level account_usage import for test-patch surface
CI caught two failures in tests/gateway/test_usage_command.py that I
missed locally:
AttributeError: 'module' object at gateway.run has no attribute 'fetch_account_usage'
The test uses monkeypatch.setattr('gateway.run.fetch_account_usage', ...)
to inject a fake account-fetch call. Moving the import inside the
handler deleted that module-level attribute, breaking the patch surface.
Restoring the top-level import in gateway/run.py gives up the ~230 ms
gateway-boot savings from that one lazy, but:
1. the gateway is a long-running daemon — boot cost is paid once per
install, not per turn;
2. the other four lazy-imports (firecrawl, openai, anthropic, cli's
account_usage) remain in place and still account for the bulk of
the savings reported in the PR body;
3. preserving the patch surface keeps the established
'gateway.run.fetch_account_usage' monkeypatch pattern working
without touching tests.
Verified: tests/gateway/test_usage_command.py — 8 passed, 0 failed.
Full targeted sweep (2336 tests across agent/gateway/hermes_cli/run_agent):
2332 passed, 4 failed — all 4 pre-existing on main.
---------
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
Previously, check_browser_requirements() only checked for the agent-browser
CLI, not the Chromium binary it drives. When the CLI was present but
Chromium wasn't (common in Docker images predating the playwright install
step), the browser tool was advertised to the agent, every call hung for
the full command timeout (~30s each, ~220s for a chained navigate), and
the agent eventually gave up with no useful error — users saw 'browser
not working' with empty errors.log.
Changes:
- tools/browser_tool.py: add _chromium_installed() checking
PLAYWRIGHT_BROWSERS_PATH + default Playwright cache paths for
chromium-* / chromium_headless_shell-* dirs; wire into
check_browser_requirements() for local mode (cloud providers
unaffected). _run_browser_command fails fast with an actionable
Docker vs. host message instead of hanging. _running_in_docker()
checks /.dockerenv and /proc/1/cgroup.
- hermes_cli/tools_config.py: post_setup for 'Local Browser' now runs
'agent-browser install --with-deps' after npm install to actually
download Chromium. In Docker, points user at the updated image pull
instead of trying to install into a read-only layer. Cloud-provider
post_setup (browserbase) skips Chromium install entirely.
- tests/tools/test_browser_chromium_check.py: new tests covering
search roots, install detection, requirements branches (local/cloud/
camofox), and the fast-fail guard in docker/non-docker contexts.
- tests/tools/test_browser_homebrew_paths.py: 5 existing subprocess-path
tests now mock _chromium_installed=True since they exercise the
post-guard subprocess path.
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
Mechanical cleanup across 43 files — removes 46 unused imports
(F401) and 14 unused local variables (F841) detected by
`ruff check --select F401,F841`. Net: -49 lines.
Also fixes a latent NameError in rl_cli.py where `get_hermes_home()`
was called at module line 32 before its import at line 65 — the
module never imported successfully on main. The ruff audit surfaced
this because it correctly saw the symbol as imported-but-unused
(the call happened before the import ran); the fix moves the import
to the top of the file alongside other stdlib imports.
One `# noqa: F401` kept in hermes_cli/status.py for `subprocess`:
tests monkeypatch `hermes_cli.status.subprocess` as a regression
guard that systemctl isn't called on Termux, so the name must
exist at module scope even though the module body doesn't reference
it. Docstring explains the reason.
Also fixes an invalid `# noqa:` directive in
gateway/platforms/discord.py:308 that lacked a rule code.
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
delegate_task runs inside the parent turn and is cancelled when the parent is interrupted (new user message, /stop, /new). The child status payload (status=interrupted, exit_reason=interrupted) is already honest, but the tool schema and user-facing docs did not set the expectation, so users reasonably assumed delegated subagents would keep running in the background after interrupting the parent.
Updates:
- tools/delegate_tool.py DELEGATE_TASK_SCHEMA description adds a WHEN NOT TO USE bullet pointing at cronjob / terminal(background=True, notify_on_complete=True) for durable long-running work.
- website/docs/user-guide/features/delegation.md gains a Lifetime and Durability callout above Key Properties.
- website/docs/guides/delegation-patterns.md expands the Use something else list and the Constraints section with the same guidance.
Reported by LizLiz (@lizliz404) via Teknium.
Co-authored-by: teknium1 <teknium@users.noreply.github.com>
Extract the islink/realpath guard from the 16743 fix into a single
atomic_replace() helper in utils.py, then migrate every os.replace()
call site in the codebase to use it.
The original PR #16777 correctly identified and fixed the bug, but
only patched 9 of ~24 call sites. The same bug class (managed
deployments that symlink state files silently losing the link on
every write) still existed at auth.json, sessions file, gateway
config, env_loader, webhook subscriptions, debug store, model
catalog, pairing, google OAuth, nous rate guard, and more.
Rather than add another 10+ copies of the same three-line guard,
consolidate into atomic_replace(tmp, target) which:
- resolves symlinks via os.path.realpath before os.replace
- returns the resolved real path so callers can re-apply permissions
- is a drop-in replacement for os.replace at the use sites
Changes:
- utils.py: new atomic_replace() helper + atomic_json_write /
atomic_yaml_write now call it instead of inlining the guard
- 16 files: all os.replace() call sites migrated to atomic_replace()
- agent/{google_oauth, nous_rate_guard, shell_hooks}.py
- cron/jobs.py
- gateway/{pairing, session, platforms/telegram}.py
- hermes_cli/{auth, config, debug, env_loader, model_catalog, webhook}.py
- tools/{memory_tool, skill_manager_tool, skills_sync}.py
Tests: tests/test_atomic_replace_symlinks.py pins the invariant for
atomic_replace + atomic_json_write + atomic_yaml_write, covers plain
files, first-time creates, broken symlinks, and permission preservation.
Refs #16743
Builds on #16777 by @vominh1919.
os.replace(tmp, path) replaces the symlink itself with a regular file,
breaking users who symlink config.yaml, SOUL.md, or .env from ~/.hermes/
to a dotfiles repo or managed profile package.
Fix: resolve symlinks via os.path.realpath() before os.replace(), so the
real file is overwritten in-place while the symlink survives.
Fixed in 7 files covering all os.replace call sites:
- utils.py (atomic_json_write, atomic_yaml_write — fixes save_config)
- hermes_cli/config.py (env sanitizer, save_env_value, remove_env_value)
- tools/skill_manager_tool.py (_atomic_write_text — SOUL.md writes)
- tools/memory_tool.py (memory file writes)
- tools/skills_sync.py (manifest writes)
- cron/jobs.py (job state + output file writes)
- agent/shell_hooks.py (hook file writes)
FixesNousResearch/hermes-agent#16743
Adds tools.schema_sanitizer.strip_nullable_unions as the single
implementation for collapsing anyOf/oneOf nullable unions. Both the
MCP input-schema normalizer and the Anthropic tool-schema guard now
delegate to it instead of re-implementing the same walk three times.
The global sanitizer also gains a final pass so any tool that slips
past the two earlier hooks (plugin tools, non-MCP custom tools with
Pydantic-shaped schemas) still gets safe input_schemas on Anthropic.
- tools/schema_sanitizer.py:
* New public strip_nullable_unions(schema, keep_nullable_hint=True).
* _sanitize_single_tool() calls it as a final pass (hint preserved
so coerce_tool_args can still map string "null" to None).
- tools/mcp_tool.py: _normalize_mcp_input_schema delegates.
- agent/anthropic_adapter.py: _normalize_tool_input_schema delegates
with keep_nullable_hint=False (Anthropic does not recognize nullable).
No behavioral change for the fix itself; tests (73/73 targeted +
E2E across MCP→sanitizer→Anthropic paths) pass.
Extends the cua-driver computer-use backend to drive backgrounded macOS
windows without stealing keyboard or mouse focus from the foreground app.
All changes target the cua-driver MCP backend and the shared dispatcher.
## cua_backend.py
**Window-aware capture**: capture() now calls list_windows + get_window_state
instead of the removed capture tool. Prefers structuredContent.windows
(MCP 2024-11-05+ cua-driver) for zero-parse window enumeration; falls back
to regex-parsed text for older builds. Stores the selected (pid, window_id)
as sticky context so subsequent action calls do not need a redundant round-trip.
**Action routing**: click/scroll/type_text/key all carry the sticky pid
(and window_id for element-indexed clicks). type_text routes through
type_text_chars (individual key events) rather than AX attribute write --
WebKit AXTextFields reject attribute writes from backgrounded processes.
**Key parsing**: _parse_key_combo splits cmd+s-style strings into
(key, [modifiers]) and routes to hotkey (modifier present) or
press_key (bare key) -- cua-driver actual tool names.
**set_value method**: new set_value(value, element) calls the cua-driver
set_value MCP tool. For AXPopUpButton / HTML select in a backgrounded Safari,
AXPress opens the native macOS popup which closes immediately when the app is
non-frontmost; set_value AX-presses the matching child option directly
(no menu required, no focus steal).
**focus_app**: reimplemented as a pure window-selector (enumerates
list_windows, sets sticky pid/window_id) without ever raising the window
or stealing focus.
**list_apps**: fixed tool name from listApps to list_apps; handles plain-text
response via regex when structured data is absent.
**Structured-content extraction**: _extract_tool_result now surfaces
structuredContent from MCP results, enabling the list_windows window array
without text parsing.
**Helpers**: _parse_windows_from_text, _parse_elements_from_tree,
_split_tree_text, _parse_key_combo extracted as module-level functions.
## schema.py
Added set_value to the action enum with a description explaining when to
prefer it over click (select/popup elements, sliders, no focus steal).
Added value field for set_value payloads.
## tool.py
Routed set_value action through _dispatch to backend.set_value.
Added set_value to _DESTRUCTIVE_ACTIONS (approval-gated).
Fixed MIME-type detection in _capture_response: cua-driver may return
JPEG; detect from base64 magic bytes (/9j/ -> image/jpeg, else image/png)
rather than hardcoding image/png.
## agent/display.py + run_agent.py
Guard _detect_tool_failure and result-preview logic against non-string
function_result values: multimodal tool results (dicts with _multimodal=True)
are not string-sliceable; treat them as successes and fall back to str()
for length/preview.
Background macOS desktop control via cua-driver MCP — does NOT steal the
user's cursor or keyboard focus, works with any tool-capable model.
Replaces the Anthropic-native `computer_20251124` approach from the
abandoned #4562 with a generic OpenAI function-calling schema plus SOM
(set-of-mark) captures so Claude, GPT, Gemini, and open models can all
drive the desktop via numbered element indices.
- `tools/computer_use/` package — swappable ComputerUseBackend ABC +
CuaDriverBackend (stdio MCP client to trycua/cua's cua-driver binary).
- Universal `computer_use` tool with one schema for all providers.
Actions: capture (som/vision/ax), click, double_click, right_click,
middle_click, drag, scroll, type, key, wait, list_apps, focus_app.
- Multimodal tool-result envelope (`_multimodal=True`, OpenAI-style
`content: [text, image_url]` parts) that flows through
handle_function_call into the tool message. Anthropic adapter converts
into native `tool_result` image blocks; OpenAI-compatible providers
get the parts list directly.
- Image eviction in convert_messages_to_anthropic: only the 3 most
recent screenshots carry real image data; older ones become text
placeholders to cap per-turn token cost.
- Context compressor image pruning: old multimodal tool results have
their image parts stripped instead of being skipped.
- Image-aware token estimation: each image counts as a flat 1500 tokens
instead of its base64 char length (~1MB would have registered as
~250K tokens before).
- COMPUTER_USE_GUIDANCE system-prompt block — injected when the toolset
is active.
- Session DB persistence strips base64 from multimodal tool messages.
- Trajectory saver normalises multimodal messages to text-only.
- `hermes tools` post-setup installs cua-driver via the upstream script
and prints permission-grant instructions.
- CLI approval callback wired so destructive computer_use actions go
through the same prompt_toolkit approval dialog as terminal commands.
- Hard safety guards at the tool level: blocked type patterns
(curl|bash, sudo rm -rf, fork bomb), blocked key combos (empty trash,
force delete, lock screen, log out).
- Skill `apple/macos-computer-use/SKILL.md` — universal (model-agnostic)
workflow guide.
- Docs: `user-guide/features/computer-use.md` plus reference catalog
entries.
44 new tests in tests/tools/test_computer_use.py covering schema
shape (universal, not Anthropic-native), dispatch routing, safety
guards, multimodal envelope, Anthropic adapter conversion, screenshot
eviction, context compressor pruning, image-aware token estimation,
run_agent helpers, and universality guarantees.
469/469 pass across tests/tools/test_computer_use.py + the affected
agent/ test suites.
- `model_tools.py` provider-gating: the tool is available to every
provider. Providers without multi-part tool message support will see
text-only tool results (graceful degradation via `text_summary`).
- Anthropic server-side `clear_tool_uses_20250919` — deferred;
client-side eviction + compressor pruning cover the same cost ceiling
without a beta header.
- macOS only. cua-driver uses private SkyLight SPIs
(SLEventPostToPid, SLPSPostEventRecordTo,
_AXObserverAddNotificationAndCheckRemote) that can break on any macOS
update. Pin with HERMES_CUA_DRIVER_VERSION.
- Requires Accessibility + Screen Recording permissions — the post-setup
prints the Settings path.
Supersedes PR #4562 (pyautogui/Quartz foreground backend, Anthropic-
native schema). Credit @0xbyt4 for the original #3816 groundwork whose
context/eviction/token design is preserved here in generic form.
When delegation.provider is configured (e.g. minimax-cn), subagents
inherited the parent's acp_command unconditionally. This caused
run_agent.py to initialize CopilotACPClient, which bypassed the
override credentials entirely and used its own default model
(provider=copilot-acp model=qwen3.5-397b-a17b) instead of the
configured delegation.provider and delegation.model.
Fix: when override_provider is set but override_acp_command is not,
clear effective_acp_command and effective_acp_args so the child agent
uses direct API calls with the configured provider credentials.
The existing override_acp_command path is unchanged — explicit ACP
transport overrides still force provider=copilot-acp as before.
Fixes#16816
* Port from Kilo-Org/kilocode#9448: roll up subagent costs into parent session total
Child subagents built by delegate_task() each track their own
session_estimated_cost_usd, but the parent agent's total never folded
those numbers in. On runs where the parent mostly delegates and the
children do the expensive work, the footer/UI was reporting a fraction
of the actual spend — sometimes $0.00 when the parent itself made no
billed calls.
Fix:
- Capture each child's session_estimated_cost_usd into _child_cost_usd
on the result entry (before child.close() drops the counter).
- After the existing subagent_stop hook loop, sum the children's costs
and add the total to parent.session_estimated_cost_usd.
- Promote session_cost_source from 'none' -> 'subagent' when the parent
had no direct spend but children did, so the UI doesn't label the
total as having unknown provenance. Real sources (openrouter,
anthropic, etc.) are preserved.
Nested orchestrator -> worker trees roll up naturally: each layer's own
delegate_task() folds its direct children in, and when the orchestrator
itself returns, its parent folds the orchestrator's now-inflated total
on top.
Internal fields (_child_cost_usd, _child_role) are stripped from the
results dict before it's serialised back to the model — same contract
as _child_role already followed.
Tests: TestSubagentCostRollup (5 cases) covers single-child, batch,
zero-cost-children, preserved-source, and legacy-fixture paths.
Source: https://github.com/Kilo-Org/kilocode/pull/9448
* fix(web): scope dashboard config Reset button to the current tab
Reported by @ykmfb001 via X: clicking 'Restore Defaults' (恢复默认值) on
the Auxiliary page wiped the entire config.yaml to defaults, not just
the auxiliary section. The button sits next to the category tabs and
users reasonably assumed 'reset this tab', not 'reset everything'.
Changes:
- handleReset now scopes to the fields in the current view:
active category's fields (form mode) or search-matched fields
(search mode). Only those keys are copied from defaults; the rest
of the config is left alone.
- Added a window.confirm() with the scope name before applying.
- Button is hidden in YAML mode (scoping doesn't apply there).
- Tooltip/aria-label now name the scope, e.g. 'Reset Auxiliary to
defaults'.
- i18n: new resetScopeTooltip / confirmResetScope / resetScopeToast
strings in en + zh; resetDefaults key preserved for compat.
Plugins can now observe dangerous-command approval events in real time,
on both the CLI-interactive path and the async gateway path. This is the
missing hook surface external tools need to build approval notifiers
(macOS menu-bar allow/deny, Slack alerts, audit logs, etc.) without
forking Hermes or running a parallel gateway adapter.
Changes:
- hermes_cli/plugins.py: add two entries to VALID_HOOKS
- tools/approval.py: fire both hooks from check_all_command_guards --
around prompt_dangerous_approval (CLI surface) and around the
notify_cb + blocking event.wait loop (gateway surface)
- website/docs/user-guide/features/hooks.md: document both hooks with
a macOS-notification example
- tests/tools/test_approval_plugin_hooks.py: 5 tests covering CLI once,
CLI deny, plugin-crash resilience, gateway approve, gateway timeout
Hooks are observer-only: return values are ignored, so plugins cannot
veto or pre-answer an approval (use pre_tool_call for that). A crashing
plugin cannot break the approval flow -- invoke_hook swallows per-
callback errors, and the wrapper logs and swallows dispatch-layer
errors too.
Surface kwarg distinguishes "cli" from "gateway"; post hook reports
choice as one of once/session/always/deny/timeout.
PR #13734 fixed the concurrent-tool-executor vector (ThreadPoolExecutor
workers didn't inherit the CLI's TLS approval callback). Two vectors
remained that could still land in the deadlocking input() fallback:
1. _spawn_background_review spawns a raw threading.Thread with no
approval callback installed, so any dangerous-command guard the
review agent trips falls back to input() -> deadlock against the
parent's prompt_toolkit TUI (same class as delegate_task subagents,
fixed in 023b1bff1 / #15491). Install a _bg_review_auto_deny
callback at thread start, clear on finally.
2. prompt_dangerous_approval's fallback unconditionally spawned a
daemon thread calling input() when approval_callback was None.
That fallback can never succeed under prompt_toolkit because the
user's Enter goes to pt's raw-mode stdin capture. Detect an active
pt Application via get_app_or_none() and fail closed (deny + log)
instead, so future threads that forget to install a callback
degrade gracefully instead of hanging 60s invisibly.
Regression guards:
- tests/run_agent/test_background_review.py verifies the review
worker thread sees a callable auto-deny callback mid-run and that
the slot is cleared in the finally block.
- tests/tools/test_approval.py TestFailClosedUnderPromptToolkit
verifies prompt_dangerous_approval returns 'deny' fast under a
mocked pt Application, and that a real callback still wins over
the guard.
* feat(image-input): native multimodal routing based on model vision capability
Attach user-sent images as OpenAI-style content parts on the user turn when
the active model supports native vision, so vision-capable models see real
pixels instead of a lossy text description from vision_analyze.
Routing decision (agent/image_routing.py::decide_image_input_mode):
agent.image_input_mode = auto | native | text (default: auto)
In auto mode:
- If auxiliary.vision.provider/model is explicitly configured, keep the
text pipeline (user paid for a dedicated vision backend).
- Else if models.dev reports supports_vision=True for the active
provider/model, attach natively.
- Else fall back to text (current behaviour).
Call sites updated: gateway/run.py (all messaging platforms), tui_gateway
(dashboard/Ink), cli.py (interactive /attach + drag-drop).
run_agent.py changes:
- _prepare_anthropic_messages_for_api now passes image parts through
unchanged when the model supports vision — the Anthropic adapter
translates them to native image blocks. Previous behaviour
(vision_analyze → text) only runs for non-vision Anthropic models.
- New _prepare_messages_for_non_vision_model mirrors the same contract
for chat.completions and codex_responses paths, so non-vision models
on any provider get text-fallback instead of failing at the provider.
- New _model_supports_vision() helper reads models.dev caps.
vision_analyze description rewritten: positions it as a tool for images
NOT already visible in the conversation (URLs, tool output, deeper
inspection). Prevents the model from redundantly calling it on images
already attached natively.
Config default: agent.image_input_mode = auto.
Tests: 35 new (test_image_routing.py + test_vision_aware_preprocessing.py),
all existing tests that reference _prepare_anthropic_messages_for_api
still pass (198 targeted + new tests green).
* feat(image-input): size-cap + resize oversized images, charge image tokens in compressor
Two follow-ups that make the native image routing safer for long / heavy
sessions:
1) Oversize handling in build_native_content_parts:
- 20 MB ceiling per image (matches vision_tools._MAX_BASE64_BYTES,
the most restrictive provider — Gemini inline data).
- Delegates to vision_tools._resize_image_for_vision (Pillow-based,
already battle-tested) to downscale to 5 MB first-try.
- If Pillow is missing or resize still overshoots, the image is
dropped and reported back in skipped[]; caller falls back to text
enrichment for that image.
2) Image-token accounting in context_compressor:
- New _IMAGE_TOKEN_ESTIMATE = 1600 (matches Claude Code's constant;
within the realistic range for Anthropic/GPT-4o/Gemini billing).
- _content_length_for_budget() helper: sums text-part lengths and
charges _IMAGE_CHAR_EQUIVALENT (1600 * 4 chars) per image/image_url/
input_image part. Base64 payload inside image_url is NOT counted
as chars — dimensions don't matter, only image-presence.
- Both tail-cut sites (_prune_old_tool_results L527 and
_find_tail_cut_by_tokens L1126) now call the helper so multi-image
conversations don't slip past compression budget.
Tests: 9 new in test_image_routing.py (oversize triggers resize,
resize-fails-returns-None, oversize-skipped-reported), 11 new in
test_compressor_image_tokens.py (flat charge per image, multiple images,
Responses-API / Anthropic-native / OpenAI-chat shapes, no-inflation on
raw base64, bounds-check on the constant, integration test that an
image-heavy tail actually gets trimmed).
* fix(image-input): replace blanket 20MB ceiling with empirically-verified per-provider limits
The previous commit imposed a hardcoded 20 MB base64 ceiling on all
providers, triggering auto-resize on anything larger. This was wrong in
both directions:
* Too loose for Anthropic — actual limit is 5 MB (returns HTTP 400
'image exceeds 5 MB maximum' above that).
* Too strict for OpenAI / Codex / OpenRouter — accept 49 MB+ without
complaint (empirically verified April 2026 with progressive PNG
sizes).
New behaviour:
* _PROVIDER_BASE64_CEILING table: only anthropic and bedrock have a
ceiling (5 MB, since bedrock-on-Claude shares Anthropic's decoder).
* Providers NOT in the table get no ceiling — images attach at native
size and we trust the provider to return its own error if it
disagrees. A provider-specific 400 message is clearer than us
guessing wrong and silently degrading image quality.
* build_native_content_parts() gains a keyword-only provider arg;
gateway/CLI/TUI pass the active provider so Anthropic users get
auto-resize protection while OpenAI users don't pay it.
* Resize target dropped from 5 MB to 4 MB to slide safely under
Anthropic's boundary with header overhead.
Empirical measurements (direct API, no Hermes in the loop):
image b64 anthropic openrouter/gpt5.5 codex-oauth/gpt5.5
0.19 MB ✓ ✓ ✓
12.37 MB ✗ 400 5MB ✓ ✓
23.85 MB ✗ 400 5MB ✓ ✓
49.46 MB ✗ 413 ✓ ✓
Tests: rewrote TestOversizeHandling (5 tests): no-ceiling pass-through,
Anthropic resize fires, Anthropic skip on resize-fail, build_native_parts
routes ceiling by provider, unknown provider gets no ceiling. All 52
targeted tests pass.
* refactor(image-input): attempt native, shrink-and-retry on provider reject
Replace proactive per-provider size ceilings with a reactive shrink path
on the provider's actual rejection. All providers now attempt native
full-size attachment first; if the provider returns an image-too-large
error, the agent silently shrinks and retries once.
Why the previous design was wrong: hardcoding provider ceilings
(anthropic=5MB, others=unlimited) meant OpenAI users on a 10MB image
paid no tax, but Anthropic users lost quality on anything >5MB even
though the empirical behaviour at provider-reject time is the same
(shrink + retry). Baking the table into the routing layer also
requires updating Hermes every time a provider's limit changes.
Reactive design:
- image_routing.py: _file_to_data_url encodes native size, no ceiling.
build_native_content_parts drops its provider kwarg.
- error_classifier.py: new FailoverReason.image_too_large + pattern
match ("image exceeds", "image too large", etc.) checked BEFORE
context_overflow so Anthropic's 5MB rejection lands in the right
bucket.
- run_agent.py: new _try_shrink_image_parts_in_messages walks api
messages in-place, re-encodes oversized data: URL image parts
through vision_tools._resize_image_for_vision to fit under 4MB,
handles both chat.completions (dict image_url) and Responses
(string image_url) shapes, ignores http URLs (provider-fetched).
New image_shrink_retry_attempted flag in the retry loop fires the
shrink exactly once per turn after credential-pool recovery but
before auth retries.
E2E verified live against Anthropic claude-sonnet-4-6:
- 17.9MB PNG (23.9MB b64) attached at native size
- Anthropic returns 400 "image exceeds 5 MB maximum"
- Agent logs '📐 Image(s) exceeded provider size limit — shrank and
retrying...'
- Retry succeeds, correct response delivered in 6.8s total.
Tests: 12 new (8 shrink-helper shapes + 4 classifier signals),
replaces 5 proactive-ceiling tests with 3 simpler 'native attach works'
tests. 181 targeted tests pass. test_enum_members_exist in
test_error_classifier.py updated for the new enum value.
On macOS (bash 3.2 and some Homebrew bash builds) `source`ing a file that
contains `declare -x` statements prints each declaration to stdout. The
persistent-shell wrapper in tools/environments/base.py was only redirecting
stderr when sourcing the session snapshot, so ~60 lines of env vars leaked
into every terminal tool response — blowing out context and triggering
HTTP 400s on context-limited providers.
Fix: redirect both stdout and stderr when sourcing the snapshot. Linux
bash is silent here, so the redirect is harmless there; macOS no longer
leaks.
Closes#15459
Co-authored-by: Sanjays2402 <51058514+Sanjays2402@users.noreply.github.com>
read_file's dedup path returned a lightweight stub on re-reads of an
unchanged file, then returned early — so the consecutive-read loop
guard (hard block at count>=4) at the bottom of read_file_tool never
ran for stub-looped calls. Weaker tool-following models (local Qwen3.6
variants in the reported case) ignore the passive 'refer to earlier
result' hint and hammer the same read_file call until iteration budget
runs out.
Track per-key stub returns in task_data['dedup_hits'] and, on the
second stub for the same (path, offset, limit), return a hard BLOCKED
error mirroring the wording the real-read path already uses. A real
read, an intervening non-read tool call (notify_other_tool_call), or
reset_file_dedup (on context compression) all clear the counter so
the guard never stays engaged longer than the actual loop.
Closes#15759
Four small tool-description / skill-content tweaks addressing recurring
model mistakes seen in @versun's docx feedback (Kimi 2.6, but the patterns
apply to every model):
1. browser_navigate description: call out .md/.txt/.json/.yaml/.csv/.xml,
raw.githubusercontent.com, and API endpoints as specifically preferring
curl or web_extract. The generic "prefer web_search or web_extract" was
too weak; models kept firing up the browser for plain-text URLs.
2. delegate_task description: two additions.
(a) Pass user language / output-style preferences in 'context' when they
differ from English — otherwise subagents default to English and their
summaries contaminate the final reply (caused the bilingual digest bug).
(b) Subagent summaries are self-reports, not verified facts. For
operations with external side-effects (HTTP uploads, remote writes,
file creation at shared paths), require a verifiable handle (URL, ID,
path) and verify it yourself before claiming success.
3. agent/prompt_builder.py Skills-mandatory block: new explicit line
"Whenever the user asks to configure / set up / modify / install /
enable / disable / troubleshoot Hermes Agent itself, load the
`hermes-agent` skill first." The generic "load what's relevant" didn't
route Hermes-meta questions (like "how do I turn off redaction?") to
the one skill that has the answer.
4. skills/autonomous-ai-agents/hermes-agent/SKILL.md: new "Security &
Privacy Toggles" section covering security.redact_secrets (with the
import-time-snapshot restart-required caveat), privacy.redact_pii,
approvals.mode (manual/smart/off) + --yolo + HERMES_YOLO_MODE, shell
hooks allowlist, and how to disable network/media tools entirely.
Every command verified against the actual config keys — no invented
knobs.
Co-authored-by: teknium1 <teknium@noreply.github.com>
* feat(skills): install skills from a direct HTTP(S) URL
Adds UrlSource adapter so `hermes skills install <url-to-SKILL.md>` and
`/skills install <url>` work as first-class operations — no more
improvising with curl + patch + cp.
- Claims identifiers that start with http(s):// and end in .md
- Skips /.well-known/skills/ URLs (WellKnownSkillSource handles those)
- Skill name from YAML frontmatter, URL-slug fallback
- Single-file SKILL.md only (v1 scope — multi-file skills need a manifest)
- Trust level 'community'; full security scan still runs
- Lock file stores the URL as identifier so `hermes skills update`
re-fetches from the same URL cleanly
Scope matches real user need from @versun's docx feedback where
`https://sharethis.chat/SKILL.md` had no first-class install path.
* feat(skills): interactive name/category for URL installs + --name override
Follow-up to the UrlSource adapter. The previous commit fell back to weak
heuristics when frontmatter had no ``name:`` and could produce garbage names
like ``SKILL`` or ``unnamed-skill``. Now:
tools/skills_hub.py
- ``UrlSource._is_valid_skill_name()`` — strict identifier check
(``^[a-z][a-z0-9_-]*$``), rejects sentinel values (``SKILL``, ``README``,
``INDEX``, ``unnamed-skill``, empty, non-strings).
- ``_resolve_skill_name()`` returns ``Optional[str]`` — ``None`` when
nothing valid is resolvable. Also ignores unsafe frontmatter names
(``../evil``) and falls through to URL slug instead of returning None
immediately, so a URL with a bad frontmatter but a good path still
works.
- ``fetch()``/``inspect()`` carry an ``awaiting_name=True`` marker in
metadata/extra when resolution fails, letting ``do_install`` decide
whether to prompt, apply an override, or error out.
hermes_cli/skills_hub.py
- ``do_install`` gains a ``name_override`` parameter.
- On URL-sourced bundles with ``awaiting_name=True``:
1. If ``name_override`` is valid → use it.
2. If ``name_override`` is invalid → refuse with a clear error.
3. Else if ``skip_confirm=True`` (non-interactive: slash / TUI /
gateway / scripts) → refuse with an actionable retry hint pointing
at ``--name <your-name>`` on both CLI and slash forms.
4. Else (interactive TTY) → prompt for the name.
- Interactive TTY also prompts for a category when none is given for a
URL-sourced install, hinting existing category buckets so users can
reuse ``productivity``, ``devops``, etc. Empty input → flat install.
- ``_existing_categories()`` scans ``~/.hermes/skills/`` for subdirs that
look like category buckets (contain nested SKILL.md files); skips
top-level skills and hidden dirs.
- ``_prompt_for_skill_name()`` / ``_prompt_for_category()`` helpers
(EOF/Ctrl-C-safe, match the existing ``Confirm [y/N]`` prompt style).
hermes_cli/main.py
- ``hermes skills install`` argparse gains ``--name <name>``.
hermes_cli/skills_hub.py (slash)
- ``/skills install <url> --name <x>`` parsing added.
Tests
- tests/tools/test_skills_hub.py: updated ``UrlSource`` tests to assert
the new ``awaiting_name`` metadata; added 4 new tests for
``_is_valid_skill_name`` rejection sets and the awaiting-name marker.
- tests/hermes_cli/test_skills_hub.py: 8 new tests covering --name
override accept/reject, non-interactive error, interactive name prompt,
interactive category prompt, cancel-aborts-install, and
``_existing_categories`` scan behavior (buckets vs flat skills).
- E2E verified all four paths (no-name/no-override → error;
--name override → install; frontmatter name → install;
invalid --name → rejection).
---------
Co-authored-by: teknium1 <teknium@noreply.github.com>
_search_members() and _fetch_messages() call min(limit, 100) assuming
limit is int. Models can pass limit as a string (e.g. "10"), causing
TypeError: '<' not supported between instances of 'str' and 'int'.
Add try/except int() coercion with safe defaults at the top of both
functions, matching the pattern used in session_search fix (#10522).
Every working dir hermes ever touches gets its own shadow git repo under
~/.hermes/checkpoints/{sha256(abs_dir)[:16]}/. The per-repo _prune is a
no-op (comment in CheckpointManager._prune says so), so abandoned repos
from deleted/moved projects or one-off tmp dirs pile up forever. Field
reports put the typical offender at 1000+ repos / ~12 GB on active
contributor machines.
Adds an opt-in startup sweep that mirrors the sessions.auto_prune
pattern from #13861 / #16286:
- tools/checkpoint_manager.py: new prune_checkpoints() and
maybe_auto_prune_checkpoints() helpers. Deletes shadow repos that
are orphan (HERMES_WORKDIR marker points to a path that no longer
exists) or stale (newest in-repo mtime older than retention_days).
Idempotent via a CHECKPOINT_BASE/.last_prune marker file so it only
runs once per min_interval_hours regardless of how many hermes
processes start up.
- hermes_cli/config.py: new checkpoints.auto_prune /
retention_days / delete_orphans / min_interval_hours knobs.
Default auto_prune: false so users who rely on /rollback against
long-ago sessions never lose data silently.
- cli.py / gateway/run.py: startup hooks gated on checkpoints.auto_prune,
called right next to the existing state.db maintenance block.
- Docs updated with the new config knobs.
- 11 regression tests: orphan/stale deletion, precedence, byte-freed
tracking, non-shadow dir skip, interval gating, corrupt marker
recovery.
Refs #3015 (session-file disk growth was fixed in #16286; this covers
the checkpoint side noted out-of-scope there).
The write_file guard added in #16223 used strict equality against the
internal dedup status message. In practice, the model sometimes
prepends a short note or appends a trailing comment before calling
write_file, which slipped past the strict check.
Broaden the heuristic: reject writes whose stripped content equals
the status message OR contains it and is <=2x its length. Short,
status-dominated writes are always corruption; legitimate docs that
quote the message verbatim are always much longer.
Adds two tests: one for the small-wrapper corruption shape, one
confirming large legitimate files that quote the status still write.
write_file_tool and patch_tool both call _update_read_timestamp to
refresh the staleness tracker after writing, but they never invalidate
the dedup cache entries for the written path. The dedup cache keys are
(resolved_path, offset, limit) → mtime tuples populated by read_file_tool.
On filesystems where a read and write land in the same mtime second (or
when mtime granularity is 1s), the cached and current mtime are equal,
so the dedup check incorrectly returns a 'File unchanged since last
read' stub — even though the file was just overwritten.
The agent then sees stale content (or a stale 'File not found' error)
and enters expensive error-recovery loops, burning API calls.
Fix: add _invalidate_dedup_for_path(filepath, task_id) that removes all
dedup entries whose resolved path matches the written file. Called from
_update_read_timestamp so both write_file_tool and patch_tool benefit
automatically. Scoped to the writing task_id — other tasks' caches are
not affected.
6 regression tests added covering:
- read→write→read within same mtime second (core #13144 scenario)
- invalidation across all offset/limit combinations
- isolation: writing file A does not invalidate file B's cache
- isolation: writing in task A does not invalidate task B's cache
- _invalidate_dedup_for_path safety on missing task / empty dedup
All 25 tests pass (19 existing + 6 new).
Fixes#13144
MCP stdio servers are spawned via the SDK's stdio_client, which on
Linux uses start_new_session=True (setsid). When a cron job is
cancelled mid-way (timeout, agent finish, exception), the subprocess
often escapes the SDK's teardown and survives as a session leader.
Because setsid() detaches the child from the gateway's process group
/ cgroup tree, systemd does not reap it on service restart either —
so every cron tick that touches an MCP tool leaks a dangling server
process.
Fix:
* tools/mcp_tool.py — _run_stdio now wraps the whole stdio+session
context in try/finally. On any exit path (clean, exception,
cancellation), PIDs still alive are moved from the active
_stdio_pids set into a new _orphan_stdio_pids set. Orphan
detection is done via os.kill(pid, 0) — a cheap liveness probe
that never signals the target.
* tools/mcp_tool.py — _kill_orphaned_mcp_children gains an
include_active=False flag. Default behaviour now only reaps the
orphan set so concurrent sessions (other parallel cron jobs or
live user chats) are never disrupted. The existing shutdown path
passes include_active=True to keep the previous "kill everything"
semantics after the MCP loop is stopped.
* cron/scheduler.py — the cleanup hook is moved from run_job()'s
finally (which would race with parallel siblings after #13021)
into tick() after the ThreadPoolExecutor has joined every future.
At that point there are no in-flight sessions from this tick, so
sweeping the orphan set is always safe.
Net effect: zero regression for healthy sessions, and orphan MCP
servers no longer accumulate between gateway restarts.
Made-with: Cursor
Slack's chat.postMessage API rejects user IDs (U...) and workspace
IDs (W...) — they are not valid conversation IDs. Posting to them
fails because the API requires a channel ID (C/G/D). To DM a user,
the sender must first call conversations.open to obtain a D... ID.
Tighten _SLACK_TARGET_RE from [CGDUW] to [CGD] so the send path rejects
U/W values as explicit targets and instead falls through to channel-
name resolution (where they'll fail with a clear 'could not resolve'
error rather than silently getting stuck in a retry loop on the API).
Flip the corresponding regression test to assert U/W values are not
explicit. Matches the narrower regex briandevans proposed in #15939.
Co-authored-by: briandevans <brian@bde.io>
send_message(target='slack:<channel_id>') failed with "Could not
resolve" because _parse_target_ref had no Slack branch — Slack's
uppercase alphanumeric IDs fell through to channel-name resolution,
which only matched by name. As a fallback, the agent would retry with
bare target='slack' and post to the home channel instead.
Three fixes:
- _parse_target_ref recognizes Slack IDs (C/G/D/U/W prefix) as
explicit targets so the name-resolver is bypassed entirely.
- resolve_channel_name tries a case-sensitive raw-ID match before
the existing name match, so any platform's IDs resolve cleanly.
- _build_slack now actually calls users.conversations against each
workspace's AsyncWebClient (paginated), instead of only returning
session-history entries. This populates the directory with public
and private channels the bot has joined, so action='list' shows
them and they can also be addressed by name. Errors from one
workspace don't block others.
build_channel_directory becomes async (Slack web calls require it).
The two async-context callers in gateway/run.py are awaited; the
cron ticker thread call bridges via asyncio.run_coroutine_threadsafe.
Slack bot needs channels:read and groups:read scopes for full
enumeration; missing scopes degrade gracefully per-workspace.
addressing #15927
Before: delegate_task children each allocated their own terminal
sandbox keyed by child task_id. Starting extra containers (or Modal
sandboxes / Daytona workspaces) is expensive, and the subagent's work
is invisible to the parent — files written by the child in its
container don't exist in the parent's when the subagent returns.
After: a single `_resolve_container_task_id` helper maps any
tool-call task_id to "default" UNLESS an env override is registered
for it. The parent agent and all delegate_task children therefore
share one long-lived sandbox — installed packages, cwd, /workspace
files, and /tmp scratch carry over freely between them.
RL and benchmark environments (TerminalBench2, HermesSweEnv, ...)
opt in to isolation via `register_task_env_overrides(task_id, {...})`;
those task_ids survive the collapse and get their own sandbox,
preserving the per-task Docker image behavior these benchmarks rely on.
file_state / active-subagents registry / TUI events still key off the
original child task_id, so the 'subagent wrote a file the parent read'
warning and UI per-subagent panels keep working.
Tradeoff: parallel delegate_task children (tasks=[...]) now share one
bash/container. Concurrent cd, env-var mutations, and writes to the
same path will collide. If that bites a specific workflow, the
subagent can opt back into isolation via register_task_env_overrides.
Applied at four lookup sites:
- tools/terminal_tool.py terminal_tool() and get_active_env()
- tools/file_tools.py _get_file_ops() and _get_live_tracking_cwd()
- tools/code_execution_tool.py _get_or_create_environment()
Docs: website/docs/user-guide/configuration.md updated to reflect the
shared-container reality and document the RL/benchmark carve-out.
Tests: tests/tools/test_shared_container_task_id.py (9 cases).
When a cloud browser provider (Browserbase / Browser-Use / Firecrawl) is
configured, browser_navigate now transparently spawns a local Chromium
sidecar for URLs whose host resolves to a private/loopback/LAN address
(localhost, 127.0.0.1, 192.168.x.x, 10.x.x.x, *.local, *.lan, *.internal,
::1, 169.254.x.x). Public URLs continue to use the cloud provider in the
same conversation.
Previously, setting BROWSERBASE_API_KEY / cloud_provider: browserbase
pinned the whole tool to cloud for the process — localhost URLs were
either SSRF-blocked (default) or sent to Browserbase (where they 404'd
because the cloud can't reach your LAN). Users who wanted 'cloud for
public, local for localhost' had no way to express it short of toggling
providers mid-session.
Implementation uses a composite session key scheme: the bare task_id
serves the cloud session, and a '{task_id}::local' sidecar serves the
local Chromium. _last_active_session_key[task_id] tracks which of the
two served the most recent nav so snapshot/click/fill/etc. hit the
correct one. cleanup_browser(bare_task_id) reaps both.
Feature is on by default. Opt out via:
browser:
auto_local_for_private_urls: false
The cloud provider never sees private URLs. Post-redirect SSRF guard
is preserved: redirects from public onto private addresses still block.
Follow-up to #6616 covering the remaining user-injected prompt markers that
the original PR did not touch (reporter's second comment on #6576 explicitly
flagged these). Azure OpenAI Default/DefaultV2 content filters treat any
bracketed [SYSTEM: ...] as prompt-injection and reject with HTTP 400.
Remaining call sites renamed:
- cli.py: background-process notifications (watch_disabled, watch_match,
completion), MCP reload notice (4 live + 1 docstring)
- gateway/run.py: same notification paths + auto-loaded skill banner +
MCP reload notice (5 live + 1 docstring)
- tools/process_registry.py: comment reference
Not renamed:
- environments/hermes_base_env.py '[SYSTEM]\n{content}' — RL training
trajectory rendering only, never sent to Azure, part of a symmetric
[USER]/[ASSISTANT]/[TOOL] scheme.
AUTHOR_MAP: buraysandro9@gmail.com -> ygd58.
Stop pre-stripping the path from the configured MCP server URL before
constructing OAuthClientProvider. The MCP SDK strips the path itself via
OAuthContext.get_authorization_base_url() for authorization-server
discovery, but uses the full server_url through
resource_url_from_server_url() + check_resource_allowed() to validate
against the server's RFC 9728 Protected Resource Metadata.
For servers whose PRM advertises a path-scoped resource (e.g. Notion's
https://mcp.notion.com/mcp), our _parse_base_url() collapsed the URL to
the origin, so check_resource_allowed() saw requested='/' vs
configured='/mcp/' and refused the token. Fixes OAuth against Notion MCP
(and any other path-scoped resource).
Closes#16015.
Adds a floor below --yolo: a tiny set of commands so catastrophic they
should never run via the agent, regardless of --yolo, gateway /yolo,
approvals.mode=off, or cron approve mode. Opting into yolo is trusting
the agent with your files and services — not trusting it to wipe the
disk or power the box off.
The list is deliberately small (12 patterns), covering only
unrecoverable ops:
- rm -rf targeting /, /home, /etc, /usr, /var, /boot, /bin, /sbin,
/lib, ~, $HOME
- mkfs (any variant)
- dd + redirection to raw block devices (/dev/sd*, /dev/nvme*, etc.)
- fork bomb
- kill -1 / kill -9 -1
- shutdown, reboot, halt, poweroff, init 0/6, telinit 0/6,
systemctl poweroff/reboot/halt/kexec
Recoverable-but-costly commands (git reset --hard, rm -rf /tmp/x,
chmod -R 777, curl | sh) stay in DANGEROUS_PATTERNS where yolo can
still pass them through — that's what yolo is for.
Container backends (docker/singularity/modal/daytona) continue to
bypass both hardline and dangerous checks, since nothing they do can
touch the host.
Inspired by Mercury Agent's permission-hardened blocklist.
* fix(terminal): three-layer defense against watch_patterns notification spam
Background processes that stack notify_on_complete=True with watch_patterns
can flood the user with duplicate, delayed notifications — matches deliver
asynchronously via the completion queue and continue arriving minutes after
the process has exited. The docstring warning against this (PR #12113) has
proven insufficient; agents still misuse the combination.
Three layered defenses, each sufficient on its own:
1. Mutual exclusion (terminal_tool.py): When both flags are set on a
background process, drop watch_patterns with a warning. notify_on_complete
wins because 'let me know when it's done' is the more useful signal and
fires exactly once. Extracted as _resolve_notification_flag_conflict() so
the rule is testable in isolation.
2. Suppress-after-exit (process_registry.py): _check_watch_patterns() now
bails the moment session.exited is True. Post-exit chunks (buffered reads
draining after the process is gone) no longer produce notifications. This
is the fix flagged as future work in session 20260418_020302_79881c.
3. Global circuit breaker (process_registry.py): Per-session rate limits don't
catch the sibling-flood case — N concurrent processes can each stay under
8/10s and still collectively spam. New WATCH_GLOBAL_MAX_PER_WINDOW=15 cap
trips a 30-second cooldown across ALL sessions, emits a single
watch_overflow_tripped event, silently counts dropped events, and emits a
watch_overflow_released summary when the cooldown ends.
Also updates the tool schema + docstring to document the new behavior.
Tests: 8 new tests covering all three fixes (suppress-after-exit x2,
mutual-exclusion resolver x4, global breaker trip/cooldown/release x2).
All 60 tests across test_watch_patterns.py, test_notify_on_complete.py,
test_terminal_tool.py pass.
Real-world trigger: self-inflicted in session 20260425_051924 — three
concurrent hermes-sweeper review subprocesses each set watch_patterns=
['failed validation', 'errored'] AND notify_on_complete=True, then iterated
over multiple items, producing enough matches per process to defeat the
per-session cap while staying under the global cap that didn't yet exist.
* fix(terminal): aggressive 1-per-15s watch_patterns rate limit + strike-3 promotion
Per Teknium's direction, the watch_patterns rate limit is now much more
aggressive and self-healing.
## New rule — per session
- HARD cap: 1 watch-match notification per 15 seconds per process.
- Any match arriving inside the cooldown window is dropped and counts as
ONE strike for that window (many drops in the same window still = 1 strike).
- After 3 consecutive strike windows, watch_patterns is permanently disabled
for the session and the session is auto-promoted to notify_on_complete
semantics — exactly one notification when the process actually exits.
- A cooldown window that expires with zero drops resets the consecutive
strike counter — healthy cadence is forgiven.
## Schema + docstring rewritten
The tool schema description now gives the model explicit guidance:
- notify_on_complete is 'the right choice for almost every long-running task'
- watch_patterns is for RARE one-shot signals on LONG-LIVED processes
- Do NOT use watch_patterns with loops/batch jobs — error patterns fire every
iteration and will hit the strike limit fast
- Mutual exclusion is stated on both parameter descriptions
- 1/15s cooldown and 3-strike promotion are stated in the watch_patterns
description so the model sees the contract every turn
## Removed
- WATCH_MAX_PER_WINDOW (8/10s) and WATCH_OVERLOAD_KILL_SECONDS (45) — the
new 1/15s limit subsumes both; keeping them would double-count.
- _watch_window_hits / _watch_window_start / _watch_overload_since fields
on ProcessSession. Replaced by _watch_last_emit_at / _watch_cooldown_until
/ _watch_strike_candidate / _watch_consecutive_strikes.
## Kept
- Global circuit breaker across all sessions (15/10s → 30s cooldown) as a
secondary safety net for concurrent siblings. Still valuable when 20
short-lived processes each fire once — none individually violates the
per-session limit.
- Suppress-after-exit guard.
- Mutual exclusion resolver at the tool entry point.
## Tests
- 6 new tests in TestPerSessionRateLimit covering: first match delivers,
second in cooldown suppressed, multi-drop = single strike, 3 strikes
disables + promotes, clean window resets counter, suppressed count
carried to next emit.
- Global circuit breaker tests rewritten to use fresh sessions instead of
hacking removed per-window fields.
- 50/50 watch_patterns + notify_on_complete tests pass.
- 60/60 including test_terminal_tool.py pass.
Split the monolithic discord_server tool (14 actions) into two:
- discord: core actions (fetch_messages, search_members, create_thread)
that are useful for the agent's normal operation. Auto-enabled on
the discord platform via the pipeline fix.
- discord_admin: server management actions (list channels/roles, pins,
role assignment) that require explicit opt-in via hermes tools.
Added to CONFIGURABLE_TOOLSETS and _DEFAULT_OFF_TOOLSETS.
The tool schema promised 'On update, pass an empty array to clear' but the
update branch ignored the context_from kwarg entirely — users could set
the field at create time and never modify or clear it afterward.
- tools/cronjob_tools.py: handle context_from in the update branch the
same way script/enabled_toolsets/workdir are handled: normalize str/list
to refs, validate each referenced job exists (same check the create
branch does), store as list-or-None to match create_job()'s shape.
Empty string or empty list clears the field.
- tests/cron/test_cron_context_from.py: 6 new tests covering add/change/
clear (both shapes)/bad-ref/preserve-across-unrelated-update.
Subagents run inside a ThreadPoolExecutor. The CLI's interactive approval
callback lives in tools/terminal_tool.py's threading.local(), which worker
threads do not inherit. When a subagent hits a dangerous-command guard,
prompt_dangerous_approval() falls back to input() from the worker thread,
deadlocking against the parent's prompt_toolkit TUI that owns stdin.
Fix: install a non-interactive callback into every subagent worker thread
via ThreadPoolExecutor(initializer=set_approval_callback, initargs=(cb,)).
The callback is config-gated by delegation.subagent_auto_approve:
false (default) -> _subagent_auto_deny (safe; matches leaf tool blocklist)
true -> _subagent_auto_approve (opt-in YOLO for cron/batch)
Both emit a logger.warning audit line. Gateway sessions are unaffected
because they resolve approvals via tools/approval.py's per-session queue,
not through these TLS callbacks. Diagnosis credit: @MorAlekss (#14685).
- hermes_cli/config.py: DEFAULT_CONFIG.delegation.subagent_auto_approve: False
- cli-config.yaml.example: documented, commented (default)
- tools/delegate_tool.py: _subagent_auto_deny, _subagent_auto_approve,
_get_subagent_approval_callback, wired into the child timeout executor
- tests/tools/test_delegate.py: 7 tests covering defaults, truthy coercion,
and TLS scoping in the worker thread
skill_view response went to the model verbatim; duplicating the SKILL.md
body as raw_content on every tool call added token cost with no agent-facing
benefit. Remove the field and update tests to assert on content only.
The slash/preload caller (agent/skill_commands.py) already falls back to
content when raw_content is absent, and it calls skill_view(preprocess=False)
anyway, so content is already unrendered on that path.
A child running a legitimately long-running tool (terminal command,
browser fetch, big file read) holds current_tool set and keeps
api_call_count frozen while the tool runs. The previous stale check
treated that as idle after 5 heartbeat cycles (~150s), stopped
touching the parent, and let the gateway kill the session.
Split the threshold in two:
- _HEARTBEAT_STALE_CYCLES_IDLE=5 (~150s) — applied only when
current_tool is None (child wedged between turns)
- _HEARTBEAT_STALE_CYCLES_IN_TOOL=20 (~600s) — applied when the child
is inside a tool call
Stale counter also resets when current_tool changes (new tool =
progress). The hard child_timeout_seconds (default 600s) is still
the final cap, so genuinely stuck tools don't get to block forever.
Moves the Spotify integration from tools/ into plugins/spotify/,
matching the existing pattern established by plugins/image_gen/ for
third-party service integrations.
Why:
- tools/ should be reserved for foundational capabilities (terminal,
read_file, web_search, etc.). tools/providers/ was a one-off
directory created solely for spotify_client.py.
- plugins/ is already the home for image_gen backends, memory
providers, context engines, and standalone hook-based plugins.
Spotify is a third-party service integration and belongs alongside
those, not in tools/.
- Future service integrations (eventually: Deezer, Apple Music, etc.)
now have a pattern to copy.
Changes:
- tools/spotify_tool.py → plugins/spotify/tools.py (handlers + schemas)
- tools/providers/spotify_client.py → plugins/spotify/client.py
- tools/providers/ removed (was only used for Spotify)
- New plugins/spotify/__init__.py with register(ctx) calling
ctx.register_tool() × 7. The handler/check_fn wiring is unchanged.
- New plugins/spotify/plugin.yaml (kind: backend, bundled, auto-load).
- tests/tools/test_spotify_client.py: import paths updated.
tools_config fix — _DEFAULT_OFF_TOOLSETS now wins over plugin auto-enable:
- _get_platform_tools() previously auto-enabled unknown plugin
toolsets for new platforms. That was fine for image_gen (which has
no toolset of its own) but bad for Spotify, which explicitly
requires opt-in (don't ship 7 tool schemas to users who don't use
it). Added a check: if a plugin toolset is in _DEFAULT_OFF_TOOLSETS,
it stays off until the user picks it in 'hermes tools'.
Pre-existing test bug fix:
- tests/hermes_cli/test_plugins.py::test_list_returns_sorted
asserted names were sorted, but list_plugins() sorts by key
(path-derived, e.g. image_gen/openai). With only image_gen plugins
bundled, name and key order happened to agree. Adding plugins/spotify
broke that coincidence (spotify sorts between openai-codex and xai
by name but after xai by key). Updated test to assert key order,
which is what the code actually documents.
Validation:
- scripts/run_tests.sh tests/hermes_cli/test_plugins.py \
tests/hermes_cli/test_tools_config.py \
tests/hermes_cli/test_spotify_auth.py \
tests/tools/test_spotify_client.py \
tests/tools/test_registry.py
→ 143 passed
- E2E plugin load: 'spotify' appears in loaded plugins, all 7 tools
register into the spotify toolset, check_fn gating intact.
Three quality improvements on top of #15121 / #15130 / #15135:
1. Tool consolidation (9 → 7)
- spotify_saved_tracks + spotify_saved_albums → spotify_library with
kind='tracks'|'albums'. Handler code was ~90 percent identical
across the two old tools; the merge is a behavioral no-op.
- spotify_activity dropped. Its 'now_playing' action was a duplicate
of spotify_playback.get_currently_playing (both return identical
204/empty payloads). Its 'recently_played' action moves onto
spotify_playback as a new action — history belongs adjacent to
live state.
- Net: each API call ships 2 fewer tool schemas when the Spotify
toolset is enabled, and the action surface is more discoverable
(everything playback-related is on one tool).
2. Spotify skill (skills/media/spotify/SKILL.md)
Teaches the agent canonical usage patterns so common requests don't
balloon into 4+ tool calls:
- 'play X' = one search, then play by URI (not search + scan +
describe + play)
- 'what's playing' = single get_currently_playing (no preflight
get_state chain)
- Don't retry on '403 Premium required' or '403 No active device' —
both require user action
- URI/URL/bare-ID format normalization
- Full failure-mode reference for 204/401/403/429
3. Surfaced in 'hermes setup' tool status
Adds 'Spotify (PKCE OAuth)' to the tool status list when
auth.json has a Spotify access/refresh token. Matches the
homeassistant pattern but reads from auth.json (OAuth-based) rather
than env vars.
Docs updated to reflect the new 7-tool surface, and mention the
companion skill in the 'Using it' section.
Tests: 54 passing (client 22, auth 15, tools_config 35 — 18 = 54 after
renaming/replacing the spotify_activity tests with library +
recently_played coverage). Docusaurus build clean.
Streamable HTTP MCP servers may garbage-collect their server-side
session state while the OAuth token remains valid — idle TTL, server
restart, pod rotation, etc. Before this fix, the tool-call handler
treated the resulting "Invalid or expired session" error as a plain
tool failure with no recovery path, so **every subsequent call on
the affected server failed until the gateway was manually
restarted**. Reporter: #13383.
The OAuth-based recovery path (``_handle_auth_error_and_retry``)
already exists for 401s, but it only fires on auth errors. Session
expiry slipped through because the access token is still valid —
nothing 401'd, so the existing recovery branch was skipped.
Fix
---
Add a sibling function ``_handle_session_expired_and_retry`` that
detects MCP session-expiry via ``_is_session_expired_error`` (a
narrow allow-list of known-stable substrings: ``"invalid or expired
session"``, ``"session expired"``, ``"session not found"``,
``"unknown session"``, etc.) and then uses the existing transport
reconnect mechanism:
* Sets ``MCPServerTask._reconnect_event`` — the server task's
lifecycle loop already interprets this as "tear down the current
``streamablehttp_client`` + ``ClientSession`` and rebuild them,
reusing the existing OAuth provider instance".
* Waits up to 15 s for the new session to come back ready.
* Retries the original call once. If the retry succeeds, returns
its result and resets the circuit-breaker error count. If the
retry raises, or if the reconnect doesn't ready in time, falls
through to the caller's generic error path.
Unlike the 401 path, this does **not** call ``handle_401`` — the
access token is already valid and running an OAuth refresh would be
a pointless round-trip.
All 5 MCP handlers (``call_tool``, ``list_resources``, ``read_resource``,
``list_prompts``, ``get_prompt``) now consult both recovery paths
before falling through:
recovered = _handle_auth_error_and_retry(...) # 401 path
if recovered is not None: return recovered
recovered = _handle_session_expired_and_retry(...) # new
if recovered is not None: return recovered
# generic error response
Narrow scope — explicitly not changed
-------------------------------------
* **Detection is string-based on a 5-entry allow-list.** The MCP
SDK wraps JSON-RPC errors in ``McpError`` whose exception type +
attributes vary across SDK versions, so matching on message
substrings is the durable path. Kept narrow to avoid false
positives — a regular ``RuntimeError("Tool failed")`` will NOT
trigger spurious reconnects (pinned by
``test_is_session_expired_rejects_unrelated_errors``).
* **No change to the existing 401 recovery flow.** The new path is
consulted only after the auth path declines (returns ``None``).
* **Retry count stays at 1.** If the reconnect-then-retry also
fails, we don't loop — the error surfaces normally so the model
sees a failed tool call rather than a hang.
* **``InterruptedError`` is explicitly excluded** from session-expired
detection so user-cancel signals always short-circuit the same
way they did before (pinned by
``test_is_session_expired_rejects_interrupted_error``).
Regression coverage
-------------------
``tests/tools/test_mcp_tool_session_expired.py`` (new, 16 cases):
Unit tests for ``_is_session_expired_error``:
* ``test_is_session_expired_detects_invalid_or_expired_session`` —
reporter's exact wpcom-mcp text.
* ``test_is_session_expired_detects_expired_session_variant`` —
"Session expired" / "expired session" variants.
* ``test_is_session_expired_detects_session_not_found`` — server GC
variant ("session not found", "unknown session").
* ``test_is_session_expired_is_case_insensitive``.
* ``test_is_session_expired_rejects_unrelated_errors`` — narrow-scope
canary: random RuntimeError / ValueError / 401 don't trigger.
* ``test_is_session_expired_rejects_interrupted_error`` — user cancel
must never route through reconnect.
* ``test_is_session_expired_rejects_empty_message``.
Handler integration tests:
* ``test_call_tool_handler_reconnects_on_session_expired`` — reporter's
full repro: first call raises "Invalid or expired session", handler
signals ``_reconnect_event``, retries once, returns the retry's
success result with no ``error`` key.
* ``test_call_tool_handler_non_session_expired_error_falls_through``
— preserved-behaviour canary: random tool failures do NOT trigger
reconnect.
* ``test_session_expired_handler_returns_none_without_loop`` —
defensive: cold-start / shutdown race.
* ``test_session_expired_handler_returns_none_without_server_record``
— torn-down server falls through cleanly.
* ``test_session_expired_handler_returns_none_when_retry_also_fails``
— no retry loop on repeated failure.
Parametrised across all 4 non-``tools/call`` handlers:
* ``test_non_tool_handlers_also_reconnect_on_session_expired``
[list_resources / read_resource / list_prompts / get_prompt].
**15 of 16 fail on clean ``origin/main`` (``6fb69229``)** with
``ImportError: cannot import name '_is_session_expired_error'``
— the fix's surface symbols don't exist there yet. The 1 passing
test is an ordering artefact of pytest-xdist worker collection.
Validation
----------
``source venv/bin/activate && python -m pytest
tests/tools/test_mcp_tool_session_expired.py -q`` → **16 passed**.
Broader MCP suite (5 files:
``test_mcp_tool.py``, ``test_mcp_tool_401_handling.py``,
``test_mcp_tool_session_expired.py``, ``test_mcp_reconnect_signal.py``,
``test_mcp_oauth.py``) → **230 passed, 0 regressions**.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add event hook to httpx.AsyncClient in MCP HTTP transport that strips
Authorization headers when a redirect targets a different origin,
preventing credential leakage to third-party servers.
``tools/mcp_oauth.py`` relied on ``assert _oauth_port is not None`` to
guard the module-level port set by ``build_oauth_auth``. Python's
``-O`` / ``-OO`` optimization flags strip ``assert`` statements
entirely, so a deployment that runs ``python -O -m hermes ...``
silently loses the check: ``_oauth_port`` stays ``None`` and the
failure surfaces much later as an obscure ``int()`` or
``http.server.HTTPServer((host, None))`` TypeError rather than the
intended "OAuth callback port not set" signal.
Replace with an explicit ``if … raise RuntimeError(...)`` so the
invariant is preserved regardless of the interpreter's optimization
level. Docstring updated to document the new exception.
Found during a proactive audit of ``assert`` statements in
non-test code paths.
OAuth client information and token responses from the MCP SDK contain
Pydantic AnyUrl fields (client_uri, redirect_uris, etc.). The previous
model_dump() call returned a dict with these AnyUrl objects still as
their native Python type, which then crashed json.dumps with:
TypeError: Object of type AnyUrl is not JSON serializable
This caused any OAuth-based MCP server (e.g. alphaxiv) to fail
registration with an "OAuth flow error" traceback during startup.
Adding mode="json" tells Pydantic to serialize all fields to
JSON-compatible primitives (AnyUrl -> str, datetime -> ISO string, etc.)
before returning the dict, so the standard json.dumps can handle it.
Three call sites fixed:
- HermesTokenStorage.set_tokens
- HermesTokenStorage.set_client_info
- build_oauth_auth pre-registration write
Cron jobs can now specify a per-job working directory. When set, the job
runs as if launched from that directory: AGENTS.md / CLAUDE.md /
.cursorrules from that dir are injected into the system prompt, and the
terminal / file / code-exec tools use it as their cwd (via TERMINAL_CWD).
When unset, old behaviour is preserved (no project context files, tools
use the scheduler's cwd).
Requested by @bluthcy.
## Mechanism
- cron/jobs.py: create_job / update_job accept 'workdir'; validated to
be an absolute existing directory at create/update time.
- cron/scheduler.py run_job: if job.workdir is set, point TERMINAL_CWD
at it and flip skip_context_files to False before building the agent.
Restored in finally on every exit path.
- cron/scheduler.py tick: workdir jobs run sequentially (outside the
thread pool) because TERMINAL_CWD is process-global. Workdir-less jobs
still run in the parallel pool unchanged.
- tools/cronjob_tools.py + hermes_cli/cron.py + hermes_cli/main.py:
expose 'workdir' via the cronjob tool and 'hermes cron create/edit
--workdir ...'. Empty string on edit clears the field.
## Validation
- tests/cron/test_cron_workdir.py (21 tests): normalize, create, update,
JSON round-trip via cronjob tool, tick partition (workdir jobs run on
the main thread, not the pool), run_job env toggle + restore in finally.
- Full targeted suite (tests/cron/, test_cronjob_tools.py, test_cron.py,
test_config_cwd_bridge.py, test_worktree.py): 314/314 passed.
- Live smoke: hermes cron create --workdir $(pwd) works; relative path
rejected; list shows 'Workdir:'; edit --workdir '' clears.
When a subagent in delegate_task times out before making its first LLM
request, write a structured diagnostic file under
~/.hermes/logs/subagent-timeout-<sid>-<ts>.log capturing enough state
for the user (and us) to debug the hang. The old error message —
'Subagent timed out after Ns with no response. The child may be stuck
on a slow API call or unresponsive network request.' — gave no
observability for the 0-API-call case, which is the hardest to reason
about remotely.
The diagnostic captures:
- timeout config vs actual duration
- goal (truncated to 1000 chars)
- child config: model, provider, api_mode, base_url, max_iterations,
quiet_mode, platform, _delegate_role, _delegate_depth
- enabled_toolsets + loaded tool names
- system prompt byte/char count (catches oversized prompts that
providers silently choke on)
- tool schema count + byte size
- child's get_activity_summary() snapshot
- Python stack of the worker thread at the moment of timeout
(reveals whether the hang is in credential resolution, transport,
prompt construction, etc.)
Wiring:
- _run_single_child captures the worker thread via a small wrapper
around child.run_conversation so we can look up its stack at
timeout.
- After a FuturesTimeoutError, we pull child.get_activity_summary()
to read api_call_count. If 0 AND it was a timeout (not a raise),
_dump_subagent_timeout_diagnostic() is invoked.
- The returned path is surfaced in the error string so the parent
agent (and therefore the user / gateway) sees exactly where to look.
- api_calls > 0 timeouts keep the old 'stuck on slow API call'
phrasing since that's the correct diagnosis for those.
This does NOT change any behavior for successful subagent runs,
non-timeout errors, or subagents that made at least one API call
before hanging.
Tests: 7 cases (tests/tools/test_delegate_subagent_timeout_diagnostic.py)
- output format + required sections + field values
- long-goal truncation with [truncated] marker
- missing / already-exited worker thread branches
- unwritable HERMES_HOME/logs/ returns None without raising
- _run_single_child wiring: 0 API calls → dump + diagnostic_path in error
- _run_single_child wiring: N>0 API calls → no dump, old message
Refs: #14726
Make the main-branch test suite pass again. Most failures were tests
still asserting old shapes after recent refactors; two were real source
bugs.
Source fixes:
- tools/mcp_tool.py: _kill_orphaned_mcp_children() slept 2s on every
shutdown even when no tracked PIDs existed, making test_shutdown_is_parallel
measure ~3s for 3 parallel 1s shutdowns. Early-return when pids is empty.
- hermes_cli/tips.py: tip 105 was 157 chars; corpus max is 150.
Test fixes (mostly stale mock targets / missing fixture fields):
- test_zombie_process_cleanup, test_agent_cache: patch run_agent.cleanup_vm
(the local name bound at import), not tools.terminal_tool.cleanup_vm.
- test_browser_camofox: patch tools.browser_camofox.load_config, not
hermes_cli.config.load_config (the source module, not the resolved one).
- test_flush_memories_codex._chat_response_with_memory_call: add
finish_reason, tool_call.id, tool_call.type so the chat_completions
transport normalizer doesn't AttributeError.
- test_concurrent_interrupt: polling_tool signature now accepts
messages= kwarg that _invoke_tool() passes through.
- test_minimax_provider: add _fallback_chain=[] to the __new__'d agent
so switch_model() doesn't AttributeError.
- test_skills_config: SKILLS_DIR MagicMock + .rglob stopped working
after the scanner switched to agent.skill_utils.iter_skill_index_files
(os.walk-based). Point SKILLS_DIR at a real tmp_path and patch
agent.skill_utils.get_external_skills_dirs.
- test_browser_cdp_tool: browser_cdp toolset was intentionally split into
'browser-cdp' (commit 96b0f3700) so its stricter check_fn doesn't gate
the whole browser toolset; test now expects 'browser-cdp'.
- test_registry: add tools.browser_dialog_tool to the expected
builtin-discovery set (PR #14540 added it).
- test_file_tools TestPatchHints: patch_tool surfaces hints as a '_hint'
key on the JSON payload, not inline '[Hint: ...' text.
- test_write_deny test_hermes_env: resolve .env via get_hermes_home() so
the path matches the profile-aware denylist under hermetic HERMES_HOME.
- test_checkpoint_manager test_falls_back_to_parent: guard the walk-up
so a stray /tmp/pyproject.toml on the host doesn't pick up /tmp as the
project root.
- test_quick_commands: set cli.session_id in the __new__'d CLI so the
alias-args path doesn't trip AttributeError when fuzzy-matching leaks
a skill command across xdist test distribution.
faster-whisper's device="auto" picks CUDA when ctranslate2's wheel
ships CUDA shared libs, even on hosts without the NVIDIA runtime
(libcublas.so.12 / libcudnn*). On those hosts the model often loads
fine but transcribe() fails at first dlopen, and the broken model
stays cached in the module-global — every subsequent voice message
in the gateway process fails identically until restart.
- Add _load_local_whisper_model() wrapper: try auto, catch missing-lib
errors, retry on device=cpu compute_type=int8.
- Wrap transcribe() with the same fallback: evict cached model, reload
on CPU, retry once. Required because the dlopen failure only surfaces
at first kernel launch, not at model construction.
- Narrow marker list (libcublas, libcudnn, libcudart, 'cannot be loaded',
'no kernel image is available', 'no CUDA-capable device', driver
mismatch). Deliberately excludes 'CUDA out of memory' and similar —
those are real runtime failures that should surface, not be silently
retried on CPU.
- Tests for load-time fallback, runtime fallback (with cached-model
eviction verified), and the OOM non-fallback path.
Reported via Telegram voice-message dumps on WSL2 hosts where libcublas
isn't installed by default.
Local llama.cpp servers (e.g. ggml-org/llama.cpp:full-cuda) fail the entire
request with HTTP 400 'Unable to generate parser for this template. ...
Unrecognized schema: "object"' when any tool schema contains shapes its
json-schema-to-grammar converter can't handle:
* 'type': 'object' without 'properties'
* bare string schema values ('additionalProperties: "object"')
* 'type': ['X', 'null'] arrays (nullable form)
Cloud providers accept these silently, so they ship from external MCP
servers (Atlassian, GCloud, Datadog) and from a couple of our own tools.
Changes
- tools/schema_sanitizer.py: walks the finalized tool list right before it
leaves get_tool_definitions() and repairs the hostile shapes in a deep
copy. No-op on well-formed schemas. Recurses into properties, items,
additionalProperties, anyOf/oneOf/allOf, and $defs.
- model_tools.get_tool_definitions(): invoke the sanitizer as the last
step so all paths (built-in, MCP, plugin, dynamically-rebuilt) get
covered uniformly.
- tools/browser_cdp_tool.py, tools/mcp_tool.py: fix our own bare-object
schemas so sanitization isn't load-bearing for in-repo tools.
- tui_gateway/server.py: _load_enabled_toolsets() was passing
include_default_mcp_servers=False at runtime. That's the config-editing
variant (see PR #3252) — it silently drops every default MCP server
from the TUI's enabled_toolsets, which is why the TUI didn't hit the
llama.cpp crash (no MCP tools sent at all). Switch to True so TUI
matches CLI behavior.
Tests
tests/tools/test_schema_sanitizer.py (17 tests) covers the individual
failure modes, well-formed pass-through, deep-copy isolation, and
required-field pruning.
E2E: loaded the default 'hermes-cli' toolset with MCP discovery and
confirmed all 27 resolved tool schemas pass a llama.cpp-compatibility
walk (no 'object' node missing 'properties', no bare-string schema
values).
* docs: browser CDP supervisor design (for upcoming PR)
Design doc ahead of implementation — dialog + iframe detection/interaction
via a persistent CDP supervisor. Covers backend capability matrix (verified
live 2026-04-23), architecture, lifecycle, policy, agent surface, PR split,
non-goals, and test plan.
Supersedes #12550.
No code changes in this commit.
* feat(browser): add persistent CDP supervisor for dialog + frame detection
Single persistent CDP WebSocket per Hermes task_id that subscribes to
Page/Runtime/Target events and maintains thread-safe state for pending
dialogs, frame tree, and console errors.
Supervisor lives in its own daemon thread running an asyncio loop;
external callers use sync API (snapshot(), respond_to_dialog()) that
bridges onto the loop.
Auto-attaches to OOPIF child targets via Target.setAutoAttach{flatten:true}
and enables Page+Runtime on each so iframe-origin dialogs surface through
the same supervisor.
Dialog policies: must_respond (default, 300s safety timeout),
auto_dismiss, auto_accept.
Frame tree capped at 30 entries + OOPIF depth 2 to keep snapshot
payloads bounded on ad-heavy pages.
E2E verified against real Chrome via smoke test — detects + responds
to main-frame alerts, iframe-contentWindow alerts, preserves frame
tree, graceful no-dialog error path, clean shutdown.
No agent-facing tool wiring in this commit (comes next).
* feat(browser): add browser_dialog tool wired to CDP supervisor
Agent-facing response-only tool. Schema:
action: 'accept' | 'dismiss' (required)
prompt_text: response for prompt() dialogs (optional)
dialog_id: disambiguate when multiple dialogs queued (optional)
Handler:
SUPERVISOR_REGISTRY.get(task_id).respond_to_dialog(...)
check_fn shares _browser_cdp_check with browser_cdp so both surface and
hide together. When no supervisor is attached (Camofox, default
Playwright, or no browser session started yet), tool is hidden; if
somehow invoked it returns a clear error pointing the agent to
browser_navigate / /browser connect.
Registered in _HERMES_CORE_TOOLS and the browser / hermes-acp /
hermes-api-server toolsets alongside browser_cdp.
* feat(browser): wire CDP supervisor into session lifecycle + browser_snapshot
Supervisor lifecycle:
* _get_session_info lazy-starts the supervisor after a session row is
materialized — covers every backend code path (Browserbase, cdp_url
override, /browser connect, future providers) with one hook.
* cleanup_browser(task_id) stops the supervisor for that task first
(before the backend tears down CDP).
* cleanup_all_browsers() calls SUPERVISOR_REGISTRY.stop_all().
* /browser connect eagerly starts the supervisor for task 'default'
so the first snapshot already shows pending_dialogs.
* /browser disconnect stops the supervisor.
CDP URL resolution for the supervisor:
1. BROWSER_CDP_URL / browser.cdp_url override.
2. Fallback: session_info['cdp_url'] from cloud providers (Browserbase).
browser_snapshot merges supervisor state (pending_dialogs + frame_tree)
into its JSON output when a supervisor is active — the agent reads
pending_dialogs from the snapshot it already requests, then calls
browser_dialog to respond. No extra tool surface.
Config defaults:
* browser.dialog_policy: 'must_respond' (new)
* browser.dialog_timeout_s: 300 (new)
No version bump — new keys deep-merge into existing browser section.
Deadlock fix in supervisor event dispatch:
* _on_dialog_opening and _on_target_attached used to await CDP calls
while the reader was still processing an event — but only the reader
can set the response Future, so the call timed out.
* Both now fire asyncio.create_task(...) so the reader stays pumping.
* auto_dismiss/auto_accept now actually close the dialog immediately.
Tests (tests/tools/test_browser_supervisor.py, 11 tests, real Chrome):
* supervisor start/snapshot
* main-frame alert detection + dismiss
* iframe.contentWindow alert
* prompt() with prompt_text reply
* respond with no pending dialog -> clean error
* auto_dismiss clears on event
* registry idempotency
* registry stop -> snapshot reports inactive
* browser_dialog tool no-supervisor error
* browser_dialog invalid action
* browser_dialog end-to-end via tool handler
xdist-safe: chrome_cdp fixture uses a per-worker port.
Skipped when google-chrome/chromium isn't installed.
* docs(browser): document browser_dialog tool + CDP supervisor
- user-guide/features/browser.md: new browser_dialog section with
workflow, availability gate, and dialog_policy table
- reference/tools-reference.md: row for browser_dialog, tool count
bumped 53 -> 54, browser tools count 11 -> 12
- reference/toolsets-reference.md: browser_dialog added to browser
toolset row with note on pending_dialogs / frame_tree snapshot fields
Full design doc lives at
developer-guide/browser-supervisor.md (committed earlier).
* fix(browser): reconnect loop + recent_dialogs for Browserbase visibility
Found via Browserbase E2E test that revealed two production-critical issues:
1. **Supervisor WebSocket drops when other clients disconnect.** Browserbase's
CDP proxy tears down our long-lived WebSocket whenever a short-lived
client (e.g. agent-browser CLI's per-command CDP connection) disconnects.
Fixed with a reconnecting _run loop that re-attaches with exponential
backoff on drops. _page_session_id and _child_sessions are reset on each
reconnect; pending_dialogs and frames are preserved across reconnects.
2. **Browserbase auto-dismisses dialogs server-side within ~10ms.** Their
Playwright-based CDP proxy dismisses alert/confirm/prompt before our
Page.handleJavaScriptDialog call can respond. So pending_dialogs is
empty by the time the agent reads a snapshot on Browserbase.
Added a recent_dialogs ring buffer (capacity 20) that retains a
DialogRecord for every dialog that opened, with a closed_by tag:
* 'agent' — agent called browser_dialog
* 'auto_policy' — local auto_dismiss/auto_accept fired
* 'watchdog' — must_respond timeout auto-dismissed (300s default)
* 'remote' — browser/backend closed it on us (Browserbase)
Agents on Browserbase now see the dialog history with closed_by='remote'
so they at least know a dialog fired, even though they couldn't respond.
3. **Page.javascriptDialogClosed matching bug.** The event doesn't include a
'message' field (CDP spec has only 'result' and 'userInput') but our
_on_dialog_closed was matching on message. Fixed to match by session_id
+ oldest-first, with a safety assumption that only one dialog is in
flight per session (the JS thread is blocked while a dialog is up).
Docs + tests updated:
* browser.md: new availability matrix showing the three backends and
which mode (pending / recent / response) each supports
* developer-guide/browser-supervisor.md: three-field snapshot schema
with closed_by semantics
* test_browser_supervisor.py: +test_recent_dialogs_ring_buffer (12/12
passing against real Chrome)
E2E verified both backends:
* Local Chrome via /browser connect: detect + respond full workflow
(smoke_supervisor.py all 7 scenarios pass)
* Browserbase: detect via recent_dialogs with closed_by='remote'
(smoke_supervisor_browserbase_v2.py passes)
Camofox remains out of scope (REST-only, no CDP) — tracked for
upstream PR 3.
* feat(browser): XHR bridge for dialog response on Browserbase (FIXED)
Browserbase's CDP proxy auto-dismisses native JS dialogs within ~10ms, so
Page.handleJavaScriptDialog calls lose the race. Solution: bypass native
dialogs entirely.
The supervisor now injects Page.addScriptToEvaluateOnNewDocument with a
JavaScript override for window.alert/confirm/prompt. Those overrides
perform a synchronous XMLHttpRequest to a magic host
('hermes-dialog-bridge.invalid'). We intercept those XHRs via Fetch.enable
with a requestStage=Request pattern.
Flow when a page calls alert('hi'):
1. window.alert override intercepts, builds XHR GET to
http://hermes-dialog-bridge.invalid/?kind=alert&message=hi
2. Sync XHR blocks the page's JS thread (mirrors real dialog semantics)
3. Fetch.requestPaused fires on our WebSocket; supervisor surfaces
it as a pending dialog with bridge_request_id set
4. Agent reads pending_dialogs from browser_snapshot, calls browser_dialog
5. Supervisor calls Fetch.fulfillRequest with JSON body:
{accept: true|false, prompt_text: '...', dialog_id: 'd-N'}
6. The injected script parses the body, returns the appropriate value
from the override (undefined for alert, bool for confirm, string|null
for prompt)
This works identically on Browserbase AND local Chrome — no native dialog
ever fires, so Browserbase's auto-dismiss has nothing to race. Dialog
policies (must_respond / auto_dismiss / auto_accept) all still work.
Bridge is installed on every attached session (main page + OOPIF child
sessions) so iframe dialogs are captured too.
Native-dialog path kept as a fallback for backends that don't auto-dismiss
(so a page that somehow bypasses our override — e.g. iframes that load
after Fetch.enable but before the init-script runs — still gets observed
via Page.javascriptDialogOpening).
E2E VERIFIED:
* Local Chrome: 13/13 pytest tests green (12 original + new
test_bridge_captures_prompt_and_returns_reply_text that asserts
window.__ret === 'AGENT-SUPPLIED-REPLY' after agent responds)
* Browserbase: smoke_bb_bridge_v2.py runs 4/4 PASS:
- alert('BB-ALERT-MSG') dismiss → page.alert_ret = undefined ✓
- prompt('BB-PROMPT-MSG', 'default-xyz') accept with 'AGENT-REPLY'
→ page.prompt_ret === 'AGENT-REPLY' ✓
- confirm('BB-CONFIRM-MSG') accept → page.confirm_ret === true ✓
- confirm('BB-CONFIRM-MSG') dismiss → page.confirm_ret === false ✓
Docs updated in browser.md and developer-guide/browser-supervisor.md —
availability matrix now shows Browserbase at full parity with local
Chrome for both detection and response.
* feat(browser): cross-origin iframe interaction via browser_cdp(frame_id=...)
Adds iframe interaction to the CDP supervisor PR (was queued as PR 2).
Design: browser_cdp gets an optional frame_id parameter. When set, the
tool looks up the frame in the supervisor's frame_tree, grabs its child
cdp_session_id (OOPIF session), and dispatches the CDP call through the
supervisor's already-connected WebSocket via run_coroutine_threadsafe.
Why not stateless: on Browserbase, each fresh browser_cdp WebSocket
must re-negotiate against a signed connectUrl. The session info carries
a specific URL that can expire while the supervisor's long-lived
connection stays valid. Routing via the supervisor sidesteps this.
Agent workflow:
1. browser_snapshot → frame_tree.children[] shows OOPIFs with is_oopif=true
2. browser_cdp(method='Runtime.evaluate', frame_id=<OOPIF frame_id>,
params={'expression': 'document.title', 'returnByValue': True})
3. Supervisor dispatches the call on the OOPIF's child session
Supervisor state fixes needed along the way:
* _on_frame_detached now skips reason='swap' (frame migrating processes)
* _on_frame_detached also skips when the frame is an OOPIF with a live
child session — Browserbase fires spurious remove events when a
same-origin iframe gets promoted to OOPIF
* _on_target_detached clears cdp_session_id but KEEPS the frame record
so the agent still sees the OOPIF in frame_tree during transient
session flaps
E2E VERIFIED on Browserbase (smoke_bb_iframe_agent_path.py):
browser_cdp(method='Runtime.evaluate',
params={'expression': 'document.title', 'returnByValue': True},
frame_id=<OOPIF>)
→ {'success': True, 'result': {'value': 'Example Domain'}}
The iframe is <iframe src='https://example.com/'> inside a top-level
data: URL page on a real Browserbase session. The agent Runtime.evaluates
INSIDE the cross-origin iframe and gets example.com's title back.
Tests (tests/tools/test_browser_supervisor.py — 16 pass total):
* test_browser_cdp_frame_id_routes_via_supervisor — injects fake OOPIF,
verifies routing via supervisor, Runtime.evaluate returns 1+1=2
* test_browser_cdp_frame_id_missing_supervisor — clean error when no
supervisor attached
* test_browser_cdp_frame_id_not_in_frame_tree — clean error on bad
frame_id
Docs (browser.md and developer-guide/browser-supervisor.md) updated with
the iframe workflow, availability matrix now shows OOPIF eval as shipped
for local Chrome + Browserbase.
* test(browser): real-OOPIF E2E verified manually + chrome_cdp uses --site-per-process
When asked 'did you test the iframe stuff' I had only done a mocked
pytest (fake injected OOPIF) plus a Browserbase E2E. Closed the
local-Chrome real-OOPIF gap by writing /tmp/dialog-iframe-test/
smoke_local_oopif.py:
* 2 http servers on different hostnames (localhost:18905 + 127.0.0.1:18906)
* Chrome with --site-per-process so the cross-origin iframe becomes a
real OOPIF in its own process
* Navigate, find OOPIF in supervisor.frame_tree, call
browser_cdp(method='Runtime.evaluate', frame_id=<OOPIF>) which routes
through the supervisor's child session
* Asserts iframe document.title === 'INNER-FRAME-XYZ' (from the
inner page, retrieved via OOPIF eval)
PASSED on 2026-04-23.
Tried to embed this as a pytest but hit an asyncio version quirk between
venv (3.11) and the system python (3.13) — Page.navigate hangs in the
pytest harness but works in standalone. Left a self-documenting skip
test that points to the smoke script + describes the verification.
chrome_cdp fixture now passes --site-per-process so future iframe tests
can rely on OOPIF behavior.
Result: 16 pass + 1 documented-skip = 17 tests in
tests/tools/test_browser_supervisor.py.
* docs(browser): add dialog_policy + dialog_timeout_s to configuration.md, fix tool count
Pre-merge docs audit revealed two gaps:
1. user-guide/configuration.md browser config example was missing the
two new dialog_* knobs. Added with a short table explaining
must_respond / auto_dismiss / auto_accept semantics and a link to
the feature page for the full workflow.
2. reference/tools-reference.md header said '54 built-in tools' — real
count on main is 54, this branch adds browser_dialog so it's 55.
Fixed the header. (browser count was already correctly bumped
11 -> 12 in the earlier docs commit.)
No code changes.
* feat(config): make tool output truncation limits configurable
Port from anomalyco/opencode#23770: expose a new `tool_output` config
section so users can tune the hardcoded truncation caps that apply to
terminal output and read_file pagination.
Three knobs under `tool_output`:
- max_bytes (default 50_000) — terminal stdout/stderr cap
- max_lines (default 2000) — read_file pagination cap
- max_line_length (default 2000) — per-line cap in line-numbered view
All three keep their existing hardcoded values as defaults, so behaviour
is unchanged when the section is absent. Power users on big-context
models can raise them; small-context local models can lower them.
Implementation:
- New `tools/tool_output_limits.py` reads the section with defensive
fallback (missing/invalid values → defaults, never raises).
- `tools/terminal_tool.py` MAX_OUTPUT_CHARS now comes from
get_max_bytes().
- `tools/file_operations.py` normalize_read_pagination() and
_add_line_numbers() now pull the limits at call time.
- `hermes_cli/config.py` DEFAULT_CONFIG gains the `tool_output` section
so `hermes setup` writes defaults into fresh configs.
- Docs page `user-guide/configuration.md` gains a "Tool Output
Truncation Limits" section with large-context and small-context
example configs.
Tests (18 new in tests/tools/test_tool_output_limits.py):
- Default resolution with missing / malformed / non-dict config.
- Full and partial user overrides.
- Coercion of bad values (None, negative, wrong type, str int).
- Shortcut accessors delegate correctly.
- DEFAULT_CONFIG exposes the section with the right defaults.
- Integration: normalize_read_pagination clamps to the configured
max_lines.
* feat(skills): add design-md skill for Google's DESIGN.md spec
Built-in skill under skills/creative/ that teaches the agent to author,
lint, diff, and export DESIGN.md files — Google's open-source
(Apache-2.0) format for describing a visual identity to coding agents.
Covers:
- YAML front matter + markdown body anatomy
- Full token schema (colors, typography, rounded, spacing, components)
- Canonical section order + duplicate-heading rejection
- Component property whitelist + variants-as-siblings pattern
- CLI workflow via 'npx @google/design.md' (lint/diff/export/spec)
- Lint rule reference including WCAG contrast checks
- Common YAML pitfalls (quoted hex, negative dimensions, dotted refs)
- Starter template at templates/starter.md
Package verified live on npm (@google/design.md@0.1.1).
MCP stdio servers' stderr was being dumped directly onto the user's
terminal during hermes launch. Servers like FastMCP-based ones print a
large ASCII banner at startup; slack-mcp-server emits JSON logs; etc.
With prompt_toolkit / Rich rendering the TUI concurrently, these
unsolicited writes corrupt the terminal state — hanging the session
~80% of the time for one user with Google Ads Tools + slack-mcp
configured, forcing Ctrl+C and restart loops.
Root cause: `stdio_client(server_params)` in tools/mcp_tool.py was
called without `errlog=`, and the SDK's default is `sys.stderr` —
i.e. the real parent-process stderr, which is the TTY.
Fix: open a shared, append-mode log at $HERMES_HOME/logs/mcp-stderr.log
(created once per process, line-buffered, real fd required by asyncio's
subprocess machinery) and pass it as `errlog` to every stdio_client.
Each server's spawn writes a timestamped header so the shared log stays
readable when multiple servers are running. Falls back to /dev/null if
the log file cannot be opened.
Verified by E2E spawning a subprocess with the log fd as its stderr:
banner lines land in the log file, nothing reaches the calling TTY.
The 300s default was too tight for high-reasoning models on non-trivial
delegated tasks — e.g. gpt-5.5 xhigh reviewing 12 files would burn >5min
on reasoning tokens before issuing its first tool call, tripping the
hard wall-clock timeout with 0 api_calls logged.
- tools/delegate_tool.py: DEFAULT_CHILD_TIMEOUT 300 -> 600
- hermes_cli/config.py: surface delegation.child_timeout_seconds in
DEFAULT_CONFIG so it's discoverable (previously the key was read by
_get_child_timeout() but absent from the default config schema)
Users can still override via config.yaml delegation.child_timeout_seconds
or DELEGATION_CHILD_TIMEOUT_SECONDS env var (floor 30s, no ceiling).
Fixes a broader class of 'tools.function.parameters is not a valid
moonshot flavored json schema' errors on Nous / OpenRouter aggregators
routing to moonshotai/kimi-k2.6 with MCP tools loaded.
## Moonshot sanitizer (agent/moonshot_schema.py, new)
Model-name-routed (not base-URL-routed) so Nous / OpenRouter users are
covered alongside api.moonshot.ai. Applied in
ChatCompletionsTransport.build_kwargs when is_moonshot_model(model).
Two repairs:
1. Fill missing 'type' on every property / items / anyOf-child schema
node (structural walk — only schema-position dicts are touched, not
container maps like properties/$defs).
2. Strip 'type' at anyOf parents; Moonshot rejects it.
## MCP normalizer hardened (tools/mcp_tool.py)
Draft-07 $ref rewrite from PR #14802 now also does:
- coerce missing / null 'type' on object-shaped nodes (salvages #4897)
- prune 'required' arrays to names that exist in 'properties'
(salvages #4651; Gemini 400s on dangling required)
- apply recursively, not just top-level
These repairs are provider-agnostic so the same MCP schema is valid on
OpenAI, Anthropic, Gemini, and Moonshot in one pass.
## Crash fix: safe getattr for Tool.inputSchema
_convert_mcp_schema now uses getattr(t, 'inputSchema', None) so MCP
servers whose Tool objects omit the attribute entirely no longer abort
registration (salvages #3882).
## Validation
- tests/agent/test_moonshot_schema.py: 27 new tests (model detection,
missing-type fill, anyOf-parent strip, non-mutation, real-world MCP
shape)
- tests/tools/test_mcp_tool.py: 7 new tests (missing / null type,
required pruning, nested repair, safe getattr)
- tests/agent/transports/test_chat_completions.py: 2 new integration
tests (Moonshot route sanitizes, non-Moonshot route doesn't)
- Targeted suite: 49 passed
- E2E via execute_code with a realistic MCP tool carrying all three
Moonshot rejection modes + dangling required + draft-07 refs:
sanitizer produces a schema valid on Moonshot and Gemini
- _stdio_pids: set → Dict[int,str] tracks pid→server_name
- SIGTERM-first with 2s grace before SIGKILL escalation
- hasattr guard for SIGKILL on platforms without it
- Updated tests for dict-based tracking and 3-phase kill sequence
The original regex only matched relative paths (./foo/.env or bare
.env), so the exact command from the bug report —
`cp /opt/data/.env.local /opt/data/.env` — did not trigger approval.
Broaden the leading-path prefix to accept an absolute leading slash
alongside ./ and ../, and add regressions for the bug-report command
and its redirection variant.
Previously delegate_task exposed 'max_iterations' in its JSON schema and used
`max_iterations or default_max_iter` — so a model guessing conservatively (or
copy-pasting a docstring hint like 'Only set lower for simple tasks') could
silently shrink a subagent's budget below the user's configured
delegation.max_iterations. One such call this session capped a deep forensic
audit at 40 iterations while the user's config was set to 250.
Changes:
- Drop 'max_iterations' from DELEGATE_TASK_SCHEMA['parameters']['properties'].
Models can no longer emit it.
- In delegate_task(): ignore any caller-supplied max_iterations, always use
delegation.max_iterations from config. Log at debug if a stale schema or
internal caller still passes one through.
- Keep the Python kwarg on the function signature for internal callers
(_build_child_agent tests pass it through the plumbing layer).
- Update test_schema_valid to assert the param is now absent (intentional
contract change, not a change-detector).
Replaces the blanket 'always allow' change from the previous commit with
an opt-in config flag so users who want belt-and-suspenders security can
still get the keyword scan on skill_manage output.
## Default behavior (flag off)
skill_manage(action='create'|'edit'|'patch') no longer runs the keyword
scanner. The agent can write skills that mention risky keywords in prose
(documenting what reviewers should watch for, describing cache-bust
semantics in a PR-review skill, referencing AGENTS.md, etc.) without
getting blocked.
Rationale: the agent can already execute the same code paths via
terminal() with no gate, so the scan adds friction without meaningful
security against a compromised or malicious agent.
## Opt-in behavior (flag on)
Set skills.guard_agent_created: true in config.yaml to get the original
behavior back. Scanner runs on every skill_manage write; dangerous
verdicts surface as a tool error the agent can react to (retry without
the flagged content).
## External hub installs unaffected
trusted/community sources (hermes skills install) always get scanned
regardless of this flag. The gate is specifically for skill_manage,
which only agents call.
## Changes
- hermes_cli/config.py: add skills.guard_agent_created: False to DEFAULT_CONFIG
- tools/skill_manager_tool.py: _guard_agent_created_enabled() reads the flag;
_security_scan_skill() short-circuits to None when the flag is off
- tools/skills_guard.py: restore INSTALL_POLICY['agent-created'] =
('allow', 'allow', 'ask') so the scan remains strict when it does run
- tests/tools/test_skills_guard.py: restore original ask/force tests
- tests/tools/test_skill_manager_tool.py: new TestSecurityScanGate class
covering both flag states + config error handling
## Validation
- tests/tools/test_skills_guard.py + test_skill_manager_tool.py: 115/115 pass
- E2E: flagged-keyword skill creates with default config, blocks with flag on
The security scanner is meant to protect against hostile external skills
pulled from GitHub via hermes skills install — trusted/community policies
block or ask on dangerous verdicts accordingly. But agent-created skills
(from skill_manage) run in the same process as the agent that wrote them.
The agent can already execute the same code paths via terminal() with no
gate, so the ask-on-dangerous policy adds friction without meaningful
security.
Concrete trigger: an agent writing a PR-review skill that describes
cache-busting or persistence semantics in prose gets blocked because
those words appear in the patterns list. The skill isn't actually doing
anything dangerous — it's just documenting what reviewers should watch
for in other PRs.
Change: agent-created dangerous verdict maps to 'allow' instead of 'ask'.
External hub installs (trusted/community) keep their stricter policies
intact. Tests updated: renamed test_dangerous_agent_created_asks →
test_dangerous_agent_created_allowed; renamed force-override test and
updated assertion since force is now a no-op for agent-created (the allow
branch returns first).
The environment-snapshot login shell was auto-sourcing only ~/.bashrc when
building the PATH snapshot. On Debian/Ubuntu the default ~/.bashrc starts
with a non-interactive short-circuit:
case $- in *i*) ;; *) return;; esac
Sourcing it from a non-interactive shell returns before any PATH export
below that guard runs. Node version managers like n and nvm append their
PATH line under that guard, so Hermes was capturing a PATH without
~/n/bin — and the terminal tool saw 'node: command not found' even when
node was on the user's interactive shell PATH.
Expand the auto-source list (when auto_source_bashrc is on) to:
~/.profile → ~/.bash_profile → ~/.bashrc
~/.profile and ~/.bash_profile have no interactivity guard — installers
that write their PATH there (n's n-install, nvm's curl installer on most
setups) take effect. ~/.bashrc still runs last to preserve behaviour for
users who put PATH logic there without the guard.
Added two tests covering the new behaviour plus an E2E test that spins up
a real LocalEnvironment with a guard-prefixed ~/.bashrc and a ~/.profile
PATH export, and verifies the captured snapshot PATH contains the profile
entry.
When a newly-bundled skill's name collides with a pre-existing user
skill, sync silently kept the user's copy. Users never learned that
a bundled version shipped by that name.
Now (on non-quiet sync only) print:
⚠ <name>: bundled version shipped but you already have a local
skill by this name — yours was kept. Run `hermes skills reset
<name>` to replace it with the bundled version.
No behavior change to manifest writes or to the kept user copy —
purely additive warning on the existing collision-skip path.
When a new bundled skill's name collided with a pre-existing user skill
(from hub, custom, or leftover), sync_skills() recorded the bundled hash
in the manifest even though the on-disk copy was unrelated to bundled.
On the next sync, user_hash != origin_hash (bundled_hash) marked the
skill as "user-modified" permanently, blocking all bundled updates for
that skill until the user ran `hermes skills reset`.
Fix: only baseline the manifest entry when the user's on-disk copy is
byte-identical to bundled (safe to track — this is the reset re-sync or
coincidentally-identical install case). Otherwise skip the manifest
write entirely: the on-disk skill is unrelated to bundled and shouldn't
be tracked as if it were.
This preserves reset_bundled_skill()'s re-baseline flow (its post-delete
sync still writes to the manifest when user copy matches bundled) while
fixing the poisoning scenario for genuinely unrelated collisions.
Adds two tests following the existing test_failed_copy_does_not_poison_manifest
pattern: one verifying the manifest stays clean after a collision with
differing content, one verifying no false user_modified flag on resync.
When _resolve_tirith_path() returns None (e.g. install failed on
unsupported platform or all resolution paths exhausted), the function
passed None directly to subprocess.run(), causing a TypeError instead
of respecting the fail_open config.
Add a None check before the subprocess call that allows or blocks
according to the configured fail_open policy, matching the existing
error handling behavior for OSError and TimeoutExpired.
When override_acp_command was passed to _build_child_agent, it failed to
override effective_provider to 'copilot-acp' and effective_api_mode to
'chat_completions'. This caused the child AIAgent to inherit the parent's
native API configuration (e.g. Anthropic) and attempt real HTTP requests
using the parent's API key, leading to HTTP 401 errors and completely
bypassing the ACP subprocess.
Ensure that if an ACP command override is provided, the child agent
correctly routes through CopilotACPClient.
Refs #2653
Adds MiniMax-AI/cli to the default taps list so the mmx-cli skill
is discoverable and installable out of the box via /skills browse
and /skills install. The skill definition lives upstream at
github.com/MiniMax-AI/cli/skill/SKILL.md, keeping updates decoupled.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Replace hardcoded 'fr' default with DEFAULT_LOCAL_STT_LANGUAGE ('en')
— removes locale leak, matches other providers
- Drop redundant default=True on is_truthy_value (dict .get already defaults)
- Update auto-detect comment to include 'xai' in the chain
- Fix docstring: 21 languages (match PR body + actual xAI API)
- Update test_sends_language_and_format to set HERMES_LOCAL_STT_LANGUAGE=fr
explicitly, since default is no longer 'fr'
All 18 xAI STT tests pass locally.
In _detect_target(), platform.system() returns "Android" on Termux,
not "Linux". Without this change tirith's auto-installer skips
Android even though the Linux GNU binaries are ABI-compatible.
When an MCP server config has ssl_verify: false (e.g. local dev with
a self-signed cert), the setting was read from config.yaml but never
passed to the httpx client, causing CERTIFICATE_VERIFY_FAILED errors
and silent connection failures.
Fix: read ssl_verify from config and pass it as the 'verify' kwarg to
both code paths:
- New API (mcp >= 1.24.0): httpx.AsyncClient(verify=ssl_verify)
- Legacy API (mcp < 1.24.0): streamablehttp_client(..., verify=ssl_verify)
Fixes local dev setups using ServBay, LocalWP, MAMP, or any stack with
a self-signed TLS certificate.
On Windows, Path.open() defaults to the system ANSI code page (cp1252).
If the .env file contains UTF-8 characters, decoding fails with
'gbk codec can't decode byte 0x94'. Specify encoding='utf-8'
explicitly to ensure consistent behavior across platforms.
The Docker terminal backend runs containers with `--cap-drop ALL`
and re-adds only DAC_OVERRIDE, CHOWN, FOWNER. Since commit fee0e0d3
("run as non-root user, use virtualenv") the image entrypoint drops
from root to the `hermes` user via `gosu`, which requires CAP_SETUID
and CAP_SETGID. Without them every sandbox container exits
immediately with:
Dropping root privileges
error: failed switching to 'hermes': operation not permitted
Breaking every terminal/file tool invocation in `terminal.backend: docker`
mode.
Fix: add SETUID and SETGID to the cap-add list. The `no-new-privileges`
security-opt is kept, so gosu still cannot escalate back to root after
the one-way drop — the hardening posture is preserved.
Reproduction
------------
With any image whose ENTRYPOINT calls `gosu <user>`, the container
exits immediately under the pre-fix cap set. Post-fix, the drop
succeeds and the container proceeds normally.
docker run --rm \
--cap-drop ALL \
--cap-add DAC_OVERRIDE --cap-add CHOWN --cap-add FOWNER \
--security-opt no-new-privileges \
--entrypoint /usr/local/bin/gosu \
hermes-claude:latest hermes id
# -> error: failed switching to 'hermes': operation not permitted
# Same command with SETUID+SETGID added:
# -> uid=10000(hermes) gid=10000(hermes) groups=10000(hermes)
Tests
-----
Added `test_security_args_include_setuid_setgid_for_gosu_drop` that
asserts both caps are present and the overall hardening posture
(cap-drop ALL + no-new-privileges) is preserved.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
LocalEnvironment._run_bash() spawned subprocess.Popen without a cwd
argument, so init_session()'s pwd -P ran in the gateway process's
startup directory and overwrote self.cwd. Pass cwd=self.cwd so the
initial snapshot captures the user-configured working directory.
Tested:
- pytest tests/ -q (255 env-related tests passed)
- Full suite: 13,537 passed; 70 pre-existing failures unrelated to local env
The container_config builder in terminal_tool.py was missing
docker_forward_env and docker_env keys, causing config.yaml's
docker_forward_env setting to be silently ignored. Environment
variables listed in docker_forward_env were never injected into
Docker containers.
This fix adds both keys to the container_config dict so they are
properly passed to _create_environment().
Follow-up on helix4u's PR #14211:
- Flip default to true: narrowing toolsets=['web','browser'] expresses
'I want these extras', not 'silently strip MCP'. Parent MCP tools
(registered at runtime) should survive narrowing by default.
- Drop _config_version bump (22->23); additive nested key under
delegation.* is handled by _deep_merge, no migration needed.
- Update tests to reflect new default behavior.
browser_cdp_tool.py registers before browser_tool.py (alphabetical
import order), so its stricter check_fn (requires CDP endpoint) becomes
the toolset-level check for all 11 browser tools. This causes
'hermes doctor' to report the entire browser toolset as unavailable
even when agent-browser is correctly installed.
Move browser_cdp to toolset='browser-cdp' so it is evaluated
independently. browser_navigate et al. only need agent-browser;
browser_cdp additionally requires a reachable CDP endpoint.
Version managers like frum (Ruby), rvm, nvm, and others commonly alias
cd to a wrapper function that runs additional logic after directory
changes. When Hermes captures the shell environment into a session
snapshot, these aliases are preserved. If the wrapper function fails
in the subprocess context (e.g. frum not on PATH), every cd fails,
causing all terminal commands to exit with code 126.
Using builtin cd bypasses any aliases or functions, ensuring the
directory change always uses the real bash builtin regardless of
what version managers are installed.
Make Tavily client respect a TAVILY_BASE_URL environment variable,
defaulting to https://api.tavily.com for backward compatibility.
Consistent with FIRECRAWL_API_URL pattern already used in this module.
The code execution sandbox creates a Unix domain socket in /tmp with
default permissions, allowing any local user to connect and execute
tool calls. Restrict to 0o600 after bind.
Closes#6230
Upgrades agent-browser from 0.13.0 to 0.26.0, picking up 13 releases of
daemon reliability fixes:
- Daemon hang on Linux from waitpid(-1) race in SIGCHLD handler (#1098)
- Chrome killed after ~10s idle due to PR_SET_PDEATHSIG thread tracking (#1157)
- Orphaned Chrome processes via process-group kill on shutdown (#1137)
- Stale daemon after upgrade via .version sidecar and auto-restart (#1134)
- Idle timeout not firing (sleep future recreated each loop) (#1110)
- Navigation hanging on lifecycle events that never fire (#1059, #1092)
- CDP attach hang on Chrome 144+ (#1133)
- Windows daemon TCP bind with Hyper-V port conflicts (#1041)
- Shadow DOM traversal in accessibility tree snapshots
- doctor command for user self-diagnosis
Also wires AGENT_BROWSER_IDLE_TIMEOUT_MS into the browser subprocess
environment so the daemon self-terminates after our configured inactivity
timeout (default 300s). This is the daemon-side counterpart to the
Python-side inactivity reaper — the daemon kills itself and its Chrome
children when no commands arrive, preventing orphan accumulation even
when the Python process dies without running atexit handlers.
Addresses #7343 (daemon socket hangs, shadow DOM) and #13793 (orphan
accumulation from force-killed sessions).
Adds security.allow_private_urls / HERMES_ALLOW_PRIVATE_URLS toggle so
users on OpenWrt routers, TUN-mode proxies (Clash/Mihomo/Sing-box),
corporate split-tunnel VPNs, and Tailscale networks — where DNS resolves
public domains to 198.18.0.0/15 or 100.64.0.0/10 — can use web_extract,
browser, vision URL fetching, and gateway media downloads.
Single toggle in tools/url_safety.py; all 23 is_safe_url() call sites
inherit automatically. Cached for process lifetime.
Cloud metadata endpoints stay ALWAYS blocked regardless of the toggle:
169.254.169.254 (AWS/GCP/Azure/DO/Oracle), 169.254.170.2 (AWS ECS task
IAM creds), 169.254.169.253 (Azure IMDS wire server), 100.100.100.200
(Alibaba), fd00:ec2::254 (AWS IPv6), the entire 169.254.0.0/16
link-local range, and the metadata.google.internal / metadata.goog
hostnames (checked pre-DNS so they can't be bypassed on networks where
those names resolve to local IPs).
Supersedes #3779 (narrower HERMES_ALLOW_RFC2544 for the same class of
users).
Co-authored-by: kshitijk4poor <82637225+kshitijk4poor@users.noreply.github.com>
- delegate_task: use shared tool_error() for the paused-spawn early return
so the error envelope matches the rest of the tool.
- Disk snapshot label: treat orphaned nodes (parentId missing from the
snapshot) as top-level, matching buildSubagentTree / summarizeLabel.
Four real issues Copilot flagged:
1. delegate_tool: `_build_child_agent` never passed `toolsets` to the
progress callback, so the event payload's `toolsets` field (wired
through every layer) was always empty and the overlay's toolsets
row never populated. Thread `child_toolsets` through.
2. event handler: the race-protection on subagent.spawn_requested /
subagent.start only preserved `completed`, so a late-arriving queued
event could clobber `failed` / `interrupted` too. Preserve any
terminal status (`completed | failed | interrupted`).
3. SpawnHud: comment claimed concurrency was approximated by "widest
level in the tree" but code used `totals.activeCount` (total across
all parents). `max_concurrent_children` is a per-parent cap, so
activeCount over-warns for multi-orchestrator runs. Switch to
`max(widthByDepth(tree))`; the label now reads `⚡W/cap+extra` where
W is the widest level (drives the ratio) and `+extra` is the rest.
4. spawn_tree.list: comment said "peek header without parsing full list"
but the code json.loads()'d every snapshot. Adds a per-session
`_index.jsonl` sidecar written on save; list() reads only the index
(with a full-scan fallback for pre-index sessions). O(1) per
snapshot now vs O(file-size).
Adds a live + post-hoc audit surface for recursive delegate_task fan-out.
None of cc/oc/oclaw tackle nested subagent trees inside an Ink overlay;
this ships a view-switched dashboard that handles arbitrary depth + width.
Python
- delegate_tool: every subagent event now carries subagent_id, parent_id,
depth, model, tool_count; subagent.complete also ships input/output/
reasoning tokens, cost, api_calls, files_read/files_written, and a
tail of tool-call outputs
- delegate_tool: new subagent.spawn_requested event + _active_subagents
registry so the overlay can kill a branch by id and pause new spawns
- tui_gateway: new RPCs delegation.status, delegation.pause,
subagent.interrupt, spawn_tree.save/list/load (disk under
\$HERMES_HOME/spawn-trees/<session>/<ts>.json)
TUI
- /agents overlay: full-width list mode (gantt strip + row picker) and
Enter-to-drill full-width scrollable detail mode; inverse+amber
selection, heat-coloured branch markers, wall-clock gantt with tick
ruler, per-branch rollups
- Detail pane: collapsible accordions (Budget, Files, Tool calls, Output,
Progress, Summary); open-state persists across agents + mode switches
via a shared atom
- /replay [N|last|list|load <path>] for in-memory + disk history;
/replay-diff <a> <b> for side-by-side tree comparison
- Status-bar SpawnHud warns as depth/concurrency approaches caps;
overlay auto-follows the just-finished turn onto history[1]
- Theme: bump DARK dim #B8860B → #CC9B1F for readable secondary text
globally; keep LIGHT untouched
Tests: +29 new subagentTree unit tests; 215/215 passing.
* feat(plugins): pluggable image_gen backends + OpenAI provider
Adds a ImageGenProvider ABC so image generation backends register as
bundled plugins under `plugins/image_gen/<name>/`. The plugin scanner
gains three primitives to make this work generically:
- `kind:` manifest field (`standalone` | `backend` | `exclusive`).
Bundled `kind: backend` plugins auto-load — no `plugins.enabled`
incantation. User-installed backends stay opt-in.
- Path-derived keys: `plugins/image_gen/openai/` gets key
`image_gen/openai`, so a future `tts/openai` cannot collide.
- Depth-2 recursion into category namespaces (parent dirs without a
`plugin.yaml` of their own).
Includes `OpenAIImageGenProvider` as the first consumer (gpt-image-1.5
default, plus gpt-image-1, gpt-image-1-mini, DALL-E 3/2). Base64
responses save to `$HERMES_HOME/cache/images/`; URL responses pass
through.
FAL stays in-tree for this PR — a follow-up ports it into
`plugins/image_gen/fal/` so the in-tree `image_generation_tool.py`
slims down. The dispatch shim in `_handle_image_generate` only fires
when `image_gen.provider` is explicitly set to a non-FAL value, so
existing FAL setups are untouched.
- 41 unit tests (scanner recursion, kind parsing, gate logic,
registry, OpenAI payload shapes)
- E2E smoke verified: bundled plugin autoloads, registers, and
`_handle_image_generate` routes to OpenAI when configured
* fix(image_gen/openai): don't send response_format to gpt-image-*
The live API rejects it: 'Unknown parameter: response_format'
(verified 2026-04-21 with gpt-image-1.5). gpt-image-* models return
b64_json unconditionally, so the parameter was both unnecessary and
actively broken.
* feat(image_gen/openai): gpt-image-2 only, drop legacy catalog
gpt-image-2 is the latest/best OpenAI image model (released 2026-04-21)
and there's no reason to expose the older gpt-image-1.5 / gpt-image-1 /
dall-e-3 / dall-e-2 alongside it — slower, lower quality, or awkward
(dall-e-2 squares only). Trim the catalog down to a single model.
Live-verified end-to-end: landscape 1536x1024 render of a Moog-style
synth matches prompt exactly, 2.4MB PNG saved to cache.
* feat(image_gen/openai): expose gpt-image-2 as three quality tiers
Users pick speed/fidelity via the normal model picker instead of a
hidden quality knob. All three tier IDs resolve to the single underlying
gpt-image-2 API model with a different quality parameter:
gpt-image-2-low ~15s fast iteration
gpt-image-2-medium ~40s default
gpt-image-2-high ~2min highest fidelity
Live-measured on OpenAI's API today: 15.4s / 40.8s / 116.9s for the
same 1024x1024 prompt.
Config:
image_gen.openai.model: gpt-image-2-high
# or
image_gen.model: gpt-image-2-low
# or env var for scripts/tests
OPENAI_IMAGE_MODEL=gpt-image-2-medium
Live-verified end-to-end with the low tier: 18.8s landscape render of a
golden retriever in wildflowers, vision-confirmed exact match.
* feat(tools_config): plugin image_gen providers inject themselves into picker
'hermes tools' → Image Generation now shows plugin-registered backends
alongside Nous Subscription and FAL.ai without tools_config.py needing
to know about them. OpenAI appears as a third option today; future
backends appear automatically as they're added.
Mechanism:
- ImageGenProvider gains an optional get_setup_schema() hook
(name, badge, tag, env_vars). Default derived from display_name.
- tools_config._plugin_image_gen_providers() pulls the schemas from
every registered non-FAL plugin provider.
- _visible_providers() appends those rows when rendering the Image
Generation category.
- _configure_provider() handles the new image_gen_plugin_name marker:
writes image_gen.provider and routes to the plugin's list_models()
catalog for the model picker.
- _toolset_needs_configuration_prompt('image_gen') stops demanding a
FAL key when any plugin provider reports is_available().
FAL is skipped in the plugin path because it already has hardcoded
TOOL_CATEGORIES rows — when it gets ported to a plugin in a follow-up
PR the hardcoded rows go away and it surfaces through the same path
as OpenAI.
Verified live: picker shows Nous Subscription / FAL.ai / OpenAI.
Picking OpenAI prompts for OPENAI_API_KEY, then shows the
gpt-image-2-low/medium/high model picker sourced from the plugin.
397 tests pass across plugins/, tools_config, registry, and picker.
* fix(image_gen): close final gaps for plugin-backend parity with FAL
Two small places that still hardcoded FAL:
- hermes_cli/setup.py status line: an OpenAI-only setup showed
'Image Generation: missing FAL_KEY'. Now probes plugin providers
and reports '(OpenAI)' when one is_available() — or falls back to
'missing FAL_KEY or OPENAI_API_KEY' if nothing is configured.
- image_generate tool schema description: said 'using FAL.ai, default
FLUX 2 Klein 9B'. Rewrote provider-neutral — 'backend and model are
user-configured' — and notes the 'image' field can be a URL or an
absolute path, which the gateway delivers either way via
extract_local_files().
- Wrap child.run_conversation() in a ThreadPoolExecutor with configurable
timeout (delegation.child_timeout_seconds, default 300s) to prevent
indefinite blocking when a subagent's API call or tool HTTP request hangs.
- Add heartbeat stale detection: if a child's api_call_count doesn't
advance for 5 consecutive heartbeat cycles (~2.5 min), stop touching
the parent's activity timestamp so the gateway inactivity timeout
can fire as a last resort.
- Add 'timeout' as a new exit_reason/status alongside the existing
completed/max_iterations/interrupted states.
- Use shutdown(wait=False) on the timeout executor to avoid the
ThreadPoolExecutor.__exit__ deadlock when a child is stuck on
blocking I/O.
Closes#13768
A single global MAX_TEXT_LENGTH = 4000 truncated every TTS provider at
4000 chars, causing long inputs to be silently chopped even though the
underlying APIs allow much more:
- OpenAI: 4096
- xAI: 15000
- MiniMax: 10000
- ElevenLabs: 5000 / 10000 / 30000 / 40000 (model-aware)
- Gemini: ~5000
- Edge: ~5000
The schema description also told the model 'Keep under 4000 characters',
which encouraged the agent to self-chunk long briefs into multiple TTS
calls (producing 3 separate audio files instead of one).
New behavior:
- PROVIDER_MAX_TEXT_LENGTH table + ELEVENLABS_MODEL_MAX_TEXT_LENGTH
encode the documented per-provider limits.
- _resolve_max_text_length(provider, cfg) resolves:
1. tts.<provider>.max_text_length user override
2. ElevenLabs model_id lookup
3. provider default
4. 4000 fallback
- text_to_speech_tool() and stream_tts_to_speaker() both call the
resolver; old MAX_TEXT_LENGTH alias kept for back-compat.
- Schema description no longer hardcodes 4000.
Tests: 27 new unit + E2E tests; all 53 existing TTS tests and 253
voice-command/voice-cli tests still pass.
* feat(models): hide OpenRouter models that don't advertise tool support
Port from Kilo-Org/kilocode#9068.
hermes-agent is tool-calling-first — every provider path assumes the
model can invoke tools. Models whose OpenRouter supported_parameters
doesn't include 'tools' (e.g. image-only or completion-only models)
cannot be driven by the agent loop and fail at the first tool call.
Filter them out of fetch_openrouter_models() so they never appear in
the model picker (`hermes model`, setup wizard, /model slash command).
Permissive when the field is missing — OpenRouter-compatible gateways
(Nous Portal, private mirrors, older snapshots) don't always populate
supported_parameters. Treat missing as 'unknown → allow' rather than
silently emptying the picker on those gateways. Only hide models
whose supported_parameters is an explicit list that omits tools.
Tests cover: tools present → kept, tools absent → dropped, field
missing → kept, malformed non-list → kept, non-dict item → kept,
empty list → dropped.
* feat(delegate): cross-agent file state coordination for concurrent subagents
Prevents mangled edits when concurrent subagents touch the same file
(same process, same filesystem — the mangle scenario from #11215).
Three layers, all opt-out via HERMES_DISABLE_FILE_STATE_GUARD=1:
1. FileStateRegistry (tools/file_state.py) — process-wide singleton
tracking per-agent read stamps and the last writer globally.
check_stale() names the sibling subagent in the warning when a
non-owning agent wrote after this agent's last read.
2. Per-path threading.Lock wrapped around the read-modify-write
region in write_file_tool and patch_tool. Concurrent siblings on
the same path serialize; different paths stay fully parallel.
V4A multi-file patches lock in sorted path order (deadlock-free).
3. Delegate-completion reminder in tools/delegate_tool.py: after a
subagent returns, writes_since(parent, child_start, parent_reads)
appends '[NOTE: subagent modified files the parent previously
read — re-read before editing: ...]' to entry.summary when the
child touched anything the parent had already seen.
Complements (does not replace) the existing path-overlap check in
run_agent._should_parallelize_tool_batch — batch check prevents
same-file parallel dispatch within one agent's turn (cheap prevention,
zero API cost), registry catches cross-subagent and cross-turn
staleness at write time (detection).
Behavior is warning-only, not hard-failing — matches existing project
style. Errors surface naturally: sibling writes often invalidate the
old_string in patch operations, which already errors cleanly.
Tests: tests/tools/test_file_state_registry.py — 16 tests covering
registry state transitions, per-path locking, per-path-not-global
locking, writes_since filtering, kill switch, and end-to-end
integration through the real read_file/write_file/patch handlers.
Adds role='leaf'|'orchestrator' to delegate_task. With max_spawn_depth>=2,
an orchestrator child retains the 'delegation' toolset and can spawn its
own workers; leaf children cannot delegate further (identical to today).
Default posture is flat — max_spawn_depth=1 means a depth-0 parent's
children land at the depth-1 floor and orchestrator role silently
degrades to leaf. Users opt into nested delegation by raising
max_spawn_depth to 2 or 3 in config.yaml.
Also threads acp_command/acp_args through the main agent loop's delegate
dispatch (previously silently dropped in the schema) via a new
_dispatch_delegate_task helper, and adds a DelegateEvent enum with
legacy-string back-compat for gateway/ACP/CLI progress consumers.
Config (hermes_cli/config.py defaults):
delegation.max_concurrent_children: 3 # floor-only, no upper cap
delegation.max_spawn_depth: 1 # 1=flat (default), 2-3 unlock nested
delegation.orchestrator_enabled: true # global kill switch
Salvaged from @pefontana's PR #11215. Overrides vs. the original PR:
concurrency stays at 3 (PR bumped to 5 + cap 8 — we keep the floor only,
no hard ceiling); max_spawn_depth defaults to 1 (PR defaulted to 2 which
silently enabled one level of orchestration for every user).
Co-authored-by: pefontana <fontana.pedro93@gmail.com>
Adds OpenAI's new GPT Image 2 model via FAL.ai, selectable through
`hermes tools` → Image Generation. SOTA text rendering (including CJK)
and world-aware photorealism.
- FAL_MODELS entry with image_size_preset style
- 4:3 presets on all aspect ratios — 16:9 (1024x576) falls below
GPT-Image-2's 655,360 min-pixel floor and would be rejected
- quality pinned to medium (same rule as gpt-image-1.5) for
predictable Nous Portal billing
- BYOK (openai_api_key) deliberately omitted from supports so all
users stay on shared FAL billing
- 6 new tests covering preset mapping, quality pinning, and
supports-whitelist integrity
- Docs table + aspect-ratio map updated
Live-tested end-to-end: 39.9s cold request, clean 1024x768 PNG
Two related ACP approval issues:
GHSA-96vc-wcxf-jjff — ACP's _run_agent never set HERMES_INTERACTIVE
(or any other flag recognized by tools.approval), so check_all_command_guards
took the non-interactive auto-approve path and never consulted the
ACP-supplied approval callback (conn.request_permission). Dangerous
commands executed in ACP sessions without operator approval despite
the callback being installed. Fix: set HERMES_INTERACTIVE=1 around
the agent run so check_all_command_guards routes through
prompt_dangerous_approval(approval_callback=...) — the correct shape
for ACP's per-session request_permission call. HERMES_EXEC_ASK would
have routed through the gateway-queue path instead, which requires a
notify_cb registered in _gateway_notify_cbs (not applicable to ACP).
GHSA-qg5c-hvr5-hjgr — _approval_callback and _sudo_password_callback
were module-level globals in terminal_tool. Concurrent ACP sessions
running in ThreadPoolExecutor threads each installed their own callback
into the same slot, racing. Fix: store both callbacks in threading.local()
so each thread has its own slot. CLI mode (single thread) is unaffected;
gateway mode uses a separate queue-based approval path and was never
touched.
set_approval_callback is now called INSIDE _run_agent (the executor
thread) rather than before dispatching — so the TLS write lands on the
correct thread.
Tests: 5 new in tests/acp/test_approval_isolation.py covering
thread-local isolation of both callbacks and the HERMES_INTERACTIVE
callback routing. Existing tests/acp/ (159 tests) and tests/tools/
approval-related tests continue to pass.
Fixes GHSA-96vc-wcxf-jjff
Fixes GHSA-qg5c-hvr5-hjgr
A skill declaring `required_environment_variables: [ANTHROPIC_TOKEN]` in
its SKILL.md frontmatter silently bypassed the `execute_code` sandbox's
credential-scrubbing guarantee. `register_env_passthrough` had no
blocklist, so any name a skill chose flipped `is_env_passthrough(name) =>
True`, which shortcircuits the sandbox's secret filter.
Fix: reject registration when the name appears in
`_HERMES_PROVIDER_ENV_BLOCKLIST` (the canonical list of Hermes-managed
credentials — provider keys, gateway tokens, etc.). Log a warning naming
GHSA-rhgp-j443-p4rf so operators see the rejection in logs.
Non-Hermes third-party API keys (TENOR_API_KEY for gif-search,
NOTION_TOKEN for notion skills, etc.) remain legitimately registerable —
they were never in the sandbox scrub list in the first place.
Tests: 16 -> 17 passing. Two old tests that documented the bypass
(`test_passthrough_allows_blocklisted_var`, `test_make_run_env_passthrough`)
are rewritten to assert the new fail-closed behavior. New
`test_non_hermes_api_key_still_registerable` locks in that legitimate
third-party keys are unaffected.
Reported in GHSA-rhgp-j443-p4rf by @q1uf3ng. Hardening; not CVE-worthy
on its own per the decision matrix (attacker must already have operator
consent to install a malicious skill).
Fixes#13027
Previously, `_is_skill_disabled()` only checked the explicit `platform`
argument and `os.getenv('HERMES_PLATFORM')`, missing the gateway session
context (`HERMES_SESSION_PLATFORM`). This caused `skill_view()` to expose
skills that were platform-disabled for the active gateway session.
Add `_get_session_platform()` helper that resolves the platform from
`gateway.session_context.get_session_env`, mirroring the logic in
`agent.skill_utils.get_disabled_skill_names()`.
Now the platform resolution follows the same precedence as skill_utils:
1. Explicit `platform` argument
2. `HERMES_PLATFORM` environment variable
3. `HERMES_SESSION_PLATFORM` from gateway session context
Previously the breaker was only cleared when the post-reconnect retry
call itself succeeded (via _reset_server_error at the end of the try
block). If OAuth recovery succeeded but the retry call happened to
fail for a different reason, control fell through to the
needs_reauth path which called _bump_server_error — adding to an
already-tripped count instead of the fresh count the reconnect
justified. With fix#1 in place this would still self-heal on the
next cooldown, but we should not pay a 60s stall when we already
have positive evidence the server is viable.
Move _reset_server_error(server_name) up to immediately after the
reconnect-and-ready-wait block, before the retry_call. The
subsequent retry still goes through _bump_server_error on failure,
so a genuinely broken server re-trips the breaker as normal — but
the retry starts from a clean count (1 after a failure), not a
stale one.
The MCP circuit breaker previously had no path back to the closed
state: once _server_error_counts[srv] reached _CIRCUIT_BREAKER_THRESHOLD
the gate short-circuited every subsequent call, so the only reset
path (on successful call) was unreachable. A single transient
3-failure blip (bad network, server restart, expired token) permanently
disabled every tool on that MCP server for the rest of the agent
session.
Introduce a classic closed/open/half-open state machine:
- Track a per-server breaker-open timestamp in _server_breaker_opened_at
alongside the existing failure count.
- Add _CIRCUIT_BREAKER_COOLDOWN_SEC (60s). Once the count reaches
threshold, calls short-circuit for the cooldown window.
- After the cooldown elapses, the *next* call falls through as a
half-open probe that actually hits the session. Success resets the
breaker via _reset_server_error; failure re-bumps the count via
_bump_server_error, which re-stamps the open timestamp and re-arms
the cooldown.
The error message now includes the live failure count and an
"Auto-retry available in ~Ns" hint so the model knows the breaker
will self-heal rather than giving up on the tool for the whole
session.
Covers tests 1 (half-opens after cooldown) and 2 (reopens on probe
failure); test 3 (cleared on reconnect) still fails pending fix#2.
Follow-up to PR #2504. The original fix covered the two direct FAL_KEY
checks in image_generation_tool but left four other call sites intact,
including the managed-gateway gate where a whitespace-only FAL_KEY
falsely claimed 'user has direct FAL' and *skipped* the Nous managed
gateway fallback entirely.
Introduce fal_key_is_configured() in tools/tool_backend_helpers.py as a
single source of truth (consults os.environ, falls back to .env for
CLI-setup paths) and route every FAL_KEY presence check through it:
- tools/image_generation_tool.py : _resolve_managed_fal_gateway,
image_generate_tool's upfront check, check_fal_api_key
- hermes_cli/nous_subscription.py : direct_fal detection, selected
toolset gating, tools_ready map
- hermes_cli/tools_config.py : image_gen needs-setup check
Verified by extending tests/tools/test_image_generation_env.py and by
E2E exercising whitespace + managed-gateway composition directly.
Treat whitespace-only FAL_KEY the same as unset so users who export
FAL_KEY=" " (or CI that leaves a blank token) get the expected
'not set' error path instead of a confusing downstream fal_client
failure.
Applied to the two direct FAL_KEY checks in image_generation_tool.py:
image_generate_tool's upfront credential check and check_fal_api_key().
Both keep the existing managed-gateway fallback intact.
Adapted the original whitespace/valid tests to pin the managed gateway
to None so the whitespace assertion exercises the direct-key path
rather than silently relying on gateway absence.
Follow-ups on top of @teyrebaz33's cherry-picked commit:
1. New shared helper format_no_match_hint() in fuzzy_match.py with a
startswith('Could not find') gate so the snippet only appends to
genuine no-match errors — not to 'Found N matches' (ambiguous),
'Escape-drift detected', or 'identical strings' errors, which would
all mislead the model.
2. file_tools.patch_tool suppresses the legacy generic '[Hint: old_string
not found...]' string when the rich 'Did you mean?' snippet is
already attached — no more double-hint.
3. Wire the same helper into patch_parser.py (V4A patch mode, both
_validate_operations and _apply_update) and skill_manager_tool.py so
all three fuzzy callers surface the hint consistently.
Tests: 7 new gating tests in TestFormatNoMatchHint cover every error
class (ambiguous, drift, identical, non-zero match count, None error,
no similar content, happy path). 34/34 test_fuzzy_match, 96/96
test_file_tools + test_patch_parser + test_skill_manager_tool pass.
E2E verified across all four scenarios: no-match-with-similar,
no-match-no-similar, ambiguous, success. V4A mode confirmed
end-to-end with a non-matching hunk.
When patch_replace() cannot find old_string in a file, the error message
now includes the closest matching lines from the file with line numbers
and context. This helps the LLM self-correct without a separate read_file
call.
Implements Phase 1 of #536: enhanced patch error feedback with no
architectural changes.
- tools/fuzzy_match.py: new find_closest_lines() using SequenceMatcher
- tools/file_operations.py: attach closest-lines hint to patch errors
- tests/tools/test_fuzzy_match.py: 5 new tests for find_closest_lines
Builds on @AxDSan's PR #2109 to finish the KittenTTS wiring so the
provider behaves like every other TTS backend end to end.
- tools/tts_tool.py: `_check_kittentts_available()` helper and wire
into `check_tts_requirements()`; extend Opus-conversion list to
include kittentts (WAV → Opus for Telegram voice bubbles); point the
missing-package error at `hermes setup tts`.
- hermes_cli/tools_config.py: add KittenTTS entry to the "Text-to-Speech"
toolset picker, with a `kittentts` post_setup hook that auto-installs
the wheel + soundfile via pip.
- hermes_cli/setup.py: `_install_kittentts_deps()`, new choice + install
flow in `_setup_tts_provider()`, provider_labels entry, and status row
in the `hermes setup` summary.
- website/docs/user-guide/features/tts.md: add KittenTTS to the provider
table, config example, ffmpeg note, and the zero-config voice-bubble tip.
- tests/tools/test_tts_kittentts.py: 10 unit tests covering generation,
model caching, config passthrough, ffmpeg conversion, availability
detection, and the missing-package dispatcher branch.
E2E verified against the real `kittentts` wheel:
- WAV direct output (pcm_s16le, 24kHz mono)
- MP3 conversion via ffmpeg (from WAV)
- Telegram flow (provider in Opus-conversion list) produces
`codec_name=opus`, 48kHz mono, `voice_compatible=True`, and the
`[[audio_as_voice]]` marker
- check_tts_requirements() returns True when kittentts is installed
Add support for KittenTTS - a lightweight, local TTS engine with models
ranging from 25-80MB that runs on CPU without requiring a GPU or API key.
Features:
- Support for 8 built-in voices (Jasper, Bella, Luna, etc.)
- Configurable model size (nano 25MB, micro 41MB, mini 80MB)
- Adjustable speech speed
- Model caching for performance
- Automatic WAV to Opus conversion for Telegram voice messages
Configuration example (config.yaml):
tts:
provider: kittentts
kittentts:
model: KittenML/kitten-tts-nano-0.8-int8
voice: Jasper
speed: 1.0
clean_text: true
Installation:
pip install https://github.com/KittenML/KittenTTS/releases/download/0.8.1/kittentts-0.8.1-py3-none-any.whl
Full AST-based scan of all .py files to find every case where a module
or name is imported locally inside a function body but is already
available at module level. This is the second pass — the first commit
handled the known cases from the lint report; this one catches
everything else.
Files changed (19):
cli.py — 16 removals: time as _time/_t/_tmod (×10),
re / re as _re (×2), os as _os, sys,
partial os from combo import,
from model_tools import get_tool_definitions
gateway/run.py — 8 removals: MessageEvent as _ME /
MessageType as _MT (×3), os as _os2,
MessageEvent+MessageType (×2), Platform,
BasePlatformAdapter as _BaseAdapter
run_agent.py — 6 removals: get_hermes_home as _ghh,
partial (contextlib, os as _os),
cleanup_vm, cleanup_browser,
set_interrupt as _sif (×2),
partial get_toolset_for_tool
hermes_cli/main.py — 4 removals: get_hermes_home, time as _time,
logging as _log, shutil
hermes_cli/config.py — 1 removal: get_hermes_home as _ghome
hermes_cli/runtime_provider.py
— 1 removal: load_config as _load_bedrock_config
hermes_cli/setup.py — 2 removals: importlib.util (×2)
hermes_cli/nous_subscription.py
— 1 removal: from hermes_cli.config import load_config
hermes_cli/tools_config.py
— 1 removal: from hermes_cli.config import load_config, save_config
cron/scheduler.py — 3 removals: concurrent.futures, json as _json,
from hermes_cli.config import load_config
batch_runner.py — 1 removal: list_distributions as get_all_dists
(kept print_distribution_info, not at top level)
tools/send_message_tool.py
— 2 removals: import os (×2)
tools/skills_tool.py — 1 removal: logging as _logging
tools/browser_camofox.py
— 1 removal: from hermes_cli.config import load_config
tools/image_generation_tool.py
— 1 removal: import fal_client
environments/tool_context.py
— 1 removal: concurrent.futures
gateway/platforms/bluebubbles.py
— 1 removal: httpx as _httpx
gateway/platforms/whatsapp.py
— 1 removal: import asyncio
tui_gateway/server.py — 2 removals: from datetime import datetime,
import time
All alias references (_time, _t, _tmod, _re, _os, _os2, _json, _ghh,
_ghome, _sif, _ME, _MT, _BaseAdapter, _load_bedrock_config, _httpx,
_logging, _log, get_all_dists) updated to use the top-level names.
Sweep ~74 redundant local imports across 21 files where the same module
was already imported at the top level. Also includes type fixes and lint
cleanups on the same branch.
* feat(skills): inject absolute skill dir and expand ${HERMES_SKILL_DIR} templates
When a skill loads, the activation message now exposes the absolute
skill directory and substitutes ${HERMES_SKILL_DIR} /
${HERMES_SESSION_ID} tokens in the SKILL.md body, so skills with
bundled scripts can instruct the agent to run them by absolute path
without an extra skill_view round-trip.
Also adds opt-in inline-shell expansion: !`cmd` snippets in SKILL.md
are pre-executed (with the skill directory as CWD) and their stdout is
inlined into the message before the agent reads it. Off by default —
enable via skills.inline_shell in config.yaml — because any snippet
runs on the host without approval.
Changes:
- agent/skill_commands.py: template substitution, inline-shell
expansion, absolute skill-dir header, supporting-files list now
shows both relative and absolute forms.
- hermes_cli/config.py: new skills.template_vars,
skills.inline_shell, skills.inline_shell_timeout knobs.
- tests/agent/test_skill_commands.py: coverage for header, both
template tokens (present and missing session id), template_vars
disable, inline-shell default-off, enabled, CWD, and timeout.
- website/docs/developer-guide/creating-skills.md: documents the
template tokens, the absolute-path header, and the opt-in inline
shell with its security caveat.
Validation: tests/agent/ 1591 passed (includes 9 new tests).
E2E: loaded a real skill in an isolated HERMES_HOME; confirmed
${HERMES_SKILL_DIR} resolves to the absolute path, ${HERMES_SESSION_ID}
resolves to the passed task_id, !`date` runs when opt-in is set, and
stays literal when it isn't.
* feat(terminal): source ~/.bashrc (and user-listed init files) into session snapshot
bash login shells don't source ~/.bashrc, so tools that install themselves
there — nvm, asdf, pyenv, cargo, custom PATH exports — stay invisible to
the environment snapshot Hermes builds once per session. Under systemd
or any context with a minimal parent env, that surfaces as
'node: command not found' in the terminal tool even though the binary
is reachable from every interactive shell on the machine.
Changes:
- tools/environments/local.py: before the login-shell snapshot bootstrap
runs, prepend guarded 'source <file>' lines for each resolved init
file. Missing files are skipped, each source is wrapped with a
'[ -r ... ] && . ... || true' guard so a broken rc can't abort the
bootstrap.
- hermes_cli/config.py: new terminal.shell_init_files (explicit list,
supports ~ and ${VAR}) and terminal.auto_source_bashrc (default on)
knobs. When shell_init_files is set it takes precedence; when it's
empty and auto_source_bashrc is on, ~/.bashrc gets auto-sourced.
- tests/tools/test_local_shell_init.py: 10 tests covering the resolver
(auto-bashrc, missing file, explicit override, ~/${VAR} expansion,
opt-out) and the prelude builder (quoting, guarded sourcing), plus
a real-LocalEnvironment snapshot test that confirms exports in the
init file land in subsequent commands' environment.
- website/docs/reference/faq.md: documents the fix in Troubleshooting,
including the zsh-user pattern of sourcing ~/.zshrc or nvm.sh
directly via shell_init_files.
Validation: 10/10 new tests pass; tests/tools/test_local_*.py 40/40
pass; tests/agent/ 1591/1591 pass; tests/hermes_cli/test_config.py
50/50 pass. E2E in an isolated HERMES_HOME: confirmed that a fake
~/.bashrc setting a marker var and PATH addition shows up in a real
LocalEnvironment().execute() call, that auto_source_bashrc=false
suppresses it, that an explicit shell_init_files entry wins over the
auto default, and that a missing bashrc is silently skipped.
Aslaaen's fix in the original PR covered _detect_api_mode_for_url and the
two openai/xai sites in run_agent.py. This finishes the sweep: the same
substring-match false-positive class (e.g. https://api.openai.com.evil/v1,
https://proxy/api.openai.com/v1, https://api.anthropic.com.example/v1)
existed in eight more call sites, and the hostname helper was duplicated
in two modules.
- utils: add shared base_url_hostname() (single source of truth).
- hermes_cli/runtime_provider, run_agent: drop local duplicates, import
from utils. Reuse the cached AIAgent._base_url_hostname attribute
everywhere it's already populated.
- agent/auxiliary_client: switch codex-wrap auto-detect, max_completion_tokens
gate (auxiliary_max_tokens_param), and custom-endpoint max_tokens kwarg
selection to hostname equality.
- run_agent: native-anthropic check in the Claude-style model branch
and in the AIAgent init provider-auto-detect branch.
- agent/model_metadata: Anthropic /v1/models context-length lookup.
- hermes_cli/providers.determine_api_mode: anthropic / openai URL
heuristics for custom/unknown providers (the /anthropic path-suffix
convention for third-party gateways is preserved).
- tools/delegate_tool: anthropic detection for delegated subagent
runtimes.
- hermes_cli/setup, hermes_cli/tools_config: setup-wizard vision-endpoint
native-OpenAI detection (paired with deduping the repeated check into
a single is_native_openai boolean per branch).
Tests:
- tests/test_base_url_hostname.py covers the helper directly
(path-containing-host, host-suffix, trailing dot, port, case).
- tests/hermes_cli/test_determine_api_mode_hostname.py adds the same
regression class for determine_api_mode, plus a test that the
/anthropic third-party gateway convention still wins.
Also: add asslaenn5@gmail.com → Aslaaen to scripts/release.py AUTHOR_MAP.
Users can declare shell scripts in config.yaml under a hooks: block that
fire on plugin-hook events (pre_tool_call, post_tool_call, pre_llm_call,
subagent_stop, etc). Scripts receive JSON on stdin, can return JSON on
stdout to block tool calls or inject context pre-LLM.
Key design:
- Registers closures on existing PluginManager._hooks dict — zero changes
to invoke_hook() call sites
- subprocess.run(shell=False) via shlex.split — no shell injection
- First-use consent per (event, command) pair, persisted to allowlist JSON
- Bypass via --accept-hooks, HERMES_ACCEPT_HOOKS=1, or hooks_auto_accept
- hermes hooks list/test/revoke/doctor CLI subcommands
- Adds subagent_stop hook event fired after delegate_task children exit
- Claude Code compatible response shapes accepted
Cherry-picked from PR #13143 by @pefontana.
Cherry-picked from PR #13159 by @cdanis.
Adds native media attachment delivery to Signal via signal-cli JSON-RPC
attachments param. Signal messages with media now follow the same
early-return pattern as Telegram/Discord/Matrix — attachments are sent
only with the last chunk to avoid duplicates.
Follow-up fixes on top of the original PR:
- Moved Signal into its own early-return block above the restriction
check (matches Telegram/Discord/Matrix pattern)
- Fixed media_files being sent on every chunk in the generic loop
- Restored restriction/warning guards to simple form (Signal exits early)
- Fixed non-hermetic test writing to /tmp instead of tmp_path
Adds a _resolve_path() helper that reads TERMINAL_CWD and uses it as
the base for relative path resolution. Applied to _check_sensitive_path,
read_file_tool, _update_read_timestamp, and _check_file_staleness.
Absolute paths and non-worktree sessions (no TERMINAL_CWD) are
unaffected — falls back to os.getcwd().
Fixes#12689.
Replaces the serial for-loop in tick() with ThreadPoolExecutor so all
jobs due in a single tick run concurrently. A slow job no longer blocks
others from executing, fixing silent job skipping (issue #9086).
Thread safety:
- Session/delivery env vars migrated from os.environ to ContextVars
(gateway/session_context.py) so parallel jobs can't clobber each
other's delivery targets. Each thread gets its own copied context.
- jobs.json read-modify-write cycles (advance_next_run, mark_job_run)
protected by threading.Lock to prevent concurrent save clobber.
- send_message_tool reads delivery vars via get_session_env() for
ContextVar-aware resolution with os.environ fallback.
Configuration:
- cron.max_parallel_jobs in config.yaml (null = unbounded, 1 = serial)
- HERMES_CRON_MAX_PARALLEL env var override
Based on PR #9169 by @VenomMoth1.
Fixes#9086
Cherry-picked from PR #2545 by @Mibayy.
The setup wizard could leave stt.model: "whisper-1" in config.yaml.
When using the local faster-whisper provider, this crashed with
"Invalid model size 'whisper-1'". Voice messages were silently ignored.
_normalize_local_model() now detects cloud-only names (whisper-1,
gpt-4o-transcribe, etc.) and maps them to the default local model
with a warning. Valid local sizes (tiny, base, small, medium, large-v3)
pass through unchanged.
- Renamed _normalize_local_command_model -> _normalize_local_model
(backward-compat wrapper preserved)
- 6 new tests including integration test
- Added lowercase AUTHOR_MAP alias for @Mibayy
Closes#2544
On macOS, Unix domain socket paths are capped at 104 bytes (sun_path).
SSH appends a 16-byte random suffix to the ControlPath when operating
in ControlMaster mode. With an IPv6 host embedded literally in the
filename and a deeply-nested macOS $TMPDIR like
/var/folders/XX/YYYYYYYYYYYY/T/, the full path reliably exceeds the
limit — every terminal/file-op tool call then fails immediately with
``unix_listener: path "…" too long for Unix domain socket``.
Swap the ``user@host:port.sock`` filename for a sha256-derived 16-char
hex digest. The digest is deterministic for a given (user, host, port)
triple, so ControlMaster reuse across reconnects is preserved, and the
full path fits comfortably under the limit even after SSH's random
suffix. Collision space is 2^64 — effectively unreachable for the
handful of concurrent connections any single Hermes process holds.
Regression tests cover: path length under realistic macOS $TMPDIR with
the IPv6 host from the issue report, determinism for reconnects, and
distinctness across different (user, host, port) triples.
Closes#11840
Follow-up to #12704. The SignalAdapter can resolve +E164 numbers to
UUIDs via listContacts, but _parse_target_ref() in the send_message
tool rejected '+' as non-digit and fell through to channel-name
resolution — which fails for contacts without a prior session entry.
Adds an E.164 branch in _parse_target_ref for phone-based platforms
(signal, sms, whatsapp) that preserves the leading '+' so downstream
adapters keep the format they expect. Non-phone platforms are
unaffected.
Reported by @qdrop17 on Discord after pulling #12704.
file_tools._get_file_ops() built a container_config dict for Docker/
Singularity/Modal/Daytona backends but omitted docker_mount_cwd_to_workspace
and docker_forward_env. Both are read by _create_environment() from
container_config, so file tools (read_file, write_file, patch, search)
silently ignored those config values when running in Docker.
Add the two missing keys to match the container_config already built by
terminal_tool.terminal_tool().
Fixes#2672.
The vision tool hardcoded temperature=0.1, ignoring the user's
config.yaml setting. This broke providers like Kimi/Moonshot that
require temperature=1 for vision models. Now reads temperature
from auxiliary.vision.temperature, falling back to 0.1.
bash parses `A && B &` with `&&` tighter than `&`, so it forks a subshell
for the compound and backgrounds the subshell. Inside the subshell, B
runs foreground, so the subshell waits for B. When B is a process that
doesn't naturally exit (`python3 -m http.server`, `yes > /dev/null`, a
long-running daemon), the subshell is stuck in `wait4` forever and leaks
as an orphan reparented to init.
Observed in production: agents running `cd X && python3 -m http.server
8000 &>/dev/null & sleep 1 && curl ...` as a "start a local server, then
verify it" one-liner. Outer bash exits cleanly; the subshell never does.
Across ~3 days of use, 8 unique stuck-terminal events and 7 leaked
bash+server pairs accumulated on the fleet, with some sessions appearing
hung from the user's perspective because the subshell's open stdout pipe
kept the terminal tool's drain thread blocked.
This is distinct from the `set +m` fix in 933fbd8f (which addressed
interactive-shell job-control waiting at exit). `set +m` doesn't help
here because `bash -c` is non-interactive and job control is already
off; the problem is the subshell's own internal wait for its foreground
B, not the outer shell's job-tracking.
The fix: walk the command shell-aware (respecting quotes, parens, brace
groups, `&>`/`>&` redirects), find `A && B &` / `A || B &` at depth 0
and rewrite the tail to `A && { B & }`. Brace groups don't fork a
subshell — they run in the current shell. `B &` inside the group is a
simple background (no subshell wait). The outer `&` is absorbed into
the group, so the compound no longer needs an explicit subshell.
`&&` error-propagation is preserved exactly: if A fails, `&&`
short-circuits and B never runs.
- Skips quoted strings, comment lines, and `(…)` subshells
- Handles `&>/dev/null`, `2>&1`, `>&2` without mistaking them for `&`
- Resets chain state at `;`, `|`, and newlines
- Tracks brace depth so already-rewritten output is idempotent
- Walks using the existing `_read_shell_token` tokenizer, matching the
pattern of `_rewrite_real_sudo_invocations`
Called once from `BaseEnvironment.execute` right after
`_prepare_command`, so it runs for every backend (local, ssh, docker,
modal, etc.) with no per-backend plumbing.
34 new tests covering rewrite cases, preservation cases, redirect
edge-cases, quoting/parens/backticks, idempotency, and empty/edge
inputs. End-to-end verified on a test VM: the exact vela-incident
command now returns in ~1.3s with no leaked bash, only the intentional
backgrounded server.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* [verified] fix(mcp-oauth): bridge httpx auth_flow bidirectional generator
HermesMCPOAuthProvider.async_auth_flow wrapped the SDK's auth_flow with
'async for item in super().async_auth_flow(request): yield item', which
discards httpx's .asend(response) values and resumes the inner generator
with None. This broke every OAuth MCP server on the first HTTP response
with 'NoneType' object has no attribute 'status_code' crashing at
mcp/client/auth/oauth2.py:505.
Replace with a manual bridge that forwards .asend() values into the
inner generator, preserving httpx's bidirectional auth_flow contract.
Add tests/tools/test_mcp_oauth_bidirectional.py with two regression
tests that drive the flow through real .asend() round-trips. These
catch the bug at the unit level; prior tests only exercised
_initialize() and disk-watching, never the full generator protocol.
Verified against BetterStack MCP:
Before: 'Connection failed (11564ms): NoneType...' after 3 retries
After: 'Connected (2416ms); Tools discovered: 83'
Regression from #11383.
* [verified] fix(mcp-oauth): seed token_expiry_time + pre-flight AS discovery on cold-load
PR #11383's consolidation fixed external-refresh reloading and 401 dedup
but left two latent bugs that surfaced on BetterStack and any other OAuth
MCP with a split-origin authorization server:
1. HermesTokenStorage persisted only a relative 'expires_in', which is
meaningless after a process restart. The MCP SDK's OAuthContext
does NOT seed token_expiry_time in _initialize, so is_token_valid()
returned True for any reloaded token regardless of age. Expired
tokens shipped to servers, and app-level auth failures (e.g.
BetterStack's 'No teams found. Please check your authentication.')
were invisible to the transport-layer 401 handler.
2. Even once preemptive refresh did fire, the SDK's _refresh_token
falls back to {server_url}/token when oauth_metadata isn't cached.
For providers whose AS is at a different origin (BetterStack:
mcp.betterstack.com for MCP, betterstack.com/oauth/token for the
token endpoint), that fallback 404s and drops into full browser
re-auth on every process restart.
Fix set:
- HermesTokenStorage.set_tokens persists an absolute wall-clock
expires_at alongside the SDK's OAuthToken JSON (time.time() + TTL
at write time).
- HermesTokenStorage.get_tokens reconstructs expires_in from
max(expires_at - now, 0), clamping expired tokens to zero TTL.
Legacy files without expires_at fall back to file-mtime as a
best-effort wall-clock proxy, self-healing on the next set_tokens.
- HermesMCPOAuthProvider._initialize calls super(), then
update_token_expiry on the reloaded tokens so token_expiry_time
reflects actual remaining TTL. If tokens are loaded but
oauth_metadata is missing, pre-flight PRM + ASM discovery runs
via httpx.AsyncClient using the MCP SDK's own URL builders and
response handlers (build_protected_resource_metadata_discovery_urls,
handle_auth_metadata_response, etc.) so the SDK sees the correct
token_endpoint before the first refresh attempt. Pre-flight is
skipped when there are no stored tokens to keep fresh-install
paths zero-cost.
Test coverage (tests/tools/test_mcp_oauth_cold_load_expiry.py):
- set_tokens persists absolute expires_at
- set_tokens skips expires_at when token has no expires_in
- get_tokens round-trips expires_at -> remaining expires_in
- expired tokens reload with expires_in=0
- legacy files without expires_at fall back to mtime proxy
- _initialize seeds token_expiry_time from stored tokens
- _initialize flags expired-on-disk tokens as is_token_valid=False
- _initialize pre-flights PRM + ASM discovery with mock transport
- _initialize skips pre-flight when no tokens are stored
Verified against BetterStack MCP:
hermes mcp test betterstack -> Connected (2508ms), 83 tools
mcp_betterstack_telemetry_list_teams_tool -> real team data, not
'No teams found. Please check your authentication.'
Reference: mcp-oauth-token-diagnosis skill, Fix A.
* chore: map hermes@noushq.ai to benbarclay in AUTHOR_MAP
Needed for CI attribution check on cherry-picked commits from PR #12025.
---------
Co-authored-by: Hermes Agent <hermes@noushq.ai>
Two hardening layers in the patch tool, triggered by a real silent failure
in the previous session:
(1) Post-write verification in patch_replace — after write_file succeeds,
re-read the file and confirm the bytes on disk match the intended write.
If not, return an error instead of the current success-with-diff. Catches
silent persistence failures from any cause (backend FS oddities, stdin
pipe truncation, concurrent task races, mount drift).
(2) Escape-drift guard in fuzzy_find_and_replace — when a non-exact
strategy matches and both old_string and new_string contain literal
\' or \" sequences but the matched file region does not, reject the
patch with a clear error pointing at the likely cause (tool-call
serialization adding a spurious backslash around apostrophes/quotes).
Exact matches bypass the guard, and legitimate edits that add or
preserve escape sequences in files that already have them still work.
Why: in a prior tool call, old_string was sent with \' where the file
has ' (tool-call transport drift). The fuzzy matcher's block_anchor
strategy matched anyway and produced a diff the tool reported as
successful — but the file was never modified on disk. The agent moved
on believing the edit landed when it hadn't.
Tests: added TestPatchReplacePostWriteVerification (3 cases) and
TestEscapeDriftGuard (6 cases). All pass, existing fuzzy match and
file_operations tests unaffected.
* feat: add Discord server introspection and management tool
Add a discord_server tool that gives the agent the ability to interact
with Discord servers when running on the Discord gateway. Uses Discord
REST API directly with the bot token — no dependency on the gateway
adapter's discord.py client.
The tool is only included in the hermes-discord toolset (zero cost for
users on other platforms) and gated on DISCORD_BOT_TOKEN via check_fn.
Actions (14):
- Introspection: list_guilds, server_info, list_channels, channel_info,
list_roles, member_info, search_members
- Messages: fetch_messages, list_pins, pin_message, unpin_message
- Management: create_thread, add_role, remove_role
This addresses a gap where users on Discord could not ask Hermes to
review server structure, channels, roles, or members — a task competing
agents (OpenClaw) handle out of the box.
Files changed:
- tools/discord_tool.py (new): Tool implementation + registration
- model_tools.py: Add to discovery list
- toolsets.py: Add to hermes-discord toolset only
- tests/tools/test_discord_tool.py (new): 43 tests covering all actions,
validation, error handling, registration, and toolset scoping
* feat(discord): intent-aware schema filtering + config allowlist + schema cleanup
- _detect_capabilities() hits GET /applications/@me once per process
to read GUILD_MEMBERS / MESSAGE_CONTENT privileged intent bits.
- Schema is rebuilt per-session in model_tools.get_tool_definitions:
hides search_members / member_info when GUILD_MEMBERS intent is off,
annotates fetch_messages description when MESSAGE_CONTENT is off.
- New config key discord.server_actions (comma-separated or YAML list)
lets users restrict which actions the agent can call, intersected
with intent availability. Unknown names are warned and dropped.
- Defense-in-depth: runtime handler re-checks the allowlist so a stale
cached schema cannot bypass a tightened config.
- Schema description rewritten as an action-first manifest (signature
per action) instead of per-parameter 'required for X, Y, Z' cross-refs.
~25% shorter; model can see each action's required params at a glance.
- Added bounds: limit gets minimum=1 maximum=100, auto_archive_duration
becomes an enum of the 4 valid Discord values.
- 403 enrichment: runtime 403 errors are mapped to actionable guidance
(which permission is missing and what to do about it) instead of the
raw Discord error body.
- 36 new tests: capability detection with caching and force refresh,
config allowlist parsing (string/list/invalid/unknown), intent+allowlist
intersection, dynamic schema build, runtime allowlist enforcement,
403 enrichment, and model_tools integration wiring.
The first draft of the fix called `chunk.decode("utf-8")` directly on
each 4096-byte `os.read()` result, which corrupts output whenever a
multi-byte UTF-8 character straddles a read boundary:
* `UnicodeDecodeError` fires on the valid-but-truncated byte sequence.
* The except handler clears ALL previously-decoded output and replaces
the whole buffer with `[binary output detected ...]`.
Empirically: 10000 '日' chars (30001 bytes) through the wrapper loses
all 10000 characters on the first draft; the baseline TextIOWrapper
drain (which uses `encoding='utf-8', errors='replace'` on Popen)
preserves them all. This regression affects any command emitting
non-ASCII output larger than one chunk — CJK/Arabic/emoji in
`npm install`, `pip install`, `docker logs`, `kubectl logs`, etc.
Fix: swap to `codecs.getincrementaldecoder('utf-8')(errors='replace')`,
which buffers partial multi-byte sequences across chunks and substitutes
U+FFFD for genuinely invalid bytes. Flush on drain exit via
`decoder.decode(b'', final=True)` to emit any trailing replacement
character for a dangling partial sequence.
Adds two regression tests:
* test_utf8_multibyte_across_read_boundary — 10000 U+65E5 chars,
verifies count round-trips and no fallback fires.
* test_invalid_utf8_uses_replacement_not_fallback — deliberate
\xff\xfe between valid ASCII, verifies surrounding text survives.
When a user's command backgrounds a child (`cmd &`, `setsid cmd & disown`,
etc.), the backgrounded grandchild inherits the write-end of our stdout
pipe via fork(). The old `for line in proc.stdout` drain never EOF'd
until the grandchild closed the pipe — so for a uvicorn server, the
terminal tool hung indefinitely (users reported the whole session
deadlocking when asking the agent to restart a backend).
Fix: switch _drain() to select()-based non-blocking reads and stop
draining shortly after bash exits even if the pipe hasn't EOF'd. Any
output the grandchild writes after that point goes to an orphaned pipe,
which is exactly what the user asked for when they said '&'.
Adds regression tests covering the issue's exact repro and 5 related
patterns (plain bg, setsid+disown, streaming output, high volume,
timeout, UTF-8).
Agents can now send arbitrary CDP commands to the browser. The tool is
gated on a reachable CDP endpoint at session start — it only appears in
the toolset when BROWSER_CDP_URL is set (from '/browser connect') or
'browser.cdp_url' is configured in config.yaml. Backends that don't
currently expose CDP to the Python side (Camofox, default local
agent-browser, cloud providers whose per-session cdp_url is not yet
surfaced) do not see the tool at all.
Tool schema description links to the CDP method reference at
https://chromedevtools.github.io/devtools-protocol/ so the agent can
web_extract specific method docs on demand.
Stateless per call. Browser-level methods (Target.*, Browser.*,
Storage.*) omit target_id. Page-level methods attach to the target
with flatten=true and dispatch the method on the returned sessionId.
Clean errors when the endpoint becomes unreachable mid-session or
the URL isn't a WebSocket.
Tests: 19 unit (mock CDP server + gate checks) + E2E against real
headless Chrome (Target.getTargets, Browser.getVersion,
Runtime.evaluate with target_id, Page.navigate + re-eval, bogus
method, bogus target_id, missing endpoint) + E2E of the check_fn
gate (tool hidden without CDP URL, visible with it, hidden again
after unset).
Add approvals.cron_mode config option that controls how cron jobs handle
dangerous commands. Previously, cron jobs silently auto-approved all
dangerous commands because there was no user present to approve them.
Now the behavior is configurable:
- deny (default): block dangerous commands and return a message telling
the agent to find an alternative approach. The agent loop continues —
it just can't use that specific command.
- approve: auto-approve all dangerous commands (previous behavior).
When a command is blocked, the agent receives the same response format as
a user denial in the CLI — exit_code=-1, status=blocked, with a message
explaining why and pointing to the config option. This keeps the agent
loop running and encourages it to adapt.
Implementation:
- config.py: add approvals.cron_mode to DEFAULT_CONFIG
- scheduler.py: set HERMES_CRON_SESSION=1 env var before agent runs
- approval.py: both check_command_approval() and check_all_command_guards()
now check for cron sessions and apply the configured mode
- 21 new tests covering config parsing, deny/approve behavior, and
interaction with other bypass mechanisms (yolo, containers)
Stacking both features on the same event produces duplicate, delayed
notifications — delivery is async and continues firing after the process
exits, so matches on end-of-run markers (SUMMARY, DONE, PASS) arrive
after the agent has already polled/waited and moved on.
Updates both the terminal tool JSON schema description and the
terminal_tool() function docstring to make the split explicit:
- watch_patterns: mid-process signals only (errors, readiness markers,
intermediate steps you want to react to before the process exits)
- notify_on_complete: end-of-run completion signal
No behavioural change.
Weaker models (Gemma-class) repeatedly rediscover and forget that
execute_code uses a different CWD and Python interpreter than terminal(),
causing them to flip-flop on whether user files exist and to hit import
errors on project dependencies like pandas.
Adds a new 'code_execution.mode' config key (default 'project') that
brings execute_code into line with terminal()'s filesystem/interpreter:
project (new default):
- cwd = session's TERMINAL_CWD (falls back to os.getcwd())
- python = active VIRTUAL_ENV/bin/python or CONDA_PREFIX/bin/python
with a Python 3.8+ version check; falls back cleanly to
sys.executable if no venv or the candidate fails
- result : 'import pandas' works, '.env' resolves, matches terminal()
strict (opt-in):
- cwd = staging tmpdir (today's behavior)
- python = sys.executable (today's behavior)
- result : maximum reproducibility and isolation; project deps
won't resolve
Security-critical invariants are identical across both modes and covered by
explicit regression tests:
- env scrubbing (strips *_API_KEY, *_TOKEN, *_SECRET, *_PASSWORD,
*_CREDENTIAL, *_PASSWD, *_AUTH substrings)
- SANDBOX_ALLOWED_TOOLS whitelist (no execute_code recursion, no
delegate_task, no MCP from inside scripts)
- resource caps (5-min timeout, 50KB stdout, 50 tool calls)
Deliberately avoids 'sandbox'/'isolated'/'cloud' language in tool
descriptions (regression from commit 39b83f34 where agents on local
backends falsely believed they were sandboxed and refused networking).
Override via env var: HERMES_EXECUTE_CODE_MODE=strict|project
Weaker models (Gemma-class) repeatedly rediscover and forget that execute_code's
working directory differs from terminal()/read_file()'s, leading to
os.path.exists('.env') returning False even though the file exists in the
session's CWD. They then bounce between 'the file exists' and 'the file is
missing' across tool calls.
Adds a 'Working directory' note to the execute_code schema description
pointing agents at absolute paths (os.path.expanduser) or terminal()/read_file()
for inspecting user files.
Carefully avoids the 'sandbox'/'isolated'/'cloud' language that commit
39b83f34 removed (it caused agents on local backends to refuse networking
tasks and save false sandbox beliefs to persistent memory). Purely factual
CWD guidance — no restriction implications.
Error messages that tell users to install optional extras now use
{sys.executable} -m pip install ... instead of a bare 'pip install
hermes-agent[extra]' string. Under the curl installer, bare 'pip'
resolves to system pip, which either fails with PEP 668
externally-managed-environment or installs into the wrong Python.
Affects: hermes dashboard, hermes web server startup, mcp_serve,
hermes doctor Bedrock check, CLI voice mode, voice_mode tool runtime
error, Discord voice-channel join failure message.
* fix(interrupt): propagate to concurrent-tool workers + opt-in debug trace
interrupt() previously only flagged the agent's _execution_thread_id.
Tools running inside _execute_tool_calls_concurrent execute on
ThreadPoolExecutor worker threads whose tids are distinct from the
agent's, so is_interrupted() inside those tools returned False no matter
how many times the gateway called .interrupt() — hung ssh / curl / long
make-builds ran to their own timeout.
Changes:
- run_agent.py: track concurrent-tool worker tids in a per-agent set,
fan interrupt()/clear_interrupt() out to them, and handle the
register-after-interrupt race at _run_tool entry. getattr fallback
for the tracker so test stubs built via object.__new__ keep working.
- tools/environments/base.py: opt-in _wait_for_process trace (ENTER,
per-30s HEARTBEAT with interrupt+activity-cb state, INTERRUPT
DETECTED, TIMEOUT, EXIT) behind HERMES_DEBUG_INTERRUPT=1.
- tools/interrupt.py: opt-in set_interrupt() trace (caller tid, target
tid, set snapshot) behind the same env flag.
- tests: new regression test runs a polling tool on a concurrent worker
and asserts is_interrupted() flips to True within ~1s of interrupt().
Second new test guards clear_interrupt() clearing tracked worker bits.
Validation: tests/run_agent/ all 762 pass; tests/tools/ interrupt+env
subset 216 pass.
* fix(interrupt-debug): bypass quiet_mode logger filter so trace reaches agent.log
AIAgent.__init__ sets logging.getLogger('tools').setLevel(ERROR) when
quiet_mode=True (the CLI default). This would silently swallow every
INFO-level trace line from the HERMES_DEBUG_INTERRUPT=1 instrumentation
added in the parent commit — confirmed by running hermes chat -q with
the flag and finding zero trace lines in agent.log even though
_wait_for_process was clearly executing (subprocess pid existed).
Fix: when HERMES_DEBUG_INTERRUPT=1, each traced module explicitly sets
its own logger level to INFO at import time, overriding the 'tools'
parent-level filter. Scoped to the opt-in case only, so production
(quiet_mode default) logs stay quiet as designed.
Validation: hermes chat -q with HERMES_DEBUG_INTERRUPT=1 now writes
'_wait_for_process ENTER/EXIT' lines to agent.log as expected.
* fix(cli): SIGTERM/SIGHUP no longer orphans tool subprocesses
Tool subprocesses spawned by the local environment backend use
os.setsid so they run in their own process group. Before this fix,
SIGTERM/SIGHUP to the hermes CLI killed the main thread via
KeyboardInterrupt but the worker thread running _wait_for_process
never got a chance to call _kill_process — Python exited, the child
was reparented to init (PPID=1), and the subprocess ran to its
natural end (confirmed live: sleep 300 survived 4+ min after SIGTERM
to the agent until manual cleanup).
Changes:
- cli.py _signal_handler (interactive) + _signal_handler_q (-q mode):
route SIGTERM/SIGHUP through agent.interrupt() so the worker's poll
loop sees the per-thread interrupt flag and calls _kill_process
(os.killpg) on the subprocess group. HERMES_SIGTERM_GRACE (default
1.5s) gives the worker time to complete its SIGTERM+SIGKILL
escalation before KeyboardInterrupt unwinds main.
- tools/environments/base.py _wait_for_process: wrap the poll loop in
try/except (KeyboardInterrupt, SystemExit) so the cleanup fires
even on paths the signal handlers don't cover (direct sys.exit,
unhandled KI from nested code, etc.). Emits EXCEPTION_EXIT trace
line when HERMES_DEBUG_INTERRUPT=1.
- New regression test: injects KeyboardInterrupt into a running
_wait_for_process via PyThreadState_SetAsyncExc, verifies the
subprocess process group is dead within 3s of the exception and
that KeyboardInterrupt re-raises cleanly afterward.
Validation:
| Before | After |
|---------------------------------------------------------|--------------------|
| sleep 300 survives 4+ min as PPID=1 orphan after SIGTERM | dies within 2 s |
| No INTERRUPT DETECTED in trace | INTERRUPT DETECTED fires + killing process group |
| tests/tools/test_local_interrupt_cleanup | 1/1 pass |
| tests/run_agent/test_concurrent_interrupt | 4/4 pass |
Extend forum support from PR #10145:
- REST path (_send_discord): forum thread creation now uploads media
files as multipart attachments on the starter message in a single
call. Previously media files were silently dropped on the forum
path.
- Websocket media paths (_send_file_attachment, send_voice, send_image,
send_animation — covers send_image_file, send_video, send_document
transitively): forum channels now go through a new _forum_post_file
helper that creates a thread with the file as starter content,
instead of failing via channel.send(file=...) which forums reject.
- _send_to_forum chunk follow-up failures are collected into
raw_response['warnings'] so partial-send outcomes surface.
- Process-local probe cache (_DISCORD_CHANNEL_TYPE_PROBE_CACHE) avoids
GET /channels/{id} on every uncached send after the first.
- Dedup of TestSendDiscordMedia that the PR merge-resolution left
behind.
- Docs: Forum Channels section under website/docs/user-guide/messaging/discord.md.
Tests: 117 passed (22 new for forum+media, probe cache, warnings).
ShellFileOperations captured the terminal env's cwd at __init__ time and
used that stale value for every subsequent _exec() call. When the user
ran `cd` via the terminal tool, `env.cwd` updated but `ops.cwd` did not.
Relative paths passed to patch_replace / read_file / write_file / search
then targeted the ORIGINAL directory instead of the current one.
Observed symptom in agent sessions:
terminal: cd .worktrees/my-branch
patch hermes_cli/main.py <old> <new>
→ returns {"success": true} with a plausible unified diff
→ but `git diff` in the worktree shows nothing
→ the patch landed in the main repo's checkout of main.py instead
The diff looked legitimate because patch_replace computes it from the
IN-MEMORY content vs new_content, not by re-reading the file. The
write itself DID succeed — it just wrote to the wrong directory's copy
of the same-named file.
Fix: _exec() now resolves cwd from live sources in this order:
1. Explicit `cwd` arg (if provided by the caller)
2. Live `self.env.cwd` (tracks `cd` commands run via terminal)
3. Init-time `self.cwd` (fallback when env has no cwd attribute)
Includes a 5-test regression suite covering:
- cd followed by relative read follows live cwd
- the exact reported bug: patch_replace with relative path after cd
- explicit cwd= arg still wins over env.cwd
- env without cwd attribute falls back to init-time cwd
- patch_replace success reflects real file state (safety rail)
Co-authored-by: teknium1 <teknium@nousresearch.com>
Follow-up polish on top of the cherry-picked #11023 commit.
- feishu_comment_rules.py: replace import-time "~/.hermes" expanduser fallback
with get_hermes_home() from hermes_constants (canonical, profile-safe).
- tools/feishu_doc_tool.py, tools/feishu_drive_tool.py: drop the
asyncio.get_event_loop().run_until_complete(asyncio.to_thread(...)) dance.
Tool handlers run synchronously in a worker thread with no running loop, so
the RuntimeError branch was always the one that executed. Calls client.request
directly now. Unused asyncio import removed.
- tests/gateway/test_feishu.py: add register_p2_customized_event to the mock
EventDispatcher builder so the existing adapter test matches the new handler
registration for drive.notice.comment_add_v1.
- scripts/release.py: map liujinkun@bytedance.com -> liujinkun2025 for
contributor attribution on release notes.
- Full comment handler: parse drive.notice.comment_add_v1 events, build
timeline, run agent, deliver reply with chunking support.
- 5 tools: feishu_doc_read, feishu_drive_list_comments,
feishu_drive_list_comment_replies, feishu_drive_reply_comment,
feishu_drive_add_comment.
- 3-tier access control rules (exact doc > wildcard "*" > top-level >
defaults) with per-field fallback. Config via
~/.hermes/feishu_comment_rules.json, mtime-cached hot-reload.
- Self-reply filter using generalized self_open_id (supports future
user-identity subscriptions). Receiver check: only process events
where the bot is the @mentioned target.
- Smart timeline selection, long text chunking, semantic text extraction,
session sharing per document, wiki link resolution.
Change-Id: I31e82fd6355173dbcc400b8934b6d9799e3137b9
Both fixes close process leaks observed in production (18+ orphaned
agent-browser node daemons, 15+ orphaned paste.rs sleep interpreters
accumulated over ~3 days, ~2.7 GB RSS).
## agent-browser daemon leak
Previously the orphan reaper (_reap_orphaned_browser_sessions) only ran
from _start_browser_cleanup_thread, which is only invoked on the first
browser tool call in a process. Hermes sessions that never used the
browser never swept orphans, and the cross-process orphan detection
relied on in-process _active_sessions, which doesn't see other hermes
PIDs' sessions (race risk).
- Write <session>.owner_pid alongside the socket dir recording the
hermes PID that owns the daemon (extracted into _write_owner_pid for
direct testability).
- Reaper prefers owner_pid liveness over in-process _active_sessions.
Cross-process safe: concurrent hermes instances won't reap each
other's daemons. Legacy tracked_names fallback kept for daemons
that predate owner_pid.
- atexit handler (_emergency_cleanup_all_sessions) now always runs
the reaper, not just when this process had active sessions —
every clean hermes exit sweeps accumulated orphans.
## paste.rs auto-delete leak
_schedule_auto_delete spawned a detached Python subprocess per call
that slept 6 hours then issued DELETE requests. No dedup, no tracking —
every 'hermes debug share' invocation added ~20 MB of resident Python
interpreters that stuck around until the sleep finished.
- Replaced the spawn with ~/.hermes/pastes/pending.json: records
{url, expire_at} entries.
- _sweep_expired_pastes() synchronously DELETEs past-due entries on
every 'hermes debug' invocation (run_debug() dispatcher).
- Network failures stay in pending.json for up to 24h, then give up
(paste.rs's own retention handles the 'user never runs hermes again'
edge case).
- Zero subprocesses; regression test asserts subprocess/Popen/time.sleep
never appear in the function source (skipping docstrings via AST).
## Validation
| | Before | After |
|------------------------------|---------------|--------------|
| Orphan agent-browser daemons | 18 accumulated| 2 (live) |
| paste.rs sleep interpreters | 15 accumulated| 0 |
| RSS reclaimed | - | ~2.7 GB |
| Targeted tests | - | 2253 pass |
E2E verified: alive-owner daemons NOT reaped; dead-owner daemons
SIGTERM'd and socket dirs cleaned; pending.json sweep deletes expired
entries without spawning subprocesses.
Two accretion-over-time leaks that compound over long CLI / gateway
lifetimes. Both were flagged in the memory-leak audit.
## file_tools._read_tracker
_read_tracker[task_id] holds three sub-containers that grew unbounded:
read_history set of (path, offset, limit) tuples — 1 per unique read
dedup dict of (path, offset, limit) → mtime — same growth pattern
read_timestamps dict of resolved_path → mtime — 1 per unique path
A CLI session uses one stable task_id for its lifetime, so these were
uncapped. A 10k-read session accumulated ~1.5MB of tracker state that
the tool no longer needed (only the most recent reads are relevant for
dedup, consecutive-loop detection, and write/patch external-edit
warnings).
Fix: _cap_read_tracker_data() enforces hard caps on each container
after every add. Defaults: read_history=500, dedup=1000,
read_timestamps=1000. Eviction is insertion-order (Python 3.7+ dict
guarantee) for the dicts; arbitrary for the set (which only feeds
diagnostic summaries).
## process_registry._completion_consumed
Module-level set that recorded every session_id ever polled / waited /
logged. No pruning. Each entry is ~20 bytes, so the absolute leak is
small, but on a gateway processing thousands of background commands
per day the set grows until process exit.
Fix: _prune_if_needed() now discards _completion_consumed entries
alongside the session dict evictions it already performs (both the
TTL-based prune and the LRU-over-cap prune). Adds a final
belt-and-suspenders pass that drops any dangling entries whose
session_id no longer appears in _running or _finished.
Tests: tests/tools/test_accretion_caps.py — 9 cases
* Each container bound respected, oldest evicted
* No-op when under cap (no unnecessary work)
* Handles missing sub-containers without crashing
* Live read_file_tool path enforces caps end-to-end
* _completion_consumed pruned on TTL expiry
* _completion_consumed pruned on LRU eviction
* Dangling entries (no backing session) cleared
Broader suite: 3486 tests/tools + tests/cli pass. The single flake
(test_alias_command_passes_args) reproduces on unchanged main — known
cross-test pollution under suite-order load.
- gateway/platforms/weixin.py:
- Split aiohttp.ClientSession into _poll_session and _send_session
- Add _LIVE_ADAPTERS registry so send_weixin_direct() reuses the connected gateway adapter instead of creating a competing session
- Fixes silent message loss when gateway is running (iLink token contention)
- cron/scheduler.py:
- Support comma-separated deliver values (e.g. 'feishu,weixin') for multi-target delivery
- Delay pconfig/enabled check until standalone fallback so live adapters work even when platform is not in gateway config
- tools/send_message_tool.py:
- Synthesize PlatformConfig from WEIXIN_* env vars when gateway config lacks a weixin entry
- Fall back to WEIXIN_HOME_CHANNEL env var for home channel resolution
- tests/gateway/test_weixin.py:
- Update mocks to include _send_session
The send_message tool's direct-REST QQBot path used "QQBotAccessToken {token}"
which QQ's API rejects with 401. The correct format is "QQBot {token}" — the
gateway adapter at gateway/platforms/qqbot.py uses this format in all 5 header
sites (lines 341, 551, 579, 1068, 1467); this was the one outlier.
Credit to @Quon for surfacing this in #10257 (that PR had unrelated issues in
its media-upload logic and was closed; this salvages the genuine 1-line fix).
When a user edits a bundled skill, sync flags it as user_modified and
skips it forever. The problem: if the user later tries to undo the edit
by copying the current bundled version back into ~/.hermes/skills/, the
manifest still holds the old origin hash from the last successful
sync, so the fresh bundled hash still doesn't match and the skill stays
stuck as user_modified.
Adds an escape hatch for this case.
hermes skills reset <name>
Drops the skill's entry from ~/.hermes/skills/.bundled_manifest and
re-baselines against the user's current copy. Future 'hermes update'
runs accept upstream changes again. Non-destructive.
hermes skills reset <name> --restore
Also deletes the user's copy and re-copies the bundled version.
Use when you want the pristine upstream skill back.
Also available as /skills reset in chat.
- tools/skills_sync.py: new reset_bundled_skill(name, restore=False)
- hermes_cli/skills_hub.py: do_reset() + wired into skills_command and
handle_skills_slash; added to the slash /skills help panel
- hermes_cli/main.py: argparse entry for 'hermes skills reset'
- tests/tools/test_skills_sync.py: 5 new tests covering the stuck-flag
repro, --restore, unknown-skill error, upstream-removed-skill, and
no-op on already-clean state
- website/docs/user-guide/features/skills.md: new 'Bundled skill updates'
section explaining the origin-hash mechanic + reset usage
* feat(image_gen): upgrade Recraft V3 → V4 Pro, Nano Banana → Pro
Upstream asked for these two upgrades ASAP — the old entries show
stale models when newer, higher-quality versions are available on FAL.
Recraft V3 → Recraft V4 Pro
ID: fal-ai/recraft-v3 → fal-ai/recraft/v4/pro/text-to-image
Price: $0.04/image → $0.25/image (6x — V4 Pro is premium tier)
Schema: V4 dropped the required `style` enum entirely; defaults
handle taste now. Added `colors` and `background_color`
to supports for brand-palette control. `seed` is not
supported by V4 per the API docs.
Nano Banana → Nano Banana Pro
ID: fal-ai/nano-banana → fal-ai/nano-banana-pro
Price: $0.08/image → $0.15/image (1K); $0.30 at 4K
Schema: Aspect ratio family unchanged. Added `resolution`
(1K/2K/4K, default 1K for billing predictability),
`enable_web_search` (real-time info grounding, +$0.015),
and `limit_generations` (force exactly 1 image).
Architecture: Gemini 2.5 Flash → Gemini 3 Pro Image. Quality
and reasoning depth improved; slower (~6s → ~8s).
Migration: users who had the old IDs in `image_gen.model` will
fall through the existing 'unknown model → default' warning path
in `_resolve_fal_model()` and get the Klein 9B default on the next
run. Re-run `hermes tools` → Image Generation to pick the new
version. No silent cost-upgrade aliasing — the 2-6x price jump
on these tiers warrants explicit user re-selection.
Portal note: both new model IDs need to be allowlisted on the
Nous fal-queue-gateway alongside the previous 7 additions, or
users on Nous Subscription will see the 'managed gateway rejected
model' error we added previously (which is clear and
self-remediating, just noisy).
* docs: wrap '<1s' in backticks to unblock MDX compilation
Docusaurus's MDX parser treats unquoted '<' as the start of JSX, and
'<1s' fails because '1' isn't a valid tag-name start character. This
was broken on main since PR #11265 (never noticed because
docs-site-checks was failing on OTHER issues at the time and we
admin-merged through it).
Wrapping in backticks also gives the cell monospace styling which
reads more cleanly alongside the inline-code model ID in the same row.
The other '<1s' occurrence (line 52) is inside a fenced code block
and is already safe — code fences bypass MDX parsing.
* feat(mcp-oauth): scaffold MCPOAuthManager
Central manager for per-server MCP OAuth state. Provides
get_or_build_provider (cached), remove (evicts cache + deletes
disk), invalidate_if_disk_changed (mtime watch, core fix for
external-refresh workflow), and handle_401 (dedup'd recovery).
No behavior change yet — existing call sites still use
build_oauth_auth directly. Task 1 of 8 in the MCP OAuth
consolidation (fixes Cthulhu's BetterStack reliability issues).
* feat(mcp-oauth): add HermesMCPOAuthProvider with pre-flow disk watch
Subclasses the MCP SDK's OAuthClientProvider to inject a disk
mtime check before every async_auth_flow, via the central
manager. When a subclass instance is used, external token
refreshes (cron, another CLI instance) are picked up before
the next API call.
Still dead code: the manager's _build_provider still delegates
to build_oauth_auth and returns the plain OAuthClientProvider.
Task 4 wires this subclass in. Task 2 of 8.
* refactor(mcp-oauth): extract build_oauth_auth helpers
Decomposes build_oauth_auth into _configure_callback_port,
_build_client_metadata, _maybe_preregister_client, and
_parse_base_url. Public API preserved. These helpers let
MCPOAuthManager._build_provider reuse the same logic in Task 4
instead of duplicating the construction dance.
Also updates the SDK version hint in the warning from 1.10.0 to
1.26.0 (which is what we actually require for the OAuth types
used here). Task 3 of 8.
* feat(mcp-oauth): manager now builds HermesMCPOAuthProvider directly
_build_provider constructs the disk-watching subclass using the
helpers from Task 3, instead of delegating to the plain
build_oauth_auth factory. Any consumer using the manager now gets
pre-flow disk-freshness checks automatically.
build_oauth_auth is preserved as the public API for backwards
compatibility. The code path is now:
MCPOAuthManager.get_or_build_provider ->
_build_provider ->
_configure_callback_port
_build_client_metadata
_maybe_preregister_client
_parse_base_url
HermesMCPOAuthProvider(...)
Task 4 of 8.
* feat(mcp): wire OAuth manager + add _reconnect_event
MCPServerTask gains _reconnect_event alongside _shutdown_event.
When set, _run_http / _run_stdio exit their async-with blocks
cleanly (no exception), and the outer run() loop re-enters the
transport to rebuild the MCP session with fresh credentials.
This is the recovery path for OAuth failures that the SDK's
in-place httpx.Auth cannot handle (e.g. cron externally consumed
the refresh_token, or server-side session invalidation).
_run_http now asks MCPOAuthManager for the OAuth provider
instead of calling build_oauth_auth directly. Config-time,
runtime, and reconnect paths all share one provider instance
with pre-flow disk-watch active.
shutdown() defensively sets both events so there is no race
between reconnect and shutdown signalling.
Task 5 of 8.
* feat(mcp): detect auth failures in tool handlers, trigger reconnect
All 5 MCP tool handlers (tool call, list_resources, read_resource,
list_prompts, get_prompt) now detect auth failures and route
through MCPOAuthManager.handle_401:
1. If the manager says recovery is viable (disk has fresh tokens,
or SDK can refresh in-place), signal MCPServerTask._reconnect_event
to tear down and rebuild the MCP session with fresh credentials,
then retry the tool call once.
2. If no recovery path exists, return a structured needs_reauth
JSON error so the model stops hallucinating manual refresh
attempts (the 'let me curl the token endpoint' loop Cthulhu
pasted from Discord).
_is_auth_error catches OAuthFlowError, OAuthTokenError,
OAuthNonInteractiveError, and httpx.HTTPStatusError(401). Non-auth
exceptions still surface via the generic error path unchanged.
Task 6 of 8.
* feat(mcp-cli): route add/remove through manager, add 'hermes mcp login'
cmd_mcp_add and cmd_mcp_remove now go through MCPOAuthManager
instead of calling build_oauth_auth / remove_oauth_tokens
directly. This means CLI config-time state and runtime MCP
session state are backed by the same provider cache — removing
a server evicts the live provider, adding a server populates
the same cache the MCP session will read from.
New 'hermes mcp login <name>' command:
- Wipes both the on-disk tokens file and the in-memory
MCPOAuthManager cache
- Triggers a fresh OAuth browser flow via the existing probe
path
- Intended target for the needs_reauth error Task 6 returns
to the model
Task 7 of 8.
* test(mcp-oauth): end-to-end integration tests
Five new tests exercising the full consolidation with real file
I/O and real imports (no transport mocks):
1. external_refresh_picked_up_without_restart — Cthulhu's cron
workflow. External process writes fresh tokens to disk;
on the next auth flow the manager's mtime-watch flips
_initialized and the SDK re-reads from storage.
2. handle_401_deduplicates_concurrent_callers — 10 concurrent
handlers for the same failed token fire exactly ONE recovery
attempt (thundering-herd protection).
3. handle_401_returns_false_when_no_provider — defensive path
for unknown servers.
4. invalidate_if_disk_changed_handles_missing_file — pre-auth
state returns False cleanly.
5. provider_is_reused_across_reconnects — cache stickiness so
reconnects preserve the disk-watch baseline mtime.
Task 8 of 8 — consolidation complete.
* feat(image_gen): multi-model FAL support with picker in hermes tools
Adds 8 FAL text-to-image models selectable via `hermes tools` →
Image Generation → (FAL.ai | Nous Subscription) → model picker.
Models supported:
- fal-ai/flux-2/klein/9b (new default, <1s, $0.006/MP)
- fal-ai/flux-2-pro (previous default, kept backward-compat upscaling)
- fal-ai/z-image/turbo (Tongyi-MAI, bilingual EN/CN)
- fal-ai/nano-banana (Gemini 2.5 Flash Image)
- fal-ai/gpt-image-1.5 (with quality tier: low/medium/high)
- fal-ai/ideogram/v3 (best typography)
- fal-ai/recraft-v3 (vector, brand styles)
- fal-ai/qwen-image (LLM-based)
Architecture:
- FAL_MODELS catalog declares per-model size family, defaults, supports
whitelist, and upscale flag. Three size families handled uniformly:
image_size_preset (flux family), aspect_ratio (nano-banana), and
gpt_literal (gpt-image-1.5).
- _build_fal_payload() translates unified inputs (prompt + aspect_ratio)
into model-specific payloads, merges defaults, applies caller overrides,
wires GPT quality_setting, then filters to the supports whitelist — so
models never receive rejected keys.
- IMAGEGEN_BACKENDS registry in tools_config prepares for future imagegen
providers (Replicate, Stability, etc.); each provider entry tags itself
with imagegen_backend: 'fal' to select the right catalog.
- Upscaler (Clarity) defaults off for new models (preserves <1s value
prop), on for flux-2-pro (backward-compat). Per-model via FAL_MODELS.
Config:
image_gen.model = fal-ai/flux-2/klein/9b (new)
image_gen.quality_setting = medium (new, GPT only)
image_gen.use_gateway = bool (existing)
Agent-facing schema unchanged (prompt + aspect_ratio only) — model
choice is a user-level config decision, not an agent-level arg.
Picker uses curses_radiolist (arrow keys, auto numbered-fallback on
non-TTY). Column-aligned: Model / Speed / Strengths / Price.
Docs: image-generation.md rewritten with the model table and picker
walkthrough. tools-reference, tool-gateway, overview updated to drop
the stale "FLUX 2 Pro" wording.
Tests: 42 new in tests/tools/test_image_generation.py covering catalog
integrity, all 3 size families, supports filter, default merging, GPT
quality wiring, model resolution fallback. 8 new in
tests/hermes_cli/test_tools_config.py for picker wiring (registry,
config writes, GPT quality follow-up prompt, corrupt-config repair).
* feat(image_gen): translate managed-gateway 4xx to actionable error
When the Nous Subscription managed FAL proxy rejects a model with 4xx
(likely portal-side allowlist miss or billing gate), surface a clear
message explaining:
1. The rejected model ID + HTTP status
2. Two remediation paths: set FAL_KEY for direct access, or
pick a different model via `hermes tools`
5xx, connection errors, and direct-FAL errors pass through unchanged
(those have different root causes and reasonable native messages).
Motivation: new FAL models added to this release (flux-2-klein-9b,
z-image-turbo, nano-banana, gpt-image-1.5, ideogram-v3, recraft-v3,
qwen-image) are untested against the Nous Portal proxy. If the portal
allowlists model IDs, users on Nous Subscription will hit cryptic
4xx errors without guidance on how to work around it.
Tests: 8 new cases covering status extraction across httpx/fal error
shapes and 4xx-vs-5xx-vs-ConnectionError translation policy.
Docs: brief note in image-generation.md for Nous subscribers.
Operator action (Nous Portal side): verify that fal-queue-gateway
passes through these 7 new FAL model IDs. If the proxy has an
allowlist, add them; otherwise Nous Subscription users will see the
new translated error and fall back to direct FAL.
* feat(image_gen): pin GPT-Image quality to medium (no user choice)
Previously the tools picker asked a follow-up question for GPT-Image
quality tier (low / medium / high) and persisted the answer to
`image_gen.quality_setting`. This created two problems:
1. Nous Portal billing complexity — the 22x cost spread between tiers
($0.009 low / $0.20 high) forces the gateway to meter per-tier per
user, which the portal team can't easily support at launch.
2. User footgun — anyone picking `high` by mistake burns through
credit ~6x faster than `medium`.
This commit pins quality at medium by baking it into FAL_MODELS
defaults for gpt-image-1.5 and removes all user-facing override paths:
- Removed `_resolve_gpt_quality()` runtime lookup
- Removed `honors_quality_setting` flag on the model entry
- Removed `_configure_gpt_quality_setting()` picker helper
- Removed `_GPT_QUALITY_CHOICES` constant
- Removed the follow-up prompt call in `_configure_imagegen_model()`
- Even if a user manually edits `image_gen.quality_setting` in
config.yaml, no code path reads it — always sends medium.
Tests:
- Replaced TestGptQualitySetting (6 tests) with TestGptQualityPinnedToMedium
(5 tests) — proves medium is baked in, config is ignored, flag is
removed, helper is removed, non-gpt models never get quality.
- Replaced test_picker_with_gpt_image_also_prompts_quality with
test_picker_with_gpt_image_does_not_prompt_quality — proves only 1
picker call fires when gpt-image is selected (no quality follow-up).
Docs updated: image-generation.md replaces the quality-tier table
with a short note explaining the pinning decision.
* docs(image_gen): drop stale 'wires GPT quality tier' line from internals section
Caught in a cleanup sweep after pinning quality to medium. The
"How It Works Internally" walkthrough still described the removed
quality-wiring step.
Follow-ups on top of kshitijk4poor's cherry-picked salvage of PR #8018:
tools/environments/daytona.py
- PID-suffix /tmp/.hermes_sync.<pid>.tar so concurrent sync_back calls
against the same sandbox don't collide on the remote temp path
- Move sync_back() inside the cleanup lock and after the _sandbox-None
guard, with its own try/except. Previously a no-op cleanup (sandbox
already cleared) still fired sync_back → 3-attempt retry storm against
a nil sandbox (~6s of sleep). Now short-circuits cleanly.
tools/environments/file_sync.py
- Add _SYNC_BACK_MAX_BYTES (2 GiB) defensive cap: refuse to extract a
tar larger than the limit. Protects against runaway sandboxes
producing arbitrary-size archives.
- Add 'nothing previously pushed' guard at the top of sync_back(). If
_pushed_hashes and _synced_files are both empty, the FileSyncManager
was never initialized from the host side — there is nothing coherent
to sync back. Skips the retry/backoff machinery on uninitialized
managers and eliminates test-suite slowdown from pre-existing cleanup
tests that don't mock the sync layer.
tests/tools/test_file_sync_back.py
- Update _make_manager helper to seed a _pushed_hashes entry by default
so sync_back() exercises its real path. A seed_pushed_state=False
opt-out is available for noop-path tests.
- Add TestSyncBackSizeCap with positive and negative coverage of the
new cap.
tests/tools/test_sync_back_backends.py
- Update Daytona bulk download test to assert the PID-suffixed path
pattern instead of the fixed /tmp/.hermes_sync.tar.
Salvage of PR #8018 by @alt-glitch onto current main.
On sandbox teardown, FileSyncManager now downloads the remote .hermes/
directory, diffs against SHA-256 hashes of what was originally pushed,
and applies only changed files back to the host.
Core (tools/environments/file_sync.py):
- sync_back(): orchestrates download -> unpack -> diff -> apply with:
- Retry with exponential backoff (3 attempts, 2s/4s/8s)
- SIGINT trap + defer (prevents partial writes on Ctrl-C)
- fcntl.flock serialization (concurrent gateway sandboxes)
- Last-write-wins conflict resolution with warning
- New remote files pulled back via _infer_host_path prefix matching
Backends:
- SSH: _ssh_bulk_download — tar cf - piped over SSH
- Modal: _modal_bulk_download — exec tar cf - -> proc.stdout.read
- Daytona: _daytona_bulk_download — exec tar cf -> SDK download_file
- All three call sync_back() at the top of cleanup()
Fixes applied during salvage (vs original PR #8018):
| # | Issue | Fix |
|---|-------|-----|
| C1 | import fcntl unconditional — crashes Windows | try/except with fallback; _sync_back_locked skips locking when fcntl=None |
| W1 | assert for runtime guard (stripped by -O) | Replaced with proper if/raise RuntimeError |
| W2 | O(n*m) from _get_files_fn() called per file | Cache mapping once at start of _sync_back_impl, pass to resolve/infer |
| W3 | Dead BulkDownloadFn imports in 3 backends | Removed unused imports |
| W4 | Modal hardcodes root/.hermes, no explanation | Added docstring comment explaining Modal always runs as root |
| S1 | SHA-256 computed for new files where pushed_hash=None | Skip hashing when pushed_hash is None (comparison always False) |
| S2 | Daytona /tmp/.hermes_sync.tar never cleaned up | Added rm -f after download (best-effort) |
Tests: 49 passing (17 new: _infer_host_path edge cases, SIGINT
main/worker thread, Windows fcntl=None fallback, Daytona tar cleanup).
Based on #8018 by @alt-glitch.
Users with 'commit.gpgsign = true' in their global git config got a
pinentry popup (or a failed commit) every time the agent took a
background filesystem snapshot — every write_file, patch, or diff
mid-session. With GPG_TTY unset, pinentry-qt/gtk would spawn a GUI
window, constantly interrupting the session.
The shadow repo is internal Hermes infrastructure. It must not
inherit user-level git settings (signing, hooks, aliases, credential
helpers, etc.) under any circumstance.
Fix is layered:
1. _git_env() sets GIT_CONFIG_GLOBAL=os.devnull,
GIT_CONFIG_SYSTEM=os.devnull, and GIT_CONFIG_NOSYSTEM=1. Shadow
git commands no longer see ~/.gitconfig or /etc/gitconfig at all
(uses os.devnull for Windows compat).
2. _init_shadow_repo() explicitly writes commit.gpgsign=false and
tag.gpgSign=false into the shadow's own config, so the repo is
correct even if inspected or run against directly without the
env vars, and for older git versions (<2.32) that predate
GIT_CONFIG_GLOBAL.
3. _take() passes --no-gpg-sign inline on the commit call. This
covers existing shadow repos created before this fix — they will
never re-run _init_shadow_repo (it is gated on HEAD not existing),
so they would miss layer 2. Layer 1 still protects them, but the
inline flag guarantees correctness at the commit call itself.
Existing checkpoints, rollback, list, diff, and restore all continue
to work — history is untouched. Users who had the bug stop getting
pinentry popups; users who didn't see no observable change.
Tests: 5 new regression tests in TestGpgAndGlobalConfigIsolation,
including a full E2E repro with fake HOME, global gpgsign=true, and
a deliberately broken GPG binary — checkpoint succeeds regardless.
The blocking gateway approval wait at tools/approval.py called
`entry.event.wait(timeout=...)` which never touched the agent's
activity tracker. When a user was slow to respond to a /approve prompt
(or the gateway_timeout config was set higher than the default 300s),
the agent thread sat silent long enough for the gateway's inactivity
watchdog (agent.gateway_timeout, default 1800s) to kill it — even
though the agent was doing exactly the right thing and the user was
the one causing the delay.
The fix polls the event in 1s slices and calls touch_activity_if_due
between slices, mirroring the _wait_for_process() pattern in
tools/environments/base.py that covers the subprocess-waiting side of
the same problem. At the default 10s heartbeat cadence, a 300s
approval wait now pings activity ~30 times, well under the 1800s
idle threshold.
Observed in community user logs: 12 repeated 'Agent idle 1800s,
last_activity=executing tool: terminal' events across April 12-14.
Companion to PR #10501 which covered streaming / concurrent-tool /
Modal-backend gaps but did not touch approval.py.
Test: tests/tools/test_approval_heartbeat.py — verifies (1) heartbeats
fire during the wait, (2) user responses are still near-instant, and
(3) the approval path stays functional when the heartbeat helper
can't be imported.
Adds Google Gemini TTS as the seventh voice provider, with 30 prebuilt
voices (Zephyr, Puck, Kore, Enceladus, Gacrux, etc.) and natural-language
prompt control. Integrates through the existing provider chain:
- tools/tts_tool.py: new _generate_gemini_tts() calls the
generativelanguage REST endpoint with responseModalities=[AUDIO],
wraps the returned 24kHz mono 16-bit PCM (L16) in a WAV RIFF header,
then ffmpeg-converts to MP3 or Opus depending on output extension.
For .ogg output, libopus is forced explicitly so Telegram voice
bubbles get Opus (ffmpeg defaults to Vorbis for .ogg).
- hermes_cli/tools_config.py: exposes 'Google Gemini TTS' as a provider
option in the curses-based 'hermes tools' UI.
- hermes_cli/setup.py: adds gemini to the setup wizard picker, tool
status display, and API key prompt branch (accepts existing
GEMINI_API_KEY or GOOGLE_API_KEY, falls back to Edge if neither set).
- tests/tools/test_tts_gemini.py: 15 unit tests covering WAV header
wrap correctness, env var fallback (GEMINI/GOOGLE), voice/model
overrides, snake_case vs camelCase inlineData handling, HTTP error
surfacing, and empty-audio edge cases.
- docs: TTS features page updated to list seven providers with the new
gemini config block and ffmpeg notes.
Live-tested against api key against gemini-2.5-flash-preview-tts: .wav,
.mp3, and Telegram-compatible .ogg (Opus codec) all produce valid
playable audio.
Replace the HERMES_ENABLE_NOUS_MANAGED_TOOLS env-var feature flag with
subscription-based detection. The Tool Gateway is now available to any
paid Nous subscriber without needing a hidden env var.
Core changes:
- managed_nous_tools_enabled() checks get_nous_auth_status() +
check_nous_free_tier() instead of an env var
- New use_gateway config flag per tool section (web, tts, browser,
image_gen) records explicit user opt-in and overrides direct API
keys at runtime
- New prefers_gateway(section) shared helper in tool_backend_helpers.py
used by all 4 tool runtimes (web, tts, image gen, browser)
UX flow:
- hermes model: after Nous login/model selection, shows a curses
prompt listing all gateway-eligible tools with current status.
User chooses to enable all, enable only unconfigured tools, or skip.
Defaults to Enable for new users, Skip when direct keys exist.
- hermes tools: provider selection now manages use_gateway flag —
selecting Nous Subscription sets it, selecting any other provider
clears it
- hermes status: renamed section to Nous Tool Gateway, added
free-tier upgrade nudge for logged-in free users
- curses_radiolist: new description parameter for multi-line context
that survives the screen clear
Runtime behavior:
- Each tool runtime (web_tools, tts_tool, image_generation_tool,
browser_use) checks prefers_gateway() before falling back to
direct env-var credentials
- get_nous_subscription_features() respects use_gateway flags,
suppressing direct credential detection when the user opted in
Removed:
- HERMES_ENABLE_NOUS_MANAGED_TOOLS env var and all references
- apply_nous_provider_defaults() silent TTS auto-set
- get_nous_subscription_explainer_lines() static text
- Override env var warnings (use_gateway handles this properly now)
The completion-line printing block (idx = entry['task_index'] etc.)
was outside the 'for future in done:' loop but referenced 'entry'
which is only assigned inside that loop. When concurrent.futures.wait()
returns with an empty 'done' set (timeout expired, no futures finished),
the loop body never executes and 'entry' is unbound.
Moved the completion-line printing and spinner-update code inside
the for loop so each completed future gets its own status line,
and empty poll cycles simply loop back without accessing 'entry'.
- Extract duplicated activity-callback polling into shared
touch_activity_if_due() helper in tools/environments/base.py
- Use helper from both base.py _wait_for_process and
code_execution_tool.py local polling loop (DRY)
- Add test assertion that timeout output field contains the
timeout message and emoji (#10807)
- Add stream_consumer test for tool-boundary fallback scenario
where continuation is empty but final_text differs from
visible prefix (#10807)
When execute_code times out, the result JSON had status="timeout" and an
error field, but the output field was empty. Many models treat empty
output as "nothing happened" and produce an empty/minimal response. The
gateway stream consumer then considers the response "already sent" (from
pre-tool streaming) and silently drops it — leaving the user staring at
silence.
Three changes:
1. Include the timeout message in the output field (both local and remote
paths) so the model always has visible content to relay to the user.
2. Add periodic activity callbacks to the local execution polling loop so
the gateway's inactivity monitor knows execute_code is alive during
long runs.
3. Fix stream_consumer._send_fallback_final to not silently drop content
when the continuation appears empty but the final text differs from
what was previously streamed (e.g. after a tool boundary reset).
Wraps provider.create_session() in _get_session_info() with try/except
to catch cloud provider runtime failures (timeouts, auth errors, rate
limits, invalid responses). Falls back to _create_local_session() so
browser automation continues working when cloud APIs are down.
Marks fallback sessions with fallback_from_cloud, fallback_reason, and
fallback_provider metadata for observability. If both cloud and local
fail, raises RuntimeError with chained context from both errors.
Closes#10883
Co-authored-by: konsisumer <konsisumer@users.noreply.github.com>
* fix: stop /model from silently rerouting direct providers to OpenRouter (#10300)
detect_provider_for_model() silently remapped models to OpenRouter when
the direct provider's credentials weren't found via env vars. Three bugs:
1. Credential check only looked at env vars from PROVIDER_REGISTRY,
missing credential pool entries, auth store, and OAuth tokens
2. When env var check failed, silently returned ('openrouter', slug)
instead of the direct provider the model actually belongs to
3. Users with valid credentials via non-env-var mechanisms (pool,
OAuth, Claude Code tokens) got silently rerouted
Fix:
- Expand credential check to also query credential pool and auth store
- Always return the direct provider match regardless of credential
status -- let client init handle missing creds with a clear error
rather than silently routing through the wrong provider
Same philosophy as the provider-required fix: don't guess, don't
silently reroute, error clearly when something is missing.
Closes#10300
* fix: word-wrap spinner, interruptable agent join, and delegate_task interrupt
Three fixes:
1. Spinner widget clips long tool commands — prompt_toolkit Window had
height=1 and wrap_lines=False. Now uses wrap_lines=True with dynamic
height from text length / terminal width. Long commands wrap naturally.
2. agent_thread.join() blocked forever after interrupt — if the agent
thread took time to clean up, the process_loop thread froze. Now polls
with 0.2s timeout on the interrupt path, checking _should_exit so
double Ctrl+C breaks out immediately.
3. Root cause of 5-hour CLI hang: delegate_task() used as_completed()
with no interrupt check. When subagent children got stuck, the parent
blocked forever inside the ThreadPoolExecutor. Now polls with
wait(timeout=0.5) and checks parent_agent._interrupt_requested each
iteration. Stuck children are reported as interrupted, and the parent
returns immediately.
Fixes 12 CI test failures:
1. test_cli_new_session (4): _FakeAgent missing commit_memory_session
attribute added in the memory provider refactoring. Added MagicMock.
2. test_run_progress_topics (1): already_sent detection only checked
stream consumer flags, missing the response_previewed path from
interim_assistant_callback. Restructured guard to check both paths.
3. test_timezone (1): HERMES_TIMEZONE leaked into child processes via
_SAFE_ENV_PREFIXES matching HERMES_*. The code correctly converts
it to TZ but didn't remove the original. Added child_env.pop().
4. test_session_env (1): contextvars baseline captured from a different
context couldn't be restored after clear. Changed assertion to verify
the test's value was removed rather than comparing to a fragile baseline.
5. test_discord_slash_commands (5): already fixed on current main.
When an MCP server returns errors consistently (crashed, disconnected,
auth expired), the model sees each error and retries the tool call.
With no circuit breaker, this burned through all 90 iterations — each
one a full LLM API call plus failed MCP call — producing 15-45 minutes
of zero useful output while the gateway inactivity timeout never fired
(because the agent WAS active, just uselessly).
Fix: track consecutive error counts per MCP server. After 3 consecutive
failures (connection errors, MCP-level errors, or transport exceptions),
the handler short-circuits with a message telling the model to stop
retrying and use alternative approaches. The counter resets to 0 on
any successful call.
Closes#10447
Wrap the TelegramAdapter import in _send_to_platform() with a try/except
ImportError guard, matching the existing Feishu pattern in the same function.
When python-telegram-bot is not installed, the import no longer crashes the
cron scheduler. Instead, MAX_MESSAGE_LENGTH falls back to a hardcoded 4096.
The _send_telegram() function already had its own ImportError guard for the
telegram package; this fixes the remaining bare import of TelegramAdapter
in the platform-routing function.
Text-only Matrix sends should continue using the lightweight _send_matrix()
HTTP helper (~100ms). Only route through the heavy MatrixAdapter (full sync +
E2EE setup) when media files are present. Adds test verifying text-only
messages don't take the adapter path.
Matrix media delivery was silently dropped by send_message because Matrix
wasn't wired into the native adapter-backed media path. Only Telegram,
Discord, and Weixin had native media support.
Adds _send_matrix_via_adapter() which creates a MatrixAdapter instance,
connects, sends text + media via the adapter's native upload methods
(send_document, send_image_file, send_video, send_voice), then disconnects.
Also fixes a stale URL-encoding assertion in test_send_message_missing_platforms
that broke after PR #10151 added quote() to room IDs.
Cherry-picked from PR #10486 by helix4u.
bash -lic with a PTY enables job control (set -m), which waits for all
background jobs before the shell exits. A command like
`python3 -m http.server &>/dev/null &` hangs forever because the shell
never completes.
Prefix `set +m;` to disable job control while keeping -i for .bashrc
sourcing and PTY for interactive tools.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
_load_skill_payload() reconstructed skill_dir as SKILLS_DIR / relative_path,
which is wrong for external skills from skills.external_dirs — they live
outside SKILLS_DIR entirely. Scripts and linked files failed to load.
Fix: skill_view() now includes the absolute skill_dir in its result dict.
_load_skill_payload() uses that directly when available, falling back to
the SKILLS_DIR-relative reconstruction only for legacy responses.
Closes#10313
Add "HERMES_" to _SAFE_ENV_PREFIXES in code_execution_tool.py so HERMES_HOME and other Hermes env vars pass through to execute_code subprocesses. Fixes vision_analyze and other tools that rely on get_hermes_home() failing in Docker environments with non-default HERMES_HOME.
Authored by @shin4.
* fix: show correct env var name in provider API key error (#9506)
The error message for missing provider API keys dynamically built
the env var name as PROVIDER_API_KEY (e.g. ALIBABA_API_KEY), but
some providers use different names (alibaba uses DASHSCOPE_API_KEY).
Users following the error message set the wrong variable.
Fix: look up the actual env var from PROVIDER_REGISTRY before
building the error. Falls back to the dynamic name if the registry
lookup fails.
Closes#9506
* fix: five HERMES_HOME profile-isolation leaks (#5947)
Bug A: Thread session_title from session_db to memory provider init kwargs
so honcho can derive chat-scoped session keys instead of falling back to
cwd-based naming that merges all gateway users into one session.
Bug B: Replace 14 hardcoded ~/.hermes/skills/ paths across 10 skill files
with HERMES_HOME-aware alternatives (${HERMES_HOME:-$HOME/.hermes} in
shell, os.environ.get('HERMES_HOME', ...) in Python).
Bug C: install.sh now respects HERMES_HOME env var and adds --hermes-home
flag. Previously --dir only set INSTALL_DIR while HERMES_HOME was always
hardcoded to $HOME/.hermes.
Bug D: Remove hardcoded ~/.hermes/honcho.json fallback in resolve_config_path().
Non-default profiles no longer silently inherit the default profile's honcho
config. Falls through to ~/.honcho/config.json (global) instead.
Bug E: Guard _edit_skill, _patch_skill, _delete_skill, _write_file, and
_remove_file against writing to skills found in external_dirs. Skills
outside the local SKILLS_DIR are now read-only from the agent's perspective.
Closes#5947
_install_tirith() uses shutil.move() to place the binary from tmpdir
to ~/.hermes/bin/. When these are on different filesystems (common in
Docker, NFS), shutil.move() falls back to copy2 + unlink, but copy2's
metadata step can raise PermissionError. This exception propagated
past the fail_open guard, crashing the terminal tool entirely.
Additionally, a failed install could leave a non-executable tirith
binary at the destination, causing a retry loop on every subsequent
terminal command.
Fix:
- Catch OSError from shutil.move() and fall back to shutil.copy()
(skips metadata/xattr copying that causes PermissionError)
- If even copy fails, clean up the partial dest file to prevent
the non-executable retry loop
- Return (None, 'cross_device_copy_failed') so the failure routes
through the existing install-failure caching and fail_open logic
Closes#10127
When a user runs /browser connect to attach browser tools to their real
Chrome instance via CDP, the BROWSER_CDP_URL env var is set. However,
every browser tool function checks _is_camofox_mode() first, which
short-circuits to the Camofox backend before _get_session_info() ever
checks for the CDP override.
Fix: is_camofox_mode() now returns False when BROWSER_CDP_URL is set,
so the explicit CDP connection takes priority. This is the correct
behavior — /browser connect is an intentional user override.
Reported by SkyLinx on Discord.
Models (especially open-source like qwen3.5-plus) may send non-int values
for the limit parameter — None (JSON null), string, or even a type object.
This caused TypeError: '<=' not supported between instances of 'int' and
'type' when the value reached min()/comparison operations.
Changes:
- Add defensive int coercion at session_search() entry with fallback to 3
- Clamp limit to [1, 5] range (was only capped at 5, not floored)
- Add tests for None, type object, string, negative, and zero limit values
Reported by community user ludoSifu via Discord.
Python's json.dumps() defaults to ensure_ascii=True, escaping non-ASCII
characters to \uXXXX sequences. For CJK characters this inflates
token count 3-4x — a single Chinese character like '中' becomes
'\u4e2d' (6 chars vs 3 bytes, ~6 tokens vs ~1 token).
Since MCP tool results feed directly into the model's conversation
context, this silently multiplied API costs for Chinese, Japanese,
and Korean users.
Fix: add ensure_ascii=False to all 20 json.dumps calls in mcp_tool.py.
Raw UTF-8 is valid JSON per RFC 8259 and all downstream consumers
(LLM APIs, display) handle it correctly.
Closes#10234
Multiple gaps in activity tracking could cause the gateway's inactivity
timeout to fire while the agent is actively working:
1. Streaming wait loop had no periodic heartbeat — the outer thread only
touched activity when the stale-stream detector fired (180-300s), and
for local providers (Ollama) the stale timeout was infinity, meaning
zero heartbeats. Now touches activity every 30s.
2. Concurrent tool execution never set the activity callback on worker
threads (threading.local invisible across threads) and never set
_current_tool. Workers now set the callback, and the concurrent wait
uses a polling loop with 30s heartbeats.
3. Modal backend's execute() override had its own polling loop without
any activity callback. Now matches _wait_for_process cadence (10s).
- Populate watcher_* routing fields for watch-only processes (not just
notify_on_complete), so watch-pattern events carry direct metadata
instead of relying solely on session_key parsing fallback
- Extract _parse_session_key() helper to dedupe session key parsing
at two call sites in gateway/run.py
- Add negative test proving cross-thread leakage doesn't happen
- Add edge-case tests for _build_process_event_source returning None
(empty evt, invalid platform, short session_key)
- Add unit tests for _parse_session_key helper
Tool schema descriptions and tool return values contained hardcoded
~/.hermes paths that the model sees and uses. When HERMES_HOME is set
to a custom path (Docker containers, profiles), the agent would still
reference ~/.hermes — looking at the wrong directory.
Fixes 6 locations across 5 files:
- tools/tts_tool.py: output_path schema description
- tools/cronjob_tools.py: script path schema description
- tools/skill_manager_tool.py: skill_manage schema description
- tools/skills_tool.py: two tool return messages
- agent/skill_commands.py: skill config injection text
All now use display_hermes_home() which resolves to the actual
HERMES_HOME path (e.g. /opt/data for Docker, ~/.hermes/profiles/X
for profiles, ~/.hermes for default).
Reported by: Sandeep Narahari (PrithviDevs)
- Fix file handle closed before POST: nest session.post() inside
the 'with open()' block so aiohttp can read the file during upload
- Update warning text to include weixin (also supports media delivery)
- Add 8 unit tests covering: text+media, media-only, missing files,
upload failures, multiple files, and _send_to_platform routing
Previously send_message only supported media delivery for Telegram.
Discord users received a warning that media was omitted.
- Add media_files parameter to _send_discord()
- Upload media via Discord multipart/form-data API (files[0] field)
- Handle Discord in _send_to_platform() same way as Telegram block
- Remove Discord from generic chunk loop (now handled above)
- Update error/warning strings to mention telegram and discord
* fix(gateway): suppress duplicate replies on interrupt and streaming flood control
Three fixes for the duplicate reply bug affecting all gateway platforms:
1. base.py: Suppress stale response when the session was interrupted by a
new message that hasn't been consumed yet. Checks both interrupt_event
and _pending_messages to avoid false positives. (#8221, #2483)
2. run.py (return path): Remove response_previewed guard from already_sent
check. Stream consumer's already_sent alone is authoritative — if
content was delivered via streaming, the duplicate send must be
suppressed regardless of the agent's response_previewed flag. (#8375)
3. run.py (queued-message path): Same fix — already_sent without
response_previewed now correctly marks the first response as already
streamed, preventing re-send before processing the queued message.
The response_previewed field is still produced by the agent (run_agent.py)
but is no longer required as a gate for duplicate suppression. The stream
consumer's already_sent flag is the delivery-level truth about what the
user actually saw.
Concepts from PR #8380 (konsisumer). Closes#8375, #8221, #2483.
* fix(cron): include job_id in delivery and guide models on removal workflow
Users reported cron reminders keep firing after asking the agent to stop.
Root cause: the conversational agent didn't know the job_id (not in delivery)
and models don't reliably do the list→remove two-step without guidance.
1. Include job_id in the cron delivery wrapper so users and agents can
reference it when requesting removal.
2. Replace confusing footer ('The agent cannot see this message') with
actionable guidance ('To stop or manage this job, send me a new
message').
3. Add explicit list→remove guidance in the cronjob tool schema so models
know to list first and never guess job IDs.
Matrix room IDs contain ! and : which must be percent-encoded in URI
path segments per the Matrix C-S spec. Without encoding, some
homeservers reject the PUT request.
Also adds 'matrix:!roomid:server.org' and 'matrix:@user:server.org'
to the tool schema examples so models know the correct target format.
`_parse_target_ref` has explicit-reference branches for Telegram, Feishu,
and numeric IDs, but none for Matrix. As a result, callers of
`send_message(target="matrix:!roomid:server")` or
`send_message(target="matrix:@user:server")` fall through to
`(None, None, False)` and the tool errors out with a resolution failure —
even though a raw Matrix room ID or MXID is the most unambiguous possible
target.
Three-line fix: recognize `!…` as a room ID and `@…` as a user MXID when
platform is `matrix`, and return them as explicit targets. Alias-based
targets (`#…`) continue to go through the normal resolve path.
- find_docker() now checks HERMES_DOCKER_BINARY env var first, then
docker on PATH, then podman on PATH, then macOS known locations
- Entrypoint respects HERMES_HOME env var (was hardcoded to /opt/data)
- Entrypoint uses groupmod -o to tolerate non-unique GIDs (fixes macOS
GID 20 conflict with Debian's dialout group)
- Entrypoint makes chown best-effort so rootless Podman continues
instead of failing with 'Operation not permitted'
- 5 new tests covering env var override, podman fallback, precedence
Based on work by alanjds (PR #3996) and malaiwah (PR #8115).
Closes#4084.
The original tree-wide ast.walk() would match registry.register() calls
inside functions too. Restrict to top-level ast.Expr statements so helper
modules that call registry.register() inside a function are never picked
up as tool modules.
Refactor browser tool PATH construction to include Termux directories
(/data/data/com.termux/files/usr/bin, /data/data/com.termux/files/usr/sbin)
so agent-browser and npx are discoverable on Android/Termux.
Extracts _browser_candidate_path_dirs() and _merge_browser_path() helpers
to centralize PATH construction shared between _find_agent_browser() and
_run_browser_command(), replacing duplicated inline logic.
Also fixes os.pathsep usage (was hardcoded ':') for cross-platform correctness.
Cherry-picked from PR #9846.
Add dangerous command patterns that require approval when the agent
tries to run gateway lifecycle commands via the terminal tool:
- hermes gateway stop/restart — kills all running agents mid-work
- hermes update — pulls code and restarts the gateway
- systemctl restart/stop (with optional flags like --user)
These patterns fire the approval prompt so the user must explicitly
approve before the agent can kill its own gateway process. In YOLO
mode, the commands run without approval (by design — YOLO means the
user accepts all risks).
Also fixes the existing systemctl pattern to handle flags between
the command and action (e.g. 'systemctl --user restart' was previously
undetected because the regex expected the action immediately after
'systemctl').
Root cause: issue #6666 reported agents running 'hermes gateway
restart' via terminal, killing the gateway process mid-agent-loop.
The user sees the agent suddenly stop responding with no explanation.
Combined with the SIGTERM auto-recovery from PR #9875, the gateway
now both prevents accidental self-destruction AND recovers if it
happens anyway.
Test plan:
- Updated test_systemctl_restart_not_flagged → test_systemctl_restart_flagged
- All 119 approval tests pass
- E2E verified: hermes gateway restart, hermes update, systemctl
--user restart all detected; hermes gateway status, systemctl
status remain safe
Add ctx.register_skill() API so plugins can ship SKILL.md files under
a 'plugin:skill' namespace, preventing name collisions with built-in
Hermes skills. skill_view() detects the ':' separator and routes to
the plugin registry while bare names continue through the existing
flat-tree scan unchanged.
Key additions:
- agent/skill_utils: parse_qualified_name(), is_valid_namespace()
- hermes_cli/plugins: PluginContext.register_skill(), PluginManager
skill registry (find/list/remove)
- tools/skills_tool: qualified name dispatch in skill_view(),
_serve_plugin_skill() with full guards (disabled, platform,
injection scan), bundle context banner with sibling listing,
stale registry self-heal
- Hoisted _INJECTION_PATTERNS to module level (dedup)
- Updated skill_view schema description
Based on PR #9334 by N0nb0at. Lean P1 salvage — omits autogen shim
(P2) for a simpler first merge.
Closes#8422
- Fix _camofox_eval() endpoint: /tabs/{id}/eval → /tabs/{id}/evaluate
(correct Camofox REST API path)
- Add required userId field to JS eval request body (all other Camofox
endpoints already include it)
- Update npm package from @askjo/camoufox-browser ^1.0.0 to
@askjo/camofox-browser ^1.5.2 (upstream package was renamed)
- Update tools_config.py post-setup to reference new package directory
and npx command
- Bump Node engine requirement from >=18 to >=20 (required by
camoufox-js dependency in camofox-browser v1.5.2)
- Regenerate package-lock.json
Fixes issues reported in PRs #9472, #8267, #7208 (stale).
Match cron/scheduler.py pattern — only attempt msvcrt import when
fcntl is unavailable. Pre-declare msvcrt = None at module level so
_file_lock() references don't NameError on Linux.
Production fixes:
- Add clear_session_context() to hermes_logging.py (fixes 48 teardown errors)
- Add clear_session() to tools/approval.py (fixes 9 setup errors)
- Add SyncError M_UNKNOWN_TOKEN check to Matrix _sync_loop (bug fix)
- Fall back to inline api_key in named custom providers when key_env
is absent (runtime_provider.py)
Test fixes:
- test_memory_user_id: use builtin+external provider pair, fix honcho
peer_name override test to match production behavior
- test_display_config: remove TestHelpers for non-existent functions
- test_auxiliary_client: fix OAuth tokens to match _is_oauth_token
patterns, replace get_vision_auxiliary_client with resolve_vision_provider_client
- test_cli_interrupt_subagent: add missing _execution_thread_id attr
- test_compress_focus: add model/provider/api_key/base_url/api_mode
to mock compressor
- test_auth_provider_gate: add autouse fixture to clean Anthropic env
vars that leak from CI secrets
- test_opencode_go_in_model_list: accept both 'built-in' and 'hermes'
source (models.dev API unavailable in CI)
- test_email: verify email Platform enum membership instead of source
inspection (build_channel_directory now uses dynamic enum loop)
- test_feishu: add bot_added/bot_deleted handler mocks to _Builder
- test_ws_auth_retry: add AsyncMock for sync_store.get_next_batch,
add _pending_megolm and _joined_rooms to Matrix adapter mocks
- test_restart_drain: monkeypatch-delete INVOCATION_ID (systemd sets
this in CI, changing the restart call signature)
- test_session_hygiene: add user_id to SessionSource
- test_session_env: use relative baseline for contextvar clear check
(pytest-xdist workers share context)
Improvements from our earlier #8269 salvage work applied to #7616:
- Platform token lock: acquire_scoped_lock/release_scoped_lock prevents
two profiles from double-connecting the same QQ bot simultaneously
- Send retry with exponential backoff (3 attempts, 1s/2s/4s) with
permanent vs transient error classification (matches Telegram pattern)
- Proper long-message splitting via truncate_message() instead of
hard-truncating at MAX_MESSAGE_LENGTH (preserves code blocks, adds 1/N)
- REST-based one-shot send in send_message_tool — uses QQ Bot REST API
directly with httpx instead of creating a full WebSocket adapter per
message (fixes the connect→send race condition)
- Use shared strip_markdown() from helpers.py instead of 15 lines of
inline regex with import-inside-method (DRY, same as BlueBubbles/SMS)
- format_message() now wired into send() pipeline
- Rename platform from 'qq' to 'qqbot' across all integration points
(Platform enum, toolset, config keys, import paths, file rename qq.py → qqbot.py)
- Add PLATFORM_HINTS for QQBot in prompt_builder (QQ supports markdown)
- Set SUPPORTS_MESSAGE_EDITING = False to skip streaming on QQ
(prevents duplicate messages from non-editable partial + final sends)
- Add _send_qqbot() standalone send function for cron/send_message tool
- Add interactive _setup_qq() wizard in hermes_cli/setup.py
- Restore missing _setup_signal/email/sms/dingtalk/feishu/wecom/wecom_callback
functions that were lost during the original merge
Three improvements to file search based on user feedback:
1. Fuzzy @ completions (commands.py):
- Bare @query now does project-wide fuzzy file search instead of
prefix-only directory listing
- Uses rg --files with 5-second cache for responsive completions
- Scoring: exact name (100) > prefix (80) > substring (60) >
path contains (40) > subsequence with boundary bonus (35/25)
- Bare @ with no query shows recently modified files first
2. Mtime-sorted file search (file_operations.py):
- _search_files_rg now uses --sortr=modified (rg 13+) to surface
recently edited files first
- Falls back to unsorted on older rg versions
3. Improved file-not-found suggestions (file_operations.py):
- Replaced crude character-set overlap with ranked scoring:
same basename (90) > prefix (70) > substring (60) >
reverse substring (40) > same extension (30)
- search_files path-not-found now suggests similar directories
from the parent
Cherry-picked from PR #9177 by @haileymarshall.
Adds a fitness and nutrition skill for gym-goers and health-conscious users:
- Exercise search via wger API (690+ exercises, free, no auth)
- Nutrition lookup via USDA FoodData Central (380K+ foods, DEMO_KEY fallback)
- Offline body composition calculators (BMI, TDEE, 1RM, macros, body fat %)
- Pure stdlib Python, no pip dependencies
Changes from original PR:
- Moved from skills/ to optional-skills/health/ (correct location)
- Fixed BMR formula in FORMULAS.md (removed confusing -5+10, now just +5)
- Fixed author attribution to match PR submitter
- Marked USDA_API_KEY as optional (DEMO_KEY works without signup)
Also adds optional env var support to the skill readiness checker:
- New 'optional: true' field in required_environment_variables entries
- Optional vars are preserved in metadata but don't block skill readiness
- Optional vars skip the CLI capture prompt flow
- Skills with only optional missing vars show as 'available' not 'setup_needed'
Port two improvements inspired by Kilo-Org/kilocode analysis:
1. Error classifier: add context overflow patterns for vLLM, Ollama,
and llama.cpp/llama-server. These local inference servers return
different error formats than cloud providers (e.g., 'exceeds the
max_model_len', 'context length exceeded', 'slot context'). Without
these patterns, context overflow errors from local servers are
misclassified as format errors, causing infinite retries instead
of triggering compression.
2. MCP initial connection retry: previously, if the very first
connection attempt to an MCP server failed (e.g., transient DNS
blip at startup), the server was permanently marked as failed with
no retry. Post-connect reconnection had 5 retries with exponential
backoff, but initial connection had zero. Now initial connections
retry up to 3 times with backoff before giving up, matching the
resilience of post-connect reconnection.
(Inspired by Kilo Code's MCP server disappearing fix in v1.3.3)
Tests: 6 new error classifier tests, 4 new MCP retry tests, 1
updated existing test. All 276 affected tests pass.
On macOS, /etc is a symlink to /private/etc, so os.path.realpath()
resolves /etc/hosts to /private/etc/hosts. The sensitive path check
only matched /etc/ prefixes against the resolved path, allowing
writes to system files on macOS.
- Add /private/etc/ and /private/var/ to _SENSITIVE_PATH_PREFIXES
- Check both realpath-resolved and normpath-normalized paths
- Add regression tests for macOS symlink bypass
Closes#8734
Co-authored-by: ElhamDevelopmentStudio (PR #8829)
Three-tier match strategy for _truncate_around_matches():
1. Full-phrase search (exact query string positions)
2. Proximity co-occurrence (all terms within 200 chars)
3. Individual terms (fallback, preserves existing behavior)
Sliding window picks the start offset covering the most matches.
Moved inline import re to module level.
Co-authored-by: Al Sayed Hoota <78100282+AlsayedHoota@users.noreply.github.com>
The terminal and execute_code tool schemas unconditionally mentioned
'cloud sandboxes' in their descriptions sent to the model. This caused
agents running on local backends to believe they were in a sandboxed
environment, refusing networking tasks and other operations. Worse,
agents sometimes saved this false belief to persistent memory, making
it persist across sessions.
Reported by multiple users (XLion, 林泽).
Port from nearai/ironclaw#2304: Telegram's 4096 character limit is
measured in UTF-16 code units, not Unicode codepoints. Characters
outside the Basic Multilingual Plane (emoji like 😀, CJK Extension B,
musical symbols) are surrogate pairs: 1 Python char but 2 UTF-16 units.
Previously, truncate_message() used Python's len() which counts
codepoints. This could produce chunks exceeding Telegram's actual limit
when messages contain many astral-plane characters.
Changes:
- Add utf16_len() helper and _prefix_within_utf16_limit() for
UTF-16-aware string measurement and truncation
- Add _custom_unit_to_cp() binary-search helper that maps a custom-unit
budget to the largest safe codepoint slice position
- Update truncate_message() to accept optional len_fn parameter
- Telegram adapter now passes len_fn=utf16_len when splitting messages
- Fix fallback truncation in Telegram error handler to use
_prefix_within_utf16_limit instead of codepoint slicing
- Update send_message_tool.py to use utf16_len for Telegram platform
- Add comprehensive tests: utf16_len, _prefix_within_utf16_limit,
truncate_message with len_fn (emoji splitting, content preservation,
code block handling)
- Update mock lambdas in reply_mode tests to accept **kw for len_fn
Add a CI-built skills index served from the docs site. The index is
crawled daily by GitHub Actions, resolves all GitHub paths upfront, and
is cached locally by the client. When the index is available:
- Search uses the cached index (0 GitHub API calls, was 23+)
- Install uses resolved paths from index (6 API calls for file
downloads only, was 31-45 for discovery + downloads)
Total: 68 → 6 GitHub API calls for a typical search + install flow.
Unauthenticated users (60 req/hr) can now search and install without
hitting rate limits.
Components:
- scripts/build_skills_index.py: Crawl all sources (skills.sh, GitHub
taps, official, clawhub, lobehub), batch-resolve GitHub paths via
tree API, output JSON index
- tools/skills_hub.py: HermesIndexSource class — search/fetch/inspect
backed by the index, with lazy GitHubSource for file downloads
- parallel_search_sources() skips external API sources when index is
available (0 GitHub calls for search)
- .github/workflows/skills-index.yml: twice-daily CI build + deploy
- .github/workflows/deploy-site.yml: also builds index during docs deploy
Graceful degradation: when the index is unavailable (first run, network
down, stale), all methods return empty/None and downstream sources
handle the request via direct API as before.
Skills.sh installs hit the GitHub API 45 times per install because the
same repo tree was fetched 6 times redundantly. Combined with search
(23 API calls), this totals 68 — exceeding the unauthenticated rate
limit of 60 req/hr, causing 'Could not fetch' errors for users without
a GITHUB_TOKEN.
Changes:
- Add _get_repo_tree() cache to GitHubSource — repo info + recursive
tree fetched once per repo per source instance, eliminating 10
redundant API calls (6 tree + 4 candidate 404s)
- _download_directory_via_tree returns {} (not None) when cached tree
shows path doesn't exist, skipping unnecessary Contents API fallback
- _check_rate_limit_response() detects exhausted quota and sets
is_rate_limited flag
- do_install() shows actionable hint when rate limited: set
GITHUB_TOKEN or install gh CLI
Before: 45 API calls per install (68 total with search)
After: 31 API calls per install (54 total with search — under 60/hr)
Reported by community user from Vietnam (no GitHub auth configured).
Centralize container detection in hermes_constants.is_container() with
process-lifetime caching, matching existing is_wsl()/is_termux() patterns.
Dedup _is_inside_container() in config.py to delegate to the new function.
Add _run_systemctl() wrapper that converts FileNotFoundError to RuntimeError
for defense-in-depth — all 10 bare subprocess.run(_systemctl_cmd(...)) call
sites now route through it.
Make supports_systemd_services() return False in containers and when
systemctl binary is absent (shutil.which check).
Add Docker-specific guidance in gateway_command() for install/uninstall/start
subcommands — exit 0 with helpful instructions instead of crashing.
Make 'hermes status' show 'Manager: docker (foreground)' and 'hermes dump'
show 'running (docker, pid N)' inside containers.
Fix setup_gateway() to use supports_systemd instead of _is_linux for all
systemd-related branches, and show Docker restart policy instructions in
containers.
Replace inline /.dockerenv check in voice_mode.py with is_container().
Fixes#7420
Co-authored-by: teknium1 <teknium1@users.noreply.github.com>
* fix: list all available toolsets in delegate_task schema description
The delegate_task tool's toolsets parameter description only mentioned
'terminal', 'file', and 'web' as examples. Models (especially smaller
ones like Gemma) would substitute 'web' for 'browser' because they
didn't know 'browser' was a valid option.
Now dynamically builds the toolset list from the TOOLSETS dict at import
time, excluding blocked, composite, and platform-specific toolsets.
Auto-updates when new toolsets are added.
Reported by jeffutter on Discord.
* chore: exclude moa and rl from delegate_task toolset list
When the agent calls process(action='wait') or process(action='poll')
and gets the exited status, the completion_queue notification is
redundant — the agent already has the output from the tool return.
Previously, the drain loops in CLI and gateway would still inject
the [SYSTEM: Background process completed] message, causing the
agent to receive the same information twice.
Fix: track session IDs in _completion_consumed set when wait/poll/log
returns an exited process. Drain loops in cli.py and gateway watcher
skip completion events for consumed sessions. Watch pattern events
are never suppressed (they have independent semantics).
Adds 4 tests covering wait/poll/log marking and running-process
negative case.
Rewrite the cronjob tool's 'deliver' parameter description to strongly
guide models toward omitting the parameter (which auto-detects origin
including thread/topic). The previous description listed all platform
names equally, inviting models to construct explicit targets like
'telegram:<chat_id>' which silently drops the thread_id.
New description:
- Leads with 'Omit this parameter' as the recommended path
- Explicitly warns that platform:chat_id without :thread_id loses topics
- Removes the long flat list of platform names that invited construction
Also adds diagnostic logging at two key points:
- _origin_from_env(): logs when thread_id is captured during job creation
- _deliver_result(): warns when origin has thread_id but delivery target
lost it; logs at debug when delivering to a specific thread
Helps diagnose user-reported issue where cron responses from Telegram
topics are delivered to the main chat instead of the originating topic.
* perf(ssh,modal): bulk file sync via tar pipe and tar/base64 archive
SSH: symlink-staging + tar -ch piped over SSH in a single TCP stream.
Eliminates per-file scp round-trips. Handles timeout (kills both
processes), SSH Popen failure (kills tar), and tar create failure.
Modal: in-memory gzipped tar archive, base64-encoded, decoded+extracted
in one exec call. Checks exit code and raises on failure.
Both backends use shared helpers extracted into file_sync.py:
- quoted_mkdir_command() — mirrors existing quoted_rm_command()
- unique_parent_dirs() — deduplicates parent dirs from file pairs
Migrates _ensure_remote_dirs to use the new helpers.
28 new tests (21 SSH + 7 Modal), all passing.
Closes#7465Closes#7467
* fix(modal): pipe stdin to avoid ARG_MAX, clean up review findings
- Modal bulk upload: stream base64 payload through proc.stdin in 1MB
chunks instead of embedding in command string (Modal SDK enforces
64KB ARG_MAX_BYTES — typical payloads are ~4.3MB)
- Modal single-file upload: same stdin fix, add exit code checking
- Remove what-narrating comments in ssh.py and modal.py (keep WHY
comments: symlink staging rationale, SIGPIPE, deadlock avoidance)
- Remove unnecessary `sandbox = self._sandbox` alias in modal bulk
- Daytona: use shared helpers (unique_parent_dirs, quoted_mkdir_command)
instead of inlined duplicates
---------
Co-authored-by: kshitijk4poor <82637225+kshitijk4poor@users.noreply.github.com>
The check_interval parameter on terminal_tool sent periodic output
updates to the gateway chat, but these were display-only — the agent
couldn't see or act on them. This added schema bloat and introduced
a bug where notify_on_complete=True was silently dropped when
check_interval was also set (the not-check_interval guard skipped
fast-watcher registration, and the check_interval watcher dict
was missing the notify_on_complete key).
Removing check_interval entirely:
- Eliminates the notify_on_complete interaction bug
- Reduces tool schema size (one fewer parameter for the model)
- Simplifies the watcher registration path
- notify_on_complete (agent wake-on-completion) still works
- watch_patterns (output alerting) still works
- process(action='poll') covers manual status checking
Closes#7947 (root cause eliminated rather than patched).
Deduplicate todo items by ID before writing to the store, keeping the
last occurrence. Prevents ghost entries when the model sends duplicate
IDs in a single write() call, which corrupts subsequent merge operations.
Co-authored-by: WAXLYY <WAXLYY@users.noreply.github.com>
Three root causes of the 'agent stops mid-task' gateway bug:
1. Compression threshold floor (64K tokens minimum)
- The 50% threshold on a 100K-context model fired at 50K tokens,
causing premature compression that made models lose track of
multi-step plans. Now threshold_tokens = max(50% * context, 64K).
- Models with <64K context are rejected at startup with a clear error.
2. Budget warning removal — grace call instead
- Removed the 70%/90% iteration budget warnings entirely. These
injected '[BUDGET WARNING: Provide your final response NOW]' into
tool results, causing models to abandon complex tasks prematurely.
- Now: no warnings during normal execution. When the budget is
actually exhausted (90/90), inject a user message asking the model
to summarise, allow one grace API call, and only then fall back
to _handle_max_iterations.
3. Activity touches during long terminal execution
- _wait_for_process polls every 0.2s but never reported activity.
The gateway's inactivity timeout (default 1800s) would fire during
long-running commands that appeared 'idle.'
- Now: thread-local activity callback fires every 10s during the
poll loop, keeping the gateway's activity tracker alive.
- Agent wires _touch_activity into the callback before each tool call.
Also: docs update noting 64K minimum context requirement.
Closes#7915 (root cause was agent-loop termination, not Weixin delivery limits).
Complete the contextvars migration by adding HERMES_SESSION_KEY to the
unified _VAR_MAP in session_context.py. Without this, concurrent gateway
handlers race on os.environ["HERMES_SESSION_KEY"].
- Add _SESSION_KEY ContextVar to _VAR_MAP, set_session_vars(), clear_session_vars()
- Wire session_key through _set_session_env() from SessionContext
- Replace os.getenv fallback in tools/approval.py with get_session_env()
(function-level import to avoid cross-layer coupling)
- Keep os.environ set as CLI/cron fallback
Cherry-picked from PR #7878 by 0xbyt4.
Add a second WeCom integration mode for regular enterprise self-built
applications. Unlike the existing bot/websocket adapter (wecom.py),
this handles WeCom's standard callback flow: WeCom POSTs encrypted XML
to an HTTP endpoint, the adapter decrypts, queues for the agent, and
immediately acknowledges. The agent's reply is delivered proactively
via the message/send API.
Key design choice: always acknowledge immediately and use proactive
send — agent sessions take 3-30 minutes, so the 5-second inline reply
window is never useful. The original PR's Future/pending-reply
machinery was removed in favour of this simpler architecture.
Features:
- AES-CBC encrypt/decrypt (BizMsgCrypt-compatible)
- Multi-app routing scoped by corp_id:user_id
- Legacy bare user_id fallback for backward compat
- Access-token management with auto-refresh
- WECOM_CALLBACK_* env var overrides
- Port-in-use pre-check before binding
- Health endpoint at /health
Salvaged from PR #7774 by @chqchshj. Simplified by removing the
inline reply Future system and fixing: secrets.choice for nonce
generation, immediate plain-text acknowledgment (not encrypted XML
containing 'success'), and initial token refresh error handling.
Adds _normalize_path() helper that calls expanduser().resolve() to
properly handle tilde paths (e.g. ~/.hermes, ~/.config). Previously
Path.resolve() alone treated ~ as a literal directory name, producing
invalid paths like /root/~/.hermes.
Also improves _run_git() error handling to distinguish missing working
directories from missing git executable, and adds pre-flight directory
validation.
Cherry-picked from PR #7898 by faishal882.
Fixes#7807
_write_to_sandbox interpolated storage_dir and remote_path directly into
a shell command passed to env.execute(). Paths containing shell
metacharacters (spaces, semicolons, $(), backticks) could trigger
arbitrary command execution inside the sandbox.
Fix: wrap both paths with shlex.quote(). Clean paths (alphanumeric +
slashes/hyphens/dots) are left unmodified by shlex.quote, so existing
behavior is unchanged. Paths with unsafe characters get single-quoted.
Tests added for spaces, $(command) substitution, and semicolon injection.
This commit addresses a security vulnerability where unsanitized user inputs for commit_hash and file_path were passed directly to git commands in CheckpointManager.restore() and diff(). It validates commit hashes to be strictly hexadecimal characters without leading dashes (preventing flag injection like '--patch') and enforces file paths to stay within the working directory via root resolution. Regression tests test_restore_rejects_argument_injection, test_restore_rejects_invalid_hex_chars, and test_restore_rejects_path_traversal were added.
The interrupt mechanism in tools/interrupt.py used a process-global
threading.Event. In the gateway, multiple agents run concurrently in
the same process via run_in_executor. When any agent was interrupted
(user sends a follow-up message), the global flag killed ALL agents'
running tools — terminal commands, browser ops, web requests — across
all sessions.
Changes:
- tools/interrupt.py: Replace single threading.Event with a set of
interrupted thread IDs. set_interrupt() targets a specific thread;
is_interrupted() checks the current thread. Includes a backward-
compatible _ThreadAwareEventProxy for legacy _interrupt_event usage.
- run_agent.py: Store execution thread ID at start of run_conversation().
interrupt() and clear_interrupt() pass it to set_interrupt() so only
this agent's thread is affected.
- tools/code_execution_tool.py: Use is_interrupted() instead of
directly checking _interrupt_event.is_set().
- tools/process_registry.py: Same — use is_interrupted().
- tests: Update interrupt tests for per-thread semantics. Add new
TestPerThreadInterruptIsolation with two tests verifying cross-thread
isolation.
When a Python process exits uncleanly (SIGKILL, crash, gateway restart
via hermes update), in-memory _active_sessions tracking is lost but the
agent-browser node daemons and their Chromium child processes keep
running indefinitely. On a long-running system this causes unbounded
memory growth — 24 orphaned sessions consumed 7.6 GB on a production
machine over 9 days.
Add _reap_orphaned_browser_sessions() which scans the tmp directory for
agent-browser-{h_*,cdp_*} socket dirs on cleanup thread startup. For
each dir not tracked by the current process, reads the daemon PID file
and sends SIGTERM if the daemon is still alive. Handles edge cases:
dead PIDs, corrupt PID files, permission errors, foreign processes.
The reaper runs once on thread startup (not every 30s) to avoid races
with sessions being actively created by concurrent agents.
Background process watchers (notify_on_complete, check_interval) created
synthetic SessionSource objects without user_id/user_name. While the
internal=True bypass (1d8d4f28) prevented false pairing for agent-
generated notifications, the missing identity caused:
- Garbage entries in pairing rate limiters (discord:None, telegram:None)
- 'User None' in approval messages and logs
- No user identity available for future code paths that need it
Additionally, platform messages arriving without from_user (Telegram
service messages, channel forwards, anonymous admin actions) could still
trigger false pairing because they are not internal events.
Fix:
1. Propagate user_id/user_name through the full watcher chain:
session_context.py → gateway/run.py → terminal_tool.py →
process_registry.py (including checkpoint persistence/recovery)
2. Add None user_id guard in _handle_message() — silently drop
non-internal messages with no user identity instead of triggering
the pairing flow.
Salvaged from PRs #7664 (kagura-agent, ContextVar approach),
#6540 (MestreY0d4-Uninter, tests), and #7709 (guang384, None guard).
Closes#6341, #6485, #7643
Relates to #6516, #7392
Independent halving of width and height caused aspect ratio distortion
for extreme dimensions (e.g. 8000x200 panoramas). When one axis hit the
64px floor, the other kept shrinking — collapsing the ratio toward 1:1.
Use proportional scaling instead: when either dimension hits the floor,
derive the effective scale factor and apply it to both axes.
Add tests for extreme panorama (8000x200) and tall narrow (200x6000)
images to verify aspect ratio preservation.
Cherry-picked from PR #7749 by kshitijk4poor with modifications:
- Raise hard image limit from 5 MB to 20 MB (matches most restrictive provider)
- Send images at full resolution first; only auto-resize to 5 MB on API failure
- Add _is_image_size_error() helper to detect size-related API rejections
- Auto-resize uses Pillow (soft dep) with progressive downscale + JPEG quality reduction
- Fix get_model_capabilities() to check modalities.input for vision support
- Increase default vision timeout from 30s to 120s (matches hardcoded fallback intent)
- Applied retry-with-resize to both vision_analyze_tool and browser_vision
Closes#7740
* feat: add watch_patterns to background processes for output monitoring
Adds a new 'watch_patterns' parameter to terminal(background=true) that
lets the agent specify strings to watch for in process output. When a
matching line appears, a notification is queued and injected as a
synthetic message — triggering a new agent turn, similar to
notify_on_complete but mid-process.
Implementation:
- ProcessSession gets watch_patterns field + rate-limit state
- _check_watch_patterns() in ProcessRegistry scans new output chunks
from all three reader threads (local, PTY, env-poller)
- Rate limited: max 8 notifications per 10s window
- Sustained overload (45s) permanently disables watching for that process
- watch_queue alongside completion_queue, same consumption pattern
- CLI drains watch_queue in both idle loop and post-turn drain
- Gateway drains after agent runs via _inject_watch_notification()
- Checkpoint persistence + crash recovery includes watch_patterns
- Blocked in execute_code sandbox (like other bg params)
- 20 new tests covering matching, rate limiting, overload kill,
checkpoint persistence, schema, and handler passthrough
Usage:
terminal(
command='npm run dev',
background=true,
watch_patterns=['ERROR', 'WARN', 'listening on port']
)
* refactor: merge watch_queue into completion_queue
Unified queue with 'type' field distinguishing 'completion',
'watch_match', and 'watch_disabled' events. Extracted
_format_process_notification() in CLI and gateway to handle
all event types in a single drain loop. Removes duplication
across both CLI drain sites and the gateway.
Three fixes for vision_analyze returning cryptic 400 "Invalid request data":
1. Pre-flight base64 size check — base64 inflates data ~33%, so a 3.8 MB
file exceeds the 5 MB API limit. Reject early with a clear message
instead of letting the provider return a generic 400.
2. Handle file:// URIs — strip the scheme and resolve as a local path.
Previously file:///path/to/image.png fell through to the "invalid
image source" error since it matched neither is_file() nor http(s).
3. Separate invalid_request errors from "does not support vision" errors
so the user gets actionable guidance (resize/compress/retry) instead
of a misleading "model does not support vision" message.
Closes#6677
vision_tools.py: _download_image() loads the full HTTP response body into
memory via response.content (line 190) with no Content-Length check and no
max file size limit. An attacker-hosted multi-gigabyte file causes OOM.
Add a 50 MB hard cap: check Content-Length header before download, and
verify actual body size before writing to disk.
hermes_parser.py: tc_data["name"] at line 57 raises KeyError when the LLM
outputs a tool call JSON without a "name" field. The outer except catches
it silently, causing the entire tool call to be lost with zero diagnostics.
Add "name" field validation before constructing the ChatCompletionMessage.
mistral_parser.py: tc["name"] at line 101 has the same KeyError issue in
the pre-v11 format path. The fallback decoder (line 112) already checks
"name" correctly, but the primary path does not. Add validation to match.
Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
process_registry.py: _reader_loop() has process.wait() after the try-except
block (line 380). If the reader thread crashes with an unexpected exception
(e.g. MemoryError, KeyboardInterrupt), control exits the except handler but
skips wait() — leaving the child as a zombie process. Move wait() and the
cleanup into a finally block so the child is always reaped.
cron/scheduler.py: _run_job_script() only redacts secrets in stdout on the
SUCCESS path (line 417-421). When a cron script fails (non-zero exit), both
stdout and stderr are returned WITHOUT redaction (lines 407-413). A script
that accidentally prints an API key to stderr during a failure would leak it
into the LLM context. Move redaction before the success/failure branch so
both paths benefit.
skill_commands.py: _build_skill_message() enumerates supporting files using
rglob("*") but only checks is_file() (line 171) without filtering symlinks.
PR #6693 added symlink protection to scan_skill_commands() but missed this
function. A malicious skill can create symlinks in references/ pointing to
arbitrary files, exposing their paths (and potentially content via skill_view)
to the LLM. Add is_symlink() check to match the guard in scan_skill_commands.
Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
_discover_bundled_skills() used the directory name to identify skills,
but skills_tool.py and skills_hub.py use the `name:` field from SKILL.md
frontmatter. This mismatch caused 9 builtin skills whose directory name
differs from their SKILL.md name to be written to .bundled_manifest
under the wrong key, so `hermes skills list` showed them as "local"
instead of "builtin".
Read the frontmatter name field (with directory-name fallback) so the
manifest keys match what the rest of the codebase expects.
Closes#6835
- Remove unreachable `if not content_sample` branch inside the truthy
`if content_sample` block in `_is_likely_binary()` (dead code that
could never execute).
- Replace `linter_cmd.format(file=...)` with `linter_cmd.replace("{file}", ...)`
in `_check_lint()` so file paths containing curly braces (e.g.
`src/{test}.py`) no longer raise KeyError/ValueError.
- Add 16 unit tests covering both fixes and edge cases.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Add delegation.reasoning_effort config key so subagents can run at a
different thinking level than the parent agent. When set, overrides
the parent's reasoning_config; when empty, inherits as before.
Valid values: xhigh, high, medium, low, minimal, none (disables thinking).
Config path: delegation.reasoning_effort in config.yaml
Files changed:
- tools/delegate_tool.py: resolve override in _build_child_agent
- hermes_cli/config.py: add reasoning_effort to DEFAULT_CONFIG
- tests/tools/test_delegate.py: 4 new tests covering all cases
warnings.warn() is suppressed/invisible when running as a gateway
or agent. Switch to logger.warning() so the disk cap message
actually appears in logs.
Fixes#7362 (item 3).
FileSyncManager now accepts an optional bulk_upload_fn callback.
When provided, all changed files are uploaded in one call instead
of iterating one-by-one with individual HTTP POSTs.
DaytonaEnvironment wires this to sandbox.fs.upload_files() which
batches everything into a single multipart POST — ~580 files goes
from ~5 min to <2s on init.
Parent directories are pre-created in one mkdir -p call.
Fixes#7362 (item 1).
When two gateway messages arrived concurrently, _set_session_env wrote
HERMES_SESSION_PLATFORM/CHAT_ID/CHAT_NAME/THREAD_ID into the process-global
os.environ. Because asyncio tasks share the same process, Message B would
overwrite Message A's values mid-flight, causing background-task notifications
and tool calls to route to the wrong thread/chat.
Replace os.environ with Python's contextvars.ContextVar. Each asyncio task
(and any run_in_executor thread it spawns) gets its own copy, so concurrent
messages never interfere.
Changes:
- New gateway/session_context.py with ContextVar definitions, set/clear/get
helpers, and os.environ fallback for CLI/cron/test backward compatibility
- gateway/run.py: _set_session_env returns reset tokens, _clear_session_env
accepts them for proper cleanup in finally blocks
- All tool consumers updated: cronjob_tools, send_message_tool, skills_tool,
terminal_tool (both notify_on_complete AND check_interval blocks), tts_tool,
agent/skill_utils, agent/prompt_builder
- Tests updated for new contextvar-based API
Fixes#7358
Co-authored-by: teknium1 <127238744+teknium1@users.noreply.github.com>
Call child.close() in the _run_single_child finally block after
unregistering the child from the parent's active children list.
Previously child AIAgent instances were only removed from the tracking
list but never had their resources released — the OpenAI/httpx client
and any tool subprocesses relied entirely on garbage collection.
Ref: #7131
- Bug 1: replace read_file(limit=10000) with read_file_raw in _apply_update,
preventing silent truncation of files >2000 lines and corruption of lines
>2000 chars; add read_file_raw to FileOperations abstract interface and
ShellFileOperations
- Bug 2: split apply_v4a_operations into validate-then-apply phases; if any
hunk fails validation, zero writes occur (was: continue after failure,
leaving filesystem partially modified)
- Bug 3: parse_v4a_patch now returns an error for begin-marker-with-no-ops,
empty file paths, and moves missing a destination (was: always returned
error=None)
- Bug 4: raise strategy 7 (block anchor) single-candidate similarity threshold
from 0.10 to 0.50, eliminating false-positive matches in repetitive code
- Bug 5: add _strategy_unicode_normalized (new strategy 7) with position
mapping via _build_orig_to_norm_map; smart quotes and em-dashes in
LLM-generated patches now match via strategies 1-6 before falling through
to fuzzy strategies
- Bug 6: extend fuzzy_find_and_replace to return 4-tuple (content, count,
error, strategy); update all 5 call sites across patch_parser.py,
file_operations.py, and skill_manager_tool.py
- Bug 7: guard in _apply_update returns error when addition-only context hint
is ambiguous (>1 occurrences); validation phase errors on both 0 and >1
- Bug 8: _apply_delete returns error (not silent success) on missing file
- Bug 9: _validate_operations checks source existence and destination absence
for MOVE operations before any write occurs
Three locations perform `int()` conversion on environment variables or
HTTP headers without error handling, causing unhandled `ValueError` crashes
when the values are non-numeric:
1. `send_message_tool.py` — `EMAIL_SMTP_PORT` env var parsed outside the
try/except block; a non-numeric value crashes `_send_email()` instead
of returning a user-friendly error.
2. `process_registry.py` — `TERMINAL_TIMEOUT` env var parsed without
protection; a non-numeric value crashes the `wait()` method.
3. `skills_hub.py` — HTTP `Retry-After` header can contain date strings
per RFC 7231; `int()` conversion crashes on non-numeric values.
All three now fall back to their default values on `ValueError`/`TypeError`.
Two issues with sandbox container spawning:
1. PID 1 was `sleep 2h` which doesn't call wait() — every background
process that exited became a zombie (<defunct>), and the process
tool reported them as "running" because zombie PIDs still exist in
the process table. Fix: add --init to docker run, which uses
tini (Docker) or catatonit (Podman) as PID 1 to reap children
automatically. Both runtimes support --init natively.
2. The fixed 2-hour lifetime was arbitrary and sometimes too short
for long agent sessions. Fix: replace 'sleep 2h' with
'sleep infinity'. The idle reaper (_cleanup_inactive_envs, gated
by terminal.lifetime_seconds, default 300s) already handles
cleanup based on last activity timestamp — there's no need for
the container itself to have a fixed death timer.
Fixes#6908.
`delegate_task` silently truncated batch tasks to 3 — the model sends
5 tasks, gets results for 3, never told 2 were dropped. Now returns a
clear tool_error explaining the limit and how to fix it.
The limit is configurable via:
- delegation.max_concurrent_children in config.yaml (priority 1)
- DELEGATION_MAX_CONCURRENT_CHILDREN env var (priority 2)
- default: 3
Uses the same _load_config() path as the rest of delegate_task for
consistent config priority. Clamps to min 1, warns on non-integer
config values.
Also removes the hardcoded maxItems: 3 from the JSON schema — the
schema was blocking the model from even attempting >3 tasks before
the runtime check could fire. The runtime check gives a much more
actionable error message.
Backwards compatible: default remains 3, existing configs unchanged.
Isolate system tool configs (git, ssh, gh, npm) per profile by injecting
a per-profile HOME into subprocess environments only. The Python
process's own os.environ['HOME'] and Path.home() are never modified,
preserving all existing profile infrastructure.
Activation is directory-based: when {HERMES_HOME}/home/ exists on disk,
subprocesses see it as HOME. The directory is created automatically for:
- Docker: entrypoint.sh bootstraps it inside the persistent volume
- Named profiles: added to _PROFILE_DIRS in profiles.py
Injection points (all three subprocess env builders):
- tools/environments/local.py _make_run_env() — foreground terminal
- tools/environments/local.py _sanitize_subprocess_env() — background procs
- tools/code_execution_tool.py child_env — execute_code sandbox
Single source of truth: hermes_constants.get_subprocess_home()
Closes#4426
hermes skills browse ran all 7 source adapters serially with no overall
timeout and no progress indicator. On a cold cache, GitHubSource alone
could make 100+ sequential HTTP calls (directory listing + inspect per
skill per tap), taking 5+ minutes with no output — appearing to hang.
Changes:
- Add parallel_search_sources() in tools/skills_hub.py that runs all
source adapters concurrently via ThreadPoolExecutor with a 30s
overall timeout. Sources that finish in time contribute results;
slow ones are skipped gracefully with a visible notice.
- Update unified_search() to use parallel_search_sources() internally.
- Update do_browse() and do_search() in hermes_cli/skills_hub.py to
show a Rich spinner while fetching, so the user sees activity.
- Bump per-source limits (clawhub 50→500, lobehub 50→500, etc.) now
that fetching is parallel — yields far more results per browse.
- Report timed-out sources and suggest re-running for cached results.
- Replace 'inspect/install' footer with 'search deeper' tip.
Worst-case latency drops from 5+ minutes (serial) to ~30s (parallel
with timeout cap). Result count should jump from ~242 to 1000+.
When delegate_task runs, the parent agent's activity tracker freezes
because child.run_conversation() blocks and the child's own
_touch_activity() never propagates back to the parent. The gateway
inactivity timeout then fires a spurious 'No activity' warning and
eventually kills the agent, even though the subagent is actively working.
Fix: add a heartbeat thread in _run_single_child that calls
parent._touch_activity() every 30 seconds with detail from the child's
activity summary (current tool, iteration count). The thread is a daemon
that starts before child.run_conversation() and is cleaned up in the
finally block.
This also improves the gateway 'Still working...' status messages —
instead of just 'running: delegate_task', users now see what the
subagent is actually doing (e.g., 'delegate_task: subagent running
terminal (iteration 5/50)').
- Remove sys.path.insert hack (leftover from standalone dev)
- Add token lock (acquire_scoped_lock/release_scoped_lock) in
connect()/disconnect() to prevent duplicate pollers across profiles
- Fix get_connected_platforms: WEIXIN check must precede generic
token/api_key check (requires both token AND account_id)
- Add WEIXIN_HOME_CHANNEL_NAME to _EXTRA_ENV_KEYS
- Add gateway setup wizard with QR login flow
- Add platform status check for partially configured state
- Add weixin.md docs page with full adapter documentation
- Update environment-variables.md reference with all 11 env vars
- Update sidebars.ts to include weixin docs page
- Wire all gateway integration points onto current main
Salvaged from PR #6747 by Zihan Huang.
Four gaps in DANGEROUS_PATTERNS found by running 10 targeted tests that
each mapped to a specific pattern in approval.py and checked whether the
documented defense actually held.
1. **Heredoc script injection** — `python3 << 'EOF'` bypasses the
existing `-e`/`-c` flag pattern. Adds pattern for interpreter + `<<`
covering python{2,3}, perl, ruby, node.
2. **PID expansion self-termination** — `kill -9 $(pgrep hermes)` is
opaque to the existing `pkill|killall` + name pattern because command
substitution is not expanded at detection time. Adds structural
patterns matching `kill` + `$(pgrep` and backtick variants.
3. **Git destructive operations** — `git reset --hard`, `push --force`,
`push -f`, `clean -f*`, and `branch -D` were entirely absent.
Note: `branch -d` also triggers because IGNORECASE is global —
acceptable since -d is still a delete, just a safe one, and the
prompt is only a confirmation, not a hard block.
4. **chmod +x then execute** — two-step social engineering where a
script containing dangerous commands is first written to disk (not
checked by write_file), then made executable and run as `./script`.
Pattern catches `chmod +x ... [;&|]+ ./` combos. Does not solve the
deeper architectural issue (write_file not checking content) — that
is called out in the PR description as a known limitation.
Tests: 23 new cases across 4 test classes, all in test_approval.py:
- TestHeredocScriptExecution (7 cases, incl. regressions for -c)
- TestPgrepKillExpansion (5 cases, incl. safe kill PID negative)
- TestGitDestructiveOps (8 cases, incl. safe git status/push negatives)
- TestChmodExecuteCombo (3 cases, incl. safe chmod-only negative)
Full suite: 146 passed, 0 failed.
Follow-up to Dusk1e's PR #7120 (Slack send_image redirect guard):
- Rename _safe_url_for_log -> safe_url_for_log (drop underscore) since
it is now imported cross-module by the Slack adapter
- Add _ssrf_redirect_guard httpx event hook to cache_image_from_url()
and cache_audio_from_url() in base.py — same pattern as vision_tools
and the Slack adapter fix
- Update url_safety.py docstring to reflect broader coverage
- Add regression tests for image/audio redirect blocking + safe passthrough
When kill_process() sends SIGTERM, both it and the reader thread race
to call _move_to_finished() — kill_process sets exit_code=-15 and
enqueues a notification, then the reader thread's process.wait()
returns with exit_code=143 (128+SIGTERM) and enqueues a second one.
Fix: make _move_to_finished() idempotent by tracking whether the
session was actually removed from _running. The second call sees it
was already moved and skips the completion_queue.put().
Adds regression test: test_move_to_finished_idempotent_no_duplicate
Automated dead code audit using vulture + coverage.py + ast-grep intersection,
confirmed by Opus deep verification pass. Every symbol verified to have zero
production callers (test imports excluded from reachability analysis).
Removes ~1,534 lines of dead production code across 46 files and ~1,382 lines
of stale test code. 3 entire files deleted (agent/builtin_memory_provider.py,
hermes_cli/checklist.py, tests/hermes_cli/test_setup_model_selection.py).
Co-authored-by: alt-glitch <balyan.sid@gmail.com>
When an MCP server returns both content (model-oriented text) and
structuredContent (machine-oriented JSON), the client now combines
them instead of discarding content. The text content becomes the
primary result (what the agent reads), and structuredContent is
included as supplementary metadata.
Previously, structuredContent took full precedence — causing data
loss for servers like Desktop Commander that put the actual file
text in content and metadata in structuredContent.
MCP spec guidance: for conversational/agent UX, prefer content.
Legacy flat stt.model config key (from cli-config.yaml.example and older
versions) was passed as a model override to transcribe_audio() by the
gateway, bypassing provider-specific model resolution. When the provider
was 'local' (faster-whisper), this caused:
ValueError: Invalid model size 'whisper-1'
Changes:
- gateway/run.py, discord.py: stop passing model override — let
transcribe_audio() handle provider-specific model resolution internally
- get_stt_model_from_config(): now provider-aware, reads from the correct
nested section (stt.local.model, stt.openai.model, etc.); ignores
legacy flat key for local provider to prevent model name mismatch
- cli-config.yaml.example: updated STT section to show nested provider
config structure instead of legacy flat key
- config migration v13→v14: moves legacy stt.model to the correct
provider section and removes the flat key
Reported by community user on Discord.
Add Discord thread support to cron delivery and send_message_tool.
- _parse_target_ref: handle discord platform with chat_id:thread_id format
- _send_discord: add thread_id param, route to /channels/{thread_id}/messages
- _send_to_platform: pass thread_id through for Discord
- Discord adapter send(): read thread_id from metadata for gateway path
- Update tool schema description to document Discord thread targets
Cherry-picked from PR #7046 by pandacooming (maxyangcn).
Follow-up fixes:
- Restore proxy support (resolve_proxy_url/proxy_kwargs_for_aiohttp) that was
accidentally deleted — would have caused NameError at runtime
- Remove duplicate _DISCORD_TARGET_RE regex; reuse existing _TELEGRAM_TOPIC_TARGET_RE
via _NUMERIC_TOPIC_RE alias (identical pattern)
- Fix misleading test comments about Discord negative snowflake IDs
(Discord uses positive snowflakes; negative IDs are a Telegram convention)
- Rewrite misleading scheduler test that claimed to exercise home channel
fallback but actually tested the explicit platform:chat_id parsing path
The same .* pattern vulnerable to newline bypass that was fixed in
prompt_builder.py (PR #6925) also existed in skills_guard.py. Changed
to [\s\S]*? to match across newlines.
prompt_builder.py: The `hidden_div` detection pattern uses `.*` which does not
match newlines in Python regex (re.DOTALL is not passed). An attacker can bypass
detection by splitting the style attribute across lines:
`<div style="color:red;\ndisplay: none">injected content</div>`
Replace `.*` with `[\s\S]*?` to match across line boundaries.
credential_files.py: `_load_config_files()` catches all exceptions at DEBUG level
(line 171), making YAML parse failures invisible in production logs. Users whose
credential files silently fail to mount into sandboxes have no diagnostic clue.
Promote to WARNING to match the severity pattern used by the path validation
warnings at lines 150 and 158 in the same function.
webhook.py: `_reload_dynamic_routes()` logs JSON parse failures at WARNING (line
265) but the impact — stale/corrupted dynamic routes persisting silently — warrants
ERROR level to ensure operator visibility in alerting pipelines.
Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
Three silent `except Exception` blocks in approval.py (lines 345, 387, 469) return
fallback values with zero logging — making it impossible to debug callback failures,
allowlist load errors, or config read issues. Add logger.warning/error calls that
match the pattern already used by save_permanent_allowlist() and _smart_approve()
in the same file.
In mcp_oauth.py, narrow the overly-broad `except Exception` in get_tokens() and
get_client_info() to the specific exceptions Pydantic's model_validate() can raise
(ValueError, TypeError, KeyError), and include the exception message in the warning.
Also wrap the _wait_for_callback() polling loop in try/finally so the HTTPServer is
always closed — previously an asyncio.CancelledError or any exception in the loop
would leak the server socket.
Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
error_classifier.py: Message-only auth errors ("invalid api key", "unauthorized",
etc.) were classified as retryable=True (line 707), inconsistent with the HTTP 401
path (line 432) which correctly uses retryable=False + should_fallback=True. The
mismatch causes 3 wasted retries with the same broken credential before fallback,
while 401 errors immediately attempt fallback. Align the message-based path to
match: retryable=False, should_fallback=True.
web_tools.py: The _PREFIX_RE secret-detection check in web_extract_tool() runs
against the raw URL string (line 1196). URL-encoded secrets like %73k-1234... (
sk-1234...) bypass the filter because the regex expects literal ASCII. Add
urllib.parse.unquote() before the check so percent-encoded variants are also caught.
Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
Replace per-backend ad-hoc file sync with a shared FileSyncManager
that handles mtime-based change detection, remote deletion of
locally-removed files, and transactional state updates.
- New FileSyncManager class (tools/environments/file_sync.py)
with callbacks for upload/delete, rate limiting, and rollback
- Shared iter_sync_files() eliminates 3 duplicate implementations
- SSH: replace unconditional rsync with scp + mtime skip
- Modal/Daytona: replace inline _synced_files dict with manager
- All 3 backends now sync credentials + skills + cache uniformly
- Remote deletion: files removed locally are cleaned from remote
- HERMES_FORCE_FILE_SYNC=1 env var for debugging
- Base class _before_execute() simplified to empty hook
- 12 unit tests covering mtime skip, deletion, rollback, rate limiting
Change behavior from silent clamping to returning an error when the
model requests a foreground timeout exceeding FOREGROUND_MAX_TIMEOUT.
This forces the model to use background=true for long-running commands
rather than silently changing its intent.
- Config default timeouts above the cap are NOT rejected (user's choice)
- Only explicit model-requested timeouts trigger rejection
- Added boundary test for timeout exactly at the limit
When the model calls terminal() in foreground mode without background=true
(e.g. to start a server), the tool call blocks until the command exits or
the timeout expires. Without an upper bound the model can request arbitrarily
high timeouts (the schema had minimum=1 but no maximum), blocking the entire
agent session for hours until the gateway idle watchdog kills it.
Changes:
- Add FOREGROUND_MAX_TIMEOUT (600s, configurable via
TERMINAL_MAX_FOREGROUND_TIMEOUT env var) that caps foreground timeout
- Clamp effective_timeout to the cap when background=false and timeout
exceeds the limit
- Include a timeout_note in the tool result when clamped, nudging the
model to use background=true for long-running processes
- Update schema description to show the max timeout value
- Remove dead clamping code in the background branch that could never
fire (max_timeout was set to effective_timeout, so timeout > max_timeout
was always false)
- Add 7 tests covering clamping, no-clamping, config-default-exceeds-cap
edge case, background bypass, default timeout, constant value, and
schema content
Self-review fixes:
- Fixed bug where timeout_note said 'Requested timeout Nones' when
clamping fired from config default exceeding cap (timeout param is
None). Now uses unclamped_timeout instead of the raw timeout param.
- Removed unused pytest import from test file
- Extracted test config dict into _make_env_config() helper
- Fixed tautological test_default_value assertion
- Added missing test for config default > cap with no model timeout
Replace 6 identical copies of the Termux detection function across
cli.py, browser_tool.py, voice_mode.py, status.py, doctor.py, and
gateway.py with a single shared implementation in hermes_constants.py.
Each call site imports with its original local name to preserve all
existing callers (internal references and test monkeypatches).
When managed_persistence is enabled, cleanup_all now only clears local
tracking state without sending DELETE requests to the Camofox server.
This prevents persistent browser profiles (cookies, logins, localStorage)
from being destroyed during process-wide cleanup.
Ephemeral sessions still get full server-side deletion as before.
Follow-up improvements on top of the shared resolver from PR #6562:
- Add platform_env_var parameter to resolve_proxy_url() so DISCORD_PROXY
takes priority over generic HTTPS_PROXY/ALL_PROXY env vars
- Add SOCKS proxy support via aiohttp_socks.ProxyConnector with rdns=True
(critical for GFW/Shadowrocket/Clash users — issue #6649)
- proxy_kwargs_for_bot() returns connector= for SOCKS, proxy= for HTTP
- proxy_kwargs_for_aiohttp() returns split (session_kw, request_kw) for
standalone aiohttp sessions
- Add proxy support to send_message_tool.py (Discord REST, Slack, SMS)
for cron job delivery behind proxies (from PR #2208)
- Add proxy support to Discord image/document downloads
- Fix duplicate import sys in base.py
Fixes blockquote > escaping, edit_message raw markdown, ***bold italic***
handling, HTML entity double-escaping (&amp;), Wikipedia URL parens
truncation, and step numbering format. Also adds format_message to the
tool-layer _send_to_platform for consistent formatting across all
delivery paths.
Changes:
- Protect Slack entities (<@user>, <https://...|label>, <!here>) from
escaping passes
- Protect blockquote > markers before HTML entity escaping
- Unescape-before-escape for idempotent HTML entity handling
- ***bold italic*** → *_text_* conversion (before **bold** pass)
- URL regex upgraded to handle balanced parentheses
- mrkdwn:True flag on chat_postMessage payloads
- format_message applied in edit_message and send_message_tool
- 52 new tests (format, edit, streaming, splitting, tool chunking)
- Use reversed(dict) idiom for placeholder restoration
Based on PR #3715 by dashed, cherry-picked onto current main.
`_cleanup_task_resources` was unconditionally calling `cleanup_vm()` at
the end of every `run_conversation` (i.e. every user turn), tearing down
the docker/daytona/modal sandbox container regardless of its
`persistent_filesystem` setting. This contradicted the documented intent
of `terminal.lifetime_seconds` (idle reaper) and `container_persistent`,
and caused per-turn loss of `/workspace`, `~/.config`, agent CLI auth
state, and any other content living inside the sandbox.
The unconditional teardown was introduced in fbd3a2fd ("prevent leakage
of morph instances between tasks", 2025-11-04) to plug a Morph backend
leak, two days after `lifetime_seconds` shipped in faecbddd. It was
later refactored into `_cleanup_task_resources` in 70dd3a16 without
changing semantics. Code and docs have disagreed since.
Fix: introduce `terminal_tool.is_persistent_env(task_id)` and skip the
per-turn `cleanup_vm` when the active env is persistent. The idle reaper
(`_cleanup_inactive_envs`) still tears persistent envs down once
`terminal.lifetime_seconds` is exceeded. Non-persistent backends (Morph)
are unchanged — still torn down per turn, preserving the original
leak-prevention intent.
Fixes 9 test failures on current main, incorporating ideas from PR stack
#6219-#6222 by xinbenlv with corrections:
- model_metadata: sync HF context length key casing
(minimaxai/minimax-m2.5 → MiniMaxAI/MiniMax-M2.5)
- cli.py: route quick command error output through self.console
instead of creating a new ChatConsole() instance
- docker.py: explicit docker_forward_env entries now bypass the
Hermes secret blocklist (intentional opt-in wins over generic filter)
- auxiliary_client: revert _read_main_provider() to simple
provider.strip().lower() — the _normalize_aux_provider() call
introduced in 5c03f2e7 stripped the custom: prefix, breaking
named custom provider resolution
- auxiliary_client: flip vision auto-detection order to
active provider → OpenRouter → Nous → stop (was OR → Nous → active)
- test: update vision priority test to match new order
Based on PR #6219-#6222 by xinbenlv.
mistralai v2.x is a namespace package — `Mistral` class lives at
`mistralai.client`, not at the top-level `mistralai` module. The
previous `from mistralai import Mistral` raises ImportError at runtime.
Update both production code and test fixture to use the correct path.
* fix(tools): skip camofox auto-cleanup when managed persistence is enabled
When managed_persistence is enabled, cleanup_browser() was calling
camofox_close() which destroys the server-side browser context via
DELETE /sessions/{userId}, killing login sessions across cron runs.
Add camofox_soft_cleanup() — a public wrapper that drops only the
in-memory session entry when managed persistence is on, returning True.
When persistence is off it returns False so the caller falls back to
the full camofox_close(). The inactivity reaper still handles idle
resource cleanup.
Also surface a logger.warning() when _managed_persistence_enabled()
fails to load config, replacing a silent except-and-return-False.
Salvaged from #6182 by el-analista (Eduardo Perea Fernandez).
Added public API wrapper to avoid cross-module private imports,
and test coverage for both persistence paths.
Co-authored-by: Eduardo Perea Fernandez <el-analista@users.noreply.github.com>
* fix(doctor): only check the active memory provider, not all providers unconditionally
hermes doctor had hardcoded Honcho Memory and Mem0 Memory sections that
always ran regardless of the user's memory.provider config setting. After
the swappable memory provider update (#4623), users with leftover Honcho
config but no active provider saw false 'broken' errors.
Replaced both sections with a single Memory Provider section that reads
memory.provider from config.yaml and only checks the configured provider.
Users with no external provider see a green 'Built-in memory active' check.
Reported by community user michaelruiz001, confirmed by Eri (Honcho).
---------
Co-authored-by: Eduardo Perea Fernandez <el-analista@users.noreply.github.com>
Previously crash recovery recreated detached sessions as if they were
fully managed, so polls and kills could lie about liveness and the
checkpoint could forget recovered jobs after the next restart.
This commit refreshes recovered host-backed sessions from real PID
state, keeps checkpoint data durable, and preserves notify watcher
metadata while treating sandbox-only PIDs as non-recoverable.
- Persist `pid_scope` in `tools/process_registry.py` and skip
recovering sandbox-backed entries without a host-visible PID handle
- Refresh detached sessions on access so `get`/`poll`/`wait` and active
session queries observe exited processes instead of hanging forever
- Allow recovered host PIDs to be terminated honestly and requeue
`notify_on_complete` watchers during checkpoint recovery
- Add regression tests for durable checkpoints, detached exit/kill
behavior, sandbox skip logic, and recovered notify watchers
- DEFAULT_RESULT_SIZE_CHARS: 50K -> 100K (match current _LARGE_RESULT_CHARS)
- DEFAULT_PREVIEW_SIZE_CHARS: 2K -> 1.5K (match current _LARGE_RESULT_PREVIEW_CHARS)
- Per-tool overrides all set to 100K (terminal, execute_code, search_files)
- Remove pre-read byte guard (no behavioral regression vs current main)
- Revert limit signature change to int=500 (match current default)
- Restore original read_file schema description
- Update test assertions to match 100K thresholds
Add BudgetConfig dataclass to centralize and make overridable the
hardcoded constants (50K per-result, 200K per-turn, 2K preview) that
control when tool outputs get persisted to sandbox. Configurable at
the RL environment level via HermesAgentEnvConfig fields, threaded
through HermesAgentLoop to the storage layer.
Resolution: pinned (read_file=inf) > env config overrides > registry
per-tool > default. CLI override: --env.turn_budget_chars 80000
_deliver_result() now returns Optional[str] — None on success, error
message on failure. All failure paths (unknown platform, platform
disabled, config load error, send failure, unresolvable target)
return descriptive error strings.
mark_job_run() gains delivery_error param, tracked as
last_delivery_error on the job — separate from agent execution errors.
A job where the agent succeeded but delivery failed shows
last_status='ok' + last_delivery_error='...'.
The cronjob list tool now surfaces last_delivery_error so agents and
users can see when cron outputs aren't arriving.
Inspired by PR #5863 (oxngon) — reimplemented with proper wiring.
Tests: 3 new mark_job_run tests + 6 new _deliver_result return tests.
- The MCP SDK Pydantic model uses camelCase (structuredContent), not
snake_case (structured_content). The original getattr was a silent no-op.
- When structuredContent is present, return it AS the result instead of
alongside text — the structured payload is the machine-readable data.
- Move test file to tests/tools/ and fix fake class to use camelCase.
- Patch _run_on_mcp_loop in tests so the handler actually executes.
MCP CallToolResult may include structured_content (a JSON object) alongside
content blocks. The tool handler previously only forwarded concatenated text
from content blocks, silently dropping the structured payload.
This breaks MCP tools that return a minimal human text in content while
putting the actual machine-usable payload in structured_content.
Now, when structured_content is present, it is included in the returned
JSON under the 'structuredContent' key.
FixesNousResearch/hermes-agent#5874
- Add 7 unit tests for _profile_arg: default home, named profile,
hash path, nested path, invalid name, systemd integration, launchd integration
- Add stt.local.language to config.yaml (empty = auto-detect)
- Both STT code paths now read config.yaml first, env var fallback,
then default (auto-detect for faster-whisper, 'en' for CLI command)
- HERMES_LOCAL_STT_LANGUAGE env var still works as backward-compat fallback
- gateway/run.py: Strip "(The user sent a message with no text content)"
placeholder when voice transcription succeeds — it was being appended
alongside the transcript, creating duplicate user turns.
- tools/transcription_tools.py: Wire HERMES_LOCAL_STT_LANGUAGE env var
into the faster-whisper backend. It was only used by the CLI fallback
path (_transcribe_local_command), not the primary faster-whisper path.
Replace 10 callsites across 6 files that manually opened config.yaml,
called yaml.safe_load(), and handled missing-file/parse-error fallbacks
with the new read_raw_config() helper from hermes_cli/config.py.
Each migrated site previously had 5-8 lines of boilerplate:
config_path = get_hermes_home() / 'config.yaml'
if config_path.exists():
import yaml
with open(config_path) as f:
cfg = yaml.safe_load(f) or {}
Now reduced to:
from hermes_cli.config import read_raw_config
cfg = read_raw_config()
Migrated files:
- tools/browser_tool.py (4 sites): command_timeout, cloud_provider,
allow_private_urls, record_sessions
- tools/env_passthrough.py: terminal.env_passthrough
- tools/credential_files.py: terminal.credential_files
- tools/transcription_tools.py: stt.model
- hermes_cli/commands.py: config-gated command resolution
- hermes_cli/auth.py (2 sites): model config read + provider reset
Skipped (intentionally):
- gateway/run.py: 10+ sites with local aliases, critical path
- hermes_cli/profiles.py: profile-specific config path
- hermes_cli/doctor.py: reads raw then writes fixes back
- agent/model_metadata.py: different file (context_length_cache.yaml)
- tools/website_policy.py: custom config_path param + error types
* refactor: re-architect tests to mirror the codebase
* Update tests.yml
* fix: add missing tool_error imports after registry refactor
* fix(tests): replace patch.dict with monkeypatch to prevent env var leaks under xdist
patch.dict(os.environ) can leak TERMINAL_ENV across xdist workers,
causing test_code_execution tests to hit the Modal remote path.
* fix(tests): fix update_check and telegram xdist failures
- test_update_check: replace patch("hermes_cli.banner.os.getenv") with
monkeypatch.setenv("HERMES_HOME") — banner.py no longer imports os
directly, it uses get_hermes_home() from hermes_constants.
- test_telegram_conflict/approval_buttons: provide real exception classes
for telegram.error mock (NetworkError, TimedOut, BadRequest) so the
except clause in connect() doesn't fail with "catching classes that do
not inherit from BaseException" when xdist pollutes sys.modules.
* fix(tests): accept unavailable_models kwarg in _prompt_model_selection mock
16 callsites across 14 files were re-deriving the hermes home path
via os.environ.get('HERMES_HOME', ...) instead of using the canonical
get_hermes_home() from hermes_constants. This breaks profiles — each
profile has its own HERMES_HOME, and the inline fallback defaults to
~/.hermes regardless.
Fixed by importing and calling get_hermes_home() at each site. For
files already inside the hermes process (agent/, hermes_cli/, tools/,
gateway/, plugins/), this is always safe. Files that run outside the
process context (mcp_serve.py, mcp_oauth.py) already had correct
try/except ImportError fallbacks and were left alone.
Skipped: hermes_constants.py (IS the implementation), env_loader.py
(bootstrap), profiles.py (intentionally manipulates the env var),
standalone scripts (optional-skills/, skills/), and tests.
Comprehensive cleanup across 80 files based on automated (ruff, pyflakes, vulture)
and manual analysis of the entire codebase.
Changes by category:
Unused imports removed (~95 across 55 files):
- Removed genuinely unused imports from all major subsystems
- agent/, hermes_cli/, tools/, gateway/, plugins/, cron/
- Includes imports in try/except blocks that were truly unused
(vs availability checks which were left alone)
Unused variables removed (~25):
- Removed dead variables: connected, inner, channels, last_exc,
source, new_server_names, verify, pconfig, default_terminal,
result, pending_handled, temperature, loop
- Dropped unused argparse subparser assignments in hermes_cli/main.py
(12 instances of add_parser() where result was never used)
Dead code removed:
- run_agent.py: Removed dead ternary (None if False else None) and
surrounding unreachable branch in identity fallback
- run_agent.py: Removed write-only attribute _last_reported_tool
- hermes_cli/providers.py: Removed dead @property decorator on
module-level function (decorator has no effect outside a class)
- gateway/run.py: Removed unused MCP config load before reconnect
- gateway/platforms/slack.py: Removed dead SessionSource construction
Undefined name bugs fixed (would cause NameError at runtime):
- batch_runner.py: Added missing logger = logging.getLogger(__name__)
- tools/environments/daytona.py: Added missing Dict and Path imports
Unnecessary global statements removed (14):
- tools/terminal_tool.py: 5 functions declared global for dicts
they only mutated via .pop()/[key]=value (no rebinding)
- tools/browser_tool.py: cleanup thread loop only reads flag
- tools/rl_training_tool.py: 4 functions only do dict mutations
- tools/mcp_oauth.py: only reads the global
- hermes_time.py: only reads cached values
Inefficient patterns fixed:
- startswith/endswith tuple form: 15 instances of
x.startswith('a') or x.startswith('b') consolidated to
x.startswith(('a', 'b'))
- len(x)==0 / len(x)>0: 13 instances replaced with pythonic
truthiness checks (not x / bool(x))
- in dict.keys(): 5 instances simplified to in dict
- Redefined unused name: removed duplicate _strip_mdv2 import in
send_message_tool.py
Other fixes:
- hermes_cli/doctor.py: Replaced undefined logger.debug() with pass
- hermes_cli/config.py: Consolidated chained .endswith() calls
Test results: 3934 passed, 17 failed (all pre-existing on main),
19 skipped. Zero regressions.
* feat: switch managed browser provider from Browserbase to Browser Use
The Nous subscription tool gateway now routes browser automation through
Browser Use instead of Browserbase. This commit:
- Adds managed Nous gateway support to BrowserUseProvider (idempotency
keys, X-BB-API-Key auth header, external_call_id persistence)
- Removes managed gateway support from BrowserbaseProvider (now
direct-only via BROWSERBASE_API_KEY/BROWSERBASE_PROJECT_ID)
- Updates browser_tool.py fallback: prefers Browser Use over Browserbase
- Updates nous_subscription.py: gateway vendor 'browser-use', auto-config
sets cloud_provider='browser-use' for new subscribers
- Updates tools_config.py: Nous Subscription entry now uses Browser Use
- Updates setup.py, cli.py, status.py, prompt_builder.py display strings
- Updates all affected tests to match new behavior
Browserbase remains fully functional for users with direct API credentials.
The change only affects the managed/subscription path.
* chore: remove redundant Browser Use hint from system prompt
* fix: upgrade Browser Use provider to v3 API
- Base URL: api/v2 -> api/v3 (v2 is legacy)
- Unified all endpoints to use native Browser Use paths:
- POST /browsers (create session, returns cdpUrl)
- PATCH /browsers/{id} with {action: stop} (close session)
- Removed managed-mode branching that used Browserbase-style
/v1/sessions paths — v3 gateway now supports /browsers directly
- Removed unused managed_mode variable in close_session
* fix(browser-use): use X-Browser-Use-API-Key header for managed mode
The managed gateway expects X-Browser-Use-API-Key, not X-BB-API-Key
(which is a Browserbase-specific header). Using the wrong header caused
a 401 AUTH_ERROR on every managed-mode browser session create.
Simplified _headers() to always use X-Browser-Use-API-Key regardless
of direct vs managed mode.
* fix(nous_subscription): browserbase explicit provider is direct-only
Since managed Nous gateway now routes through Browser Use, the
browserbase explicit provider path should not check managed_browser_available
(which resolves against the browser-use gateway). Simplified to direct-only
with managed=False.
* fix(browser-use): port missing improvements from PR #5605
- CDP URL normalization: resolve HTTP discovery URLs to websocket after
cloud provider create_session() (prevents agent-browser failures)
- Managed session payload: send timeout=5 and proxyCountryCode=us for
gateway-backed sessions (prevents billing overruns)
- Update prompt builder, browser_close schema, and module docstring to
replace remaining Browserbase references with Browser Use
- Dynamic /browser status detection via _get_cloud_provider() instead
of hardcoded env var checks (future-proof for new providers)
- Rename post_setup key from 'browserbase' to 'agent_browser'
- Update setup hint to mention Browser Use alongside Browserbase
- Add tests: CDP normalization, browserbase direct-only guard,
managed browser-use gateway, direct browserbase fallback
---------
Co-authored-by: rob-maron <132852777+rob-maron@users.noreply.github.com>
* refactor: remove browser_close tool — auto-cleanup handles it
The browser_close tool was called in only 9% of browser sessions (13/144
navigations across 66 sessions), always redundantly — cleanup_browser()
already runs via _cleanup_task_resources() at conversation end, and the
background inactivity reaper catches anything else.
Removing it saves one tool schema slot in every browser-enabled API call.
Also fixes a latent bug: cleanup_browser() now handles Camofox sessions
too (previously only Browserbase). Camofox sessions were never auto-cleaned
per-task because they live in a separate dict from _active_sessions.
Files changed (13):
- tools/browser_tool.py: remove function, schema, registry entry; add
camofox cleanup to cleanup_browser()
- toolsets.py, model_tools.py, prompt_builder.py, display.py,
acp_adapter/tools.py: remove browser_close from all tool lists
- tests/: remove browser_close test, update toolset assertion
- docs/skills: remove all browser_close references
* fix: repeat browser_scroll 5x per call for meaningful page movement
Most backends scroll ~100px per call — barely visible on a typical
viewport. Repeating 5x gives ~500px (~half a viewport), making each
scroll tool call actually useful.
Backend-agnostic approach: works across all 7+ browser backends without
needing to configure each one's scroll amount individually. Breaks
early on error for the agent-browser path.
* feat: auto-return compact snapshot from browser_navigate
Every browser session starts with navigate → snapshot. Now navigate
returns the compact accessibility tree snapshot inline, saving one
tool call per browser task.
The snapshot captures the full page DOM (not viewport-limited), so
scroll position doesn't affect it. browser_snapshot remains available
for refreshing after interactions or getting full=true content.
Both Browserbase and Camofox paths auto-snapshot. If the snapshot
fails for any reason, navigation still succeeds — the snapshot is
a bonus, not a requirement.
Schema descriptions updated to guide models: navigate mentions it
returns a snapshot, snapshot mentions it's for refresh/full content.
* refactor: slim cronjob tool schema — consolidate model/provider, drop unused params
Session data (151 calls across 67 sessions) showed several schema
properties were never used by models. Consolidated and cleaned up:
Removed from schema (still work via backend/CLI):
- skill (singular): use skills array instead
- reason: pause-only, unnecessary
- include_disabled: now defaults to true
- base_url: extreme edge case, zero usage
- provider (standalone): merged into model object
Consolidated:
- model + provider → single 'model' object with {model, provider} fields.
If provider is omitted, the current main provider is pinned at creation
time so the job stays stable even if the user changes their default.
Kept:
- script: useful data collection feature
- skills array: standard interface for skill loading
Schema shrinks from 14 to 10 properties. All backend functionality
preserved — the Python function signature and handler lambda still
accept every parameter.
* fix: remove mixture_of_agents from core toolsets — opt-in only via hermes tools
MoA was in _HERMES_CORE_TOOLS and composite toolsets (hermes-cli,
hermes-messaging, safe), which meant it appeared in every session
for anyone with OPENROUTER_API_KEY set. The _DEFAULT_OFF_TOOLSETS
gate only works after running 'hermes tools' explicitly.
Now MoA only appears when a user explicitly enables it via
'hermes tools'. The moa toolset definition and check_fn remain
unchanged — it just needs to be opted into.
* feat: notify_on_complete for background processes
When terminal(background=true, notify_on_complete=true), the system
auto-triggers a new agent turn when the process exits — no polling needed.
Changes:
- ProcessSession: add notify_on_complete field
- ProcessRegistry: add completion_queue, populate on _move_to_finished()
- Terminal tool: add notify_on_complete parameter to schema + handler
- CLI: drain completion_queue after agent turn AND during idle loop
- Gateway: enhanced _run_process_watcher injects synthetic MessageEvent
on completion, triggering a full agent turn
- Checkpoint persistence includes notify_on_complete for crash recovery
- code_execution_tool: block notify_on_complete in sandbox scripts
- 15 new tests covering queue mechanics, checkpoint round-trip, schema
* docs: update terminal tool descriptions for notify_on_complete
- background: remove 'ONLY for servers' language, describe both patterns
(long-lived processes AND long-running tasks with notify_on_complete)
- notify_on_complete: more prescriptive about when to use it
- TERMINAL_TOOL_DESCRIPTION: remove 'Do NOT use background for builds'
guidance that contradicted the new feature
Selectively cherry-picked from PR #5501 by MestreY0d4-Uninter.
- Add _resolve_workspace_hint() to detect parent's working directory
- Inject WORKSPACE PATH into child system prompts
- Add rule: never assume /workspace/ container paths
- Excludes the cli.py queue-busy-input changes from the original PR
Cherry-picked from PR #5580 by MestreY0d4-Uninter.
- Share parent's credential pool with child agents for key rotation
- Leasing layer spreads parallel children across keys (least-loaded)
- Thread-safe acquire_lease/release_lease in CredentialPool
- Reverted sneaked-in tool-name restoration change (kept original
getattr + isinstance guard pattern)
Two fixes:
1. Replace all stale 'hermes login' references with 'hermes auth' across
auth.py, auxiliary_client.py, delegate_tool.py, config.py, run_agent.py,
and documentation. The 'hermes login' command was deprecated; 'hermes auth'
now handles OAuth credential management.
2. Fix credential removal not persisting for singleton-sourced credentials
(device_code for openai-codex/nous, hermes_pkce for anthropic).
auth_remove_command already cleared env vars for env-sourced credentials,
but singleton credentials stored in the auth store were re-seeded by
_seed_from_singletons() on the next load_pool() call. Now clears the
underlying auth store entry when removing singleton-sourced credentials.
* feat(tools): add Firecrawl cloud browser provider
Adds Firecrawl (https://firecrawl.dev) as a cloud browser provider
alongside Browserbase and Browser Use. All browser tools route through
Firecrawl's cloud browser via CDP when selected.
- tools/browser_providers/firecrawl.py — FirecrawlProvider
- tools/browser_tool.py — register in _PROVIDER_REGISTRY
- hermes_cli/tools_config.py — add to onboarding provider picker
- hermes_cli/setup.py — add to setup summary
- hermes_cli/config.py — add FIRECRAWL_BROWSER_TTL config
- website/docs/ — browser docs and env var reference
Based on #4490 by @developersdigest.
Co-Authored-By: Developers Digest <124798203+developersdigest@users.noreply.github.com>
* refactor: simplify FirecrawlProvider.emergency_cleanup
Use self._headers() and self._api_url() instead of duplicating
env-var reads and header construction.
* fix: recognize Firecrawl in subscription browser detection
_resolve_browser_feature_state() now handles "firecrawl" as a direct
browser provider (same pattern as "browser-use"), so hermes setup
summary correctly shows "Browser Automation (Firecrawl)" instead of
misreporting as "Local browser".
Also fixes test_config_version_unchanged assertion (11 → 12).
---------
Co-authored-by: Developers Digest <124798203+developersdigest@users.noreply.github.com>
When parent_agent.enabled_toolsets is None (the default, meaning all tools
are enabled), subagents incorrectly fell back to DEFAULT_TOOLSETS
(['terminal', 'file', 'web']) instead of inheriting the parent's full
toolset.
Root cause:
- Line 188 used 'or' fallback: None or DEFAULT_TOOLSETS evaluates to
DEFAULT_TOOLSETS
- Line 192 checked truthiness: None is falsy, falling through to else
Fix:
- Use 'is not None' checks instead of truthiness
- When enabled_toolsets is None, derive effective toolsets from
parent_agent.valid_tool_names via the tool registry
Fixes the bug introduced in f75b1d21b and repeated in e5d14445e (PR #3269).
Shell injection via unquoted workdir interpolation in docker, singularity,
and SSH backends. When workdir contained shell metacharacters (e.g.
~/;id), arbitrary commands could execute.
Changes:
- Add shlex.quote() at each interpolation point in docker.py,
singularity.py, and ssh.py with tilde-aware quoting (keep ~
unquoted for shell expansion, quote only the subpath)
- Add _validate_workdir() allowlist in terminal_tool.py as
defense-in-depth before workdir reaches any backend
Original work by Mariano A. Nicolini (PR #5620). Salvaged with fixes
for tilde expansion (shlex.quote breaks cd ~/path) and replaced
incomplete deny-list with strict character allowlist.
Co-authored-by: Mariano A. Nicolini <entropidelic@users.noreply.github.com>
The ContextVar migration removed 'from pathlib import Path' but Path
is still used in _load_config_passthrough(). Without this import,
config-based env passthrough would raise NameError.
Subagent sessions spawned by delegate_task were created with
parent_session_id=NULL and source=cli, making them indistinguishable
from user sessions in hermes sessions list and /resume.
Changes:
- delegate_tool.py: pass parent_agent.session_id to child agent
- run_agent.py: accept parent_session_id param, pass to create_session
- hermes_state.py list_sessions_rich: filter parent_session_id IS NULL
by default (opt-in include_children=True for callers that need them)
- hermes_state.py delete_session: delete child sessions first (FK)
- hermes_state.py prune_sessions: delete children before parents (FK)
session_search already handles parent_session_id correctly — child
sessions are filtered from recent list and resolved to parent root
in full-text search results.
Fixes#5122
Before launching an MCP server via npx/uvx, queries the OSV (Open Source
Vulnerabilities) API to check if the package has known malware advisories
(MAL-* IDs). Regular CVEs are ignored — only confirmed malware is blocked.
- Free, public API (Google-maintained), ~300ms per query
- Runs once per MCP server launch, inside _run_stdio() before subprocess spawn
- Parallel with other MCP servers (asyncio.gather already in place)
- Fail-open: network errors, timeouts, unrecognized commands → allow
- Parses npm (scoped @scope/pkg@version) and PyPI (name[extras]==version)
Inspired by Block/goose extension malware check.
Add optional 'expression' parameter to browser_console that evaluates
JavaScript in the page context (like DevTools console). Returns structured
results with auto-JSON parsing.
No new tool — extends the existing browser_console schema with ~20 tokens
of overhead instead of adding a 12th browser tool.
Both backends supported:
- Browserbase: uses agent-browser 'eval' command via CDP
- Camofox: uses /tabs/{tab_id}/eval endpoint with graceful degradation
E2E verified: string eval, number eval, structured JSON, DOM manipulation,
error handling, and original console-output mode all working.
Add POST /v1/runs to start async agent runs and GET /v1/runs/{run_id}/events
for SSE streaming of typed lifecycle events (tool.started, tool.completed,
message.delta, reasoning.available, run.completed, run.failed).
Changes the internal tool_progress_callback signature from positional
(tool_name, preview, args) to event-type-first
(event_type, tool_name, preview, args, **kwargs). Existing consumers
filter on event_type and remain backward-compatible.
Adds concurrency limit (_MAX_CONCURRENT_RUNS=10) and orphaned run sweep.
Fixes logic inversion in cli.py _on_tool_progress where the original PR
would have displayed internal tools instead of non-internal ones.
Co-authored-by: Mibayy <mibayy@users.noreply.github.com>
Allow delegate_task to specify custom ACP transport per-task, so a parent
running via CLI/Discord/Telegram can spawn child agents over ACP
(e.g. claude --acp --stdio). Follows the existing override_provider pattern.
Supports per-task granularity in batch mode.
Co-authored-by: Mibayy <mibayy@users.noreply.github.com>
Route AIAgent print output to stderr via _print_fn for ACP stdio sessions.
Gate quiet-mode spinner startup on _should_start_quiet_spinner() so JSON-RPC
on stdout isn't corrupted. Child agents inherit the redirect.
Co-authored-by: Git-on-my-level <Git-on-my-level@users.noreply.github.com>
- Firecrawl scrape: 60s timeout via asyncio.wait_for + to_thread
(previously could hang indefinitely)
- Summarizer retries: 6 → 2 (one retry), reads timeout from
auxiliary.web_extract.timeout config (default 360s / 6min)
- Summarizer failure: falls back to truncated raw content (~5000 chars)
instead of useless error message, with guidance about config/model
- Config default: auxiliary.web_extract.timeout bumped 30 → 360s
for local model compatibility
Addresses Discord reports of agent hanging during web_extract.
Salvaged from PRs #3767 (chalkers), #5236 (ygd58), #2641 (buntingszn).
Three improvements to Matrix cron delivery:
1. Live adapter path: when the gateway is running, cron delivery now uses
the connected MatrixAdapter via run_coroutine_threadsafe instead of
the standalone HTTP PUT. This enables delivery to E2EE rooms where
the raw HTTP path cannot encrypt. Falls back to standalone on failure.
Threads adapters + event loop from gateway -> cron ticker -> tick() ->
_deliver_result(). (from #3767)
2. HTML formatted_body: _send_matrix() now converts markdown to HTML
using the optional markdown library, with h1-h6 to bold conversion
for Element X compatibility. Falls back to plain text if markdown
is not installed. Also adds random bytes to txn_id to prevent
collisions. (from #5236)
3. Origin fallback: when deliver="origin" but origin is null (jobs
created via API/scripts), falls back to HOME_CHANNEL env vars
in order: matrix -> telegram -> discord -> slack. (from #2641)
When commands like grep, diff, test, or find return non-zero exit codes
that aren't actual errors (grep 1 = no matches, diff 1 = files differ),
the model wastes turns investigating non-problems. This adds an
exit_code_meaning field to the terminal JSON result that explains
informational exit codes, so the agent can move on instead of debugging.
Covers grep/rg/ag/ack (no matches), diff (files differ), find (partial
access), test/[ (condition false), curl (timeouts, DNS, HTTP errors),
and git (context-dependent). Correctly extracts the last command from
pipelines and chains, strips full paths and env var assignments.
The exit_code field itself is unchanged — this is purely additive context.
When a dangerous command is approved (gateway, CLI, or smart approval),
the terminal tool now includes an 'approval' field in the result JSON
so the model knows approval was requested and granted. Previously the
model only saw normal command output with no indication that approval
happened, causing it to hallucinate that the approval system didn't fire.
Changes:
- approval.py: Return user_approved/description in all 3 approval paths
(gateway blocking, CLI interactive, smart approval)
- terminal_tool.py: Capture approval metadata and inject into both
foreground and background command results
* feat: execute_code runs on remote terminal backends (Docker/SSH/Modal/Daytona/Singularity)
When TERMINAL_ENV is not 'local', execute_code now ships the script to
the remote environment and runs it there via the terminal backend --
the same container/sandbox/SSH session used by terminal() and file tools.
Architecture:
- Local backend: unchanged (UDS RPC, subprocess.Popen)
- Remote backends: file-based RPC via execute_oneshot() polling
- Script writes request files, parent polls and dispatches tool calls
- Responses written atomically (tmp + rename) via base64/stdin
- execute_oneshot() bypasses persistent shell lock for concurrency
Changes:
- tools/environments/base.py: add execute_oneshot() (delegates to execute())
- tools/environments/persistent_shell.py: override execute_oneshot() to
bypass _shell_lock via _execute_oneshot(), enabling concurrent polling
- tools/code_execution_tool.py: add file-based transport to
generate_hermes_tools_module(), _execute_remote() with full env
get-or-create, file shipping, RPC poll loop, output post-processing
* fix: use _get_env_config() instead of raw TERMINAL_ENV env var
Read terminal backend type through the canonical config resolution
path (terminal_tool._get_env_config) instead of os.getenv directly.
* fix: use echo piping instead of stdin_data for base64 writes
Modal doesn't reliably deliver stdin_data to chained commands
(base64 -d > file && mv), producing 0-byte files. Switch to
echo 'base64' | base64 -d which works on all backends.
Verified E2E on both Docker and Modal.
Add an optional 'script' parameter to cron jobs that references a Python
script. The script runs before each agent turn, and its stdout is injected
into the prompt as context. This enables stateful monitoring — the script
handles data collection and change detection, the LLM analyzes and reports.
- cron/jobs.py: add script field to create_job(), stored in job dict
- cron/scheduler.py: add _run_job_script() executor with timeout handling,
inject script output/errors into _build_job_prompt()
- tools/cronjob_tools.py: add script to tool schema, create/update handlers,
_format_job display
- hermes_cli/cron.py: add --script to create/edit, display in list/edit output
- hermes_cli/main.py: add --script argparse for cron create/edit subcommands
- tests/cron/test_cron_script.py: 20 tests covering job CRUD, script
execution, path resolution, error handling, prompt injection, tool API
Script paths can be absolute or relative (resolved against ~/.hermes/scripts/).
Scripts run with a 120s timeout. Failures are injected as error context so
the LLM can report the problem. Empty string clears an attached script.
Bug fixes:
- agent/redact.py: catastrophic regex backtracking in _ENV_ASSIGN_RE — removed
re.IGNORECASE and changed [A-Z_]* to [A-Z0-9_]* to restrict matching to actual
env var name chars. Without this, the pattern backtracks exponentially on large
strings (e.g. 100K tool output), causing test_file_read_guards to time out.
- tools/file_operations.py: over-escaped newline in find -printf format string
produced literal backslash-n instead of a real newline, breaking file search
result parsing (total_count always 1, paths concatenated).
Test fixes:
- Remove stale pytestmark.skip from 4 test modules that were blanket-skipped as
'Hangs in non-interactive environments' but actually run fine:
- test_413_compression.py (12 tests, 25s)
- test_file_tools_live.py (71 tests, 24s)
- test_code_execution.py (61 tests, 99s)
- test_agent_loop_tool_calling.py (has proper OPENROUTER_API_KEY skip already)
- test_413_compression.py: fix threshold values in 2 preflight compression tests
where context_length was too small for the compressed output to fit in one pass.
- test_mcp_probe.py: add missing _MCP_AVAILABLE mock so tests work without MCP SDK.
- test_mcp_tool_issue_948.py: inject MCP symbols (StdioServerParameters etc.) when
SDK is not installed so patch() targets exist.
- test_approve_deny_commands.py: replace time.sleep(0.3) with deterministic polling
of _gateway_queues — fixes race condition where resolve fires before threads
register their approval entries, causing the test to hang indefinitely.
Net effect: +256 tests recovered from skip, 8 real failures fixed.
Add docker_env option to terminal config — a dict of key-value pairs that
get set inside Docker containers via -e flags at both container creation
(docker run) and per-command execution (docker exec) time.
This complements docker_forward_env (which reads values dynamically from
the host process environment). docker_env is useful when Hermes runs as a
systemd service without access to the user's shell environment — e.g.
setting SSH_AUTH_SOCK or GNUPGHOME to known stable paths for SSH/GPG
agent socket forwarding.
Precedence: docker_env provides baseline values; docker_forward_env
overrides for the same key.
Config example:
terminal:
docker_env:
SSH_AUTH_SOCK: /run/user/1000/ssh-agent.sock
GNUPGHOME: /root/.gnupg
docker_volumes:
- /run/user/1000/ssh-agent.sock:/run/user/1000/ssh-agent.sock
- /run/user/1000/gnupg/S.gpg-agent:/root/.gnupg/S.gpg-agent
Two pre-existing issues causing test_file_read_guards timeouts on CI:
1. agent/redact.py: _ENV_ASSIGN_RE used unbounded [A-Z_]* with
IGNORECASE, matching any letter/underscore to end-of-string at
each position → O(n²) backtracking on 100K+ char inputs.
Bounded to {0,50} since env var names are never that long.
2. tools/file_tools.py: redact_sensitive_text() ran BEFORE the
character-count guard, so oversized content (that would be rejected
anyway) went through the expensive regex first. Reordered to check
size limit before redaction.
The config key skills.external_dirs and core resolution (get_all_skills_dirs,
get_external_skills_dirs in agent/skill_utils.py) already existed but several
code paths still only scanned SKILLS_DIR. Now external dirs are respected
everywhere:
- skills_categories(): scan all dirs for category discovery
- _get_category_from_path(): resolve categories against any skills root
- skill_manager_tool._find_skill(): search all dirs for edit/patch/delete
- credential_files.get_skills_directory_mount(): mount all dirs into
Docker/Singularity containers (external dirs at external_skills/<idx>)
- credential_files.iter_skills_files(): list files from all dirs for
Modal/Daytona upload
- tools/environments/ssh.py: rsync all skill dirs to remote hosts
- gateway _check_unavailable_skill(): check disabled skills across all dirs
Usage in config.yaml:
skills:
external_dirs:
- ~/repos/agent-skills/hermes
- /shared/team-skills
- Add contextvars.Token[str] type hints to set/reset_current_session_key
- Use get_current_session_key(default='') in terminal_tool.py for background
process session tracking, fixing the same env var race for concurrent
gateway sessions spawning background processes
- Add .zip to SUPPORTED_DOCUMENT_TYPES so gateway platforms (Telegram,
Slack, Discord) cache uploaded zip files instead of rejecting them.
- Add get_cache_directory_mounts() and iter_cache_files() to
credential_files.py for host-side cache directory passthrough
(documents, images, audio, screenshots).
- Docker: bind-mount cache dirs read-only alongside credentials/skills.
Changes are live (bind mount semantics).
- Modal: mount cache files at sandbox creation + resync before each
command via _sync_files() with mtime+size change detection.
- Handles backward-compat with legacy dir names (document_cache,
image_cache, audio_cache, browser_screenshots) via get_hermes_dir().
- Container paths always use the new cache/<subdir> layout regardless
of host layout.
This replaces the need for a dedicated extract_archive tool (PR #4819)
— the agent can now use standard terminal commands (unzip, tar) on
uploaded files inside remote containers.
Closes: related to PR #4819 by kshitijk4poor
Three fixes for memory+profile isolation bugs:
1. memory_tool.py: Replace module-level MEMORY_DIR constant with
get_memory_dir() function that calls get_hermes_home() dynamically.
The old constant was cached at import time and could go stale if
HERMES_HOME changed after import. Internal MemoryStore methods now
call get_memory_dir() directly. MEMORY_DIR kept as backward-compat
alias.
2. profiles.py: profile create --clone now copies MEMORY.md and USER.md
from the source profile. These curated memory files are part of the
agent's identity (same as SOUL.md) and should carry over on clone.
3. holographic plugin: initialize() now expands $HERMES_HOME and
${HERMES_HOME} in the db_path config value, so users can write
'db_path: $HERMES_HOME/memory_store.db' and it resolves to the
active profile directory, not the default home.
Tests updated to mock get_memory_dir() alongside the legacy MEMORY_DIR.
Four fixes for MCP server stability issues reported by community member
(terminal lockup, zombie processes, escape sequence pollution, startup hang):
1. MCP reload timeout guard (cli.py): _check_config_mcp_changes now runs
_reload_mcp in a separate daemon thread with a 30s hard timeout. Previously,
a hung MCP server could block the process_loop thread indefinitely, freezing
the entire TUI (user can type but nothing happens, only Ctrl+D/Ctrl+\ work).
2. MCP stdio subprocess PID tracking (mcp_tool.py): Tracks child PIDs spawned
by stdio_client via before/after snapshots of /proc children. On shutdown,
_stop_mcp_loop force-kills any tracked PIDs that survived the SDK's graceful
SIGTERM→SIGKILL cleanup. Prevents zombie MCP server processes from
accumulating across sessions.
3. MCP event loop exception handler (mcp_tool.py): Installs
_mcp_loop_exception_handler on the MCP background event loop — same pattern
as the existing _suppress_closed_loop_errors on prompt_toolkit's loop.
Suppresses benign 'Event loop is closed' RuntimeError from httpx transport
__del__ during MCP shutdown. Salvaged from PR #2538 (acsezen).
4. MCP OAuth non-blocking (mcp_oauth.py): Replaces blocking input() call in
_wait_for_callback with OAuthNonInteractiveError raise. Adds _is_interactive()
TTY detection. In non-interactive environments, build_oauth_auth() still
returns a provider (cached tokens + refresh work), but the callback handler
raises immediately instead of blocking the MCP event loop for 120s. Re-raises
OAuth setup failures in _run_http so failed servers are reported cleanly
without blocking others. Salvaged from PRs #4521 (voidborne-d) and #4465
(heathley).
Closes#2537, closes#4462
Related: #4128, #3436
- Fix cron ThreadPoolExecutor blocking on timeout: use shutdown(wait=False,
cancel_futures=True) instead of context manager that waits indefinitely
- Extract _dequeue_pending_text() to deduplicate media-placeholder logic
in interrupt and normal-completion dequeue paths
- Remove hasattr guards for _running_agents_ts: add class-level default
so partial test construction works without scattered defensive checks
- Move `import concurrent.futures` to top of cron/scheduler.py
- Progress throttle: sleep remaining interval instead of busy-looping
0.1s (~15 wakeups per 1.5s window → 1 wakeup)
- Deduplicate _load_stt_config() in transcription_tools.py:
_has_openai_audio_backend() now delegates to _resolve_openai_audio_client_config()
Three targeted fixes from user-reported issues:
1. STT config resolution (transcription_tools.py):
_has_openai_audio_backend() and _resolve_openai_audio_client_config()
now check stt.openai.api_key/base_url in config.yaml FIRST, before
falling back to env vars. Fixes voice transcription breaking when
using a custom OpenAI-compatible endpoint via config.yaml.
2. Stream consumer flood control fallback (stream_consumer.py):
When an edit fails mid-stream (e.g., Telegram flood control returns
failure for waits >5s), reset _already_sent to False so the normal
final send path delivers the complete response. Previously, a
truncated partial was left as the final message.
3. Telegram edit_message comment alignment (telegram.py):
Clarify that long flood waits return failure so streaming can fall
back to a normal final send.
The PR changed prev_tools from list[str] to list[dict] with name/result
keys. The gateway's _step_callback_sync passed this directly to hooks
as 'tool_names', breaking user-authored hooks that call
', '.join(tool_names).
Now:
- 'tool_names' always contains strings (backward-compatible)
- 'tools' carries the enriched dicts for hooks that want results
Also adds summary logging to register_mcp_servers() and comprehensive
tests for all three PR changes:
- sanitize_mcp_name_component edge cases
- register_mcp_servers public API
- _register_session_mcp_servers ACP integration
- step_callback result forwarding
- gateway normalization backward compat
ACP clients pass MCP server definitions in session/new, load_session,
resume_session, and fork_session. Previously these were accepted but
silently ignored — the agent never connected to them.
This wires the mcp_servers parameter into the existing MCP registration
pipeline (tools/mcp_tool.py) so client-provided servers are connected,
their tools discovered, and the agent's tool surface refreshed before
the first prompt.
Changes:
tools/mcp_tool.py:
- Extract sanitize_mcp_name_component() to replace all non-[A-Za-z0-9_]
characters (fixes crash when server names contain / or other chars
that violate provider tool-name validation rules)
- Use it in _convert_mcp_schema, _sync_mcp_toolsets, _build_utility_schemas
- Extract register_mcp_servers(servers: dict) as a public API that takes
an explicit {name: config} map. discover_mcp_tools() becomes a thin
wrapper that loads config.yaml and calls register_mcp_servers()
acp_adapter/server.py:
- Add _register_session_mcp_servers() which converts ACP McpServerStdio /
McpServerHttp / McpServerSse objects to Hermes MCP config dicts,
registers them via asyncio.to_thread (avoids blocking the ACP event
loop), then rebuilds agent.tools, valid_tool_names, and invalidates
the cached system prompt
- Call it from new_session, load_session, resume_session, fork_session
Tested with Eden (theproxycompany.com) as ACP client — 5 MCP servers
(HTTP + stdio) registered successfully, 110 tools available to the agent.
* feat(memory): add pluggable memory provider interface with profile isolation
Introduces a pluggable MemoryProvider ABC so external memory backends can
integrate with Hermes without modifying core files. Each backend becomes a
plugin implementing a standard interface, orchestrated by MemoryManager.
Key architecture:
- agent/memory_provider.py — ABC with core + optional lifecycle hooks
- agent/memory_manager.py — single integration point in the agent loop
- agent/builtin_memory_provider.py — wraps existing MEMORY.md/USER.md
Profile isolation fixes applied to all 6 shipped plugins:
- Cognitive Memory: use get_hermes_home() instead of raw env var
- Hindsight Memory: check $HERMES_HOME/hindsight/config.json first,
fall back to legacy ~/.hindsight/ for backward compat
- Hermes Memory Store: replace hardcoded ~/.hermes paths with
get_hermes_home() for config loading and DB path defaults
- Mem0 Memory: use get_hermes_home() instead of raw env var
- RetainDB Memory: auto-derive profile-scoped project name from
hermes_home path (hermes-<profile>), explicit env var overrides
- OpenViking Memory: read-only, no local state, isolation via .env
MemoryManager.initialize_all() now injects hermes_home into kwargs so
every provider can resolve profile-scoped storage without importing
get_hermes_home() themselves.
Plugin system: adds register_memory_provider() to PluginContext and
get_plugin_memory_providers() accessor.
Based on PR #3825. 46 tests (37 unit + 5 E2E + 4 plugin registration).
* refactor(memory): drop cognitive plugin, rewrite OpenViking as full provider
Remove cognitive-memory plugin (#727) — core mechanics are broken:
decay runs 24x too fast (hourly not daily), prefetch uses row ID as
timestamp, search limited by importance not similarity.
Rewrite openviking-memory plugin from a read-only search wrapper into
a full bidirectional memory provider using the complete OpenViking
session lifecycle API:
- sync_turn: records user/assistant messages to OpenViking session
(threaded, non-blocking)
- on_session_end: commits session to trigger automatic memory extraction
into 6 categories (profile, preferences, entities, events, cases,
patterns)
- prefetch: background semantic search via find() endpoint
- on_memory_write: mirrors built-in memory writes to the session
- is_available: checks env var only, no network calls (ABC compliance)
Tools expanded from 3 to 5:
- viking_search: semantic search with mode/scope/limit
- viking_read: tiered content (abstract ~100tok / overview ~2k / full)
- viking_browse: filesystem-style navigation (list/tree/stat)
- viking_remember: explicit memory storage via session
- viking_add_resource: ingest URLs/docs into knowledge base
Uses direct HTTP via httpx (no openviking SDK dependency needed).
Response truncation on viking_read to prevent context flooding.
* fix(memory): harden Mem0 plugin — thread safety, non-blocking sync, circuit breaker
- Remove redundant mem0_context tool (identical to mem0_search with
rerank=true, top_k=5 — wastes a tool slot and confuses the model)
- Thread sync_turn so it's non-blocking — Mem0's server-side LLM
extraction can take 5-10s, was stalling the agent after every turn
- Add threading.Lock around _get_client() for thread-safe lazy init
(prefetch and sync threads could race on first client creation)
- Add circuit breaker: after 5 consecutive API failures, pause calls
for 120s instead of hammering a down server every turn. Auto-resets
after cooldown. Logs a warning when tripped.
- Track success/failure in prefetch, sync_turn, and all tool calls
- Wait for previous sync to finish before starting a new one (prevents
unbounded thread accumulation on rapid turns)
- Clean up shutdown to join both prefetch and sync threads
* fix(memory): enforce single external memory provider limit
MemoryManager now rejects a second non-builtin provider with a warning.
Built-in memory (MEMORY.md/USER.md) is always accepted. Only ONE
external plugin provider is allowed at a time. This prevents tool
schema bloat (some providers add 3-5 tools each) and conflicting
memory backends.
The warning message directs users to configure memory.provider in
config.yaml to select which provider to activate.
Updated all 47 tests to use builtin + one external pattern instead
of multiple externals. Added test_second_external_rejected to verify
the enforcement.
* feat(memory): add ByteRover memory provider plugin
Implements the ByteRover integration (from PR #3499 by hieuntg81) as a
MemoryProvider plugin instead of direct run_agent.py modifications.
ByteRover provides persistent memory via the brv CLI — a hierarchical
knowledge tree with tiered retrieval (fuzzy text then LLM-driven search).
Local-first with optional cloud sync.
Plugin capabilities:
- prefetch: background brv query for relevant context
- sync_turn: curate conversation turns (threaded, non-blocking)
- on_memory_write: mirror built-in memory writes to brv
- on_pre_compress: extract insights before context compression
Tools (3):
- brv_query: search the knowledge tree
- brv_curate: store facts/decisions/patterns
- brv_status: check CLI version and context tree state
Profile isolation: working directory at $HERMES_HOME/byterover/ (scoped
per profile). Binary resolution cached with thread-safe double-checked
locking. All write operations threaded to avoid blocking the agent
(curate can take 120s with LLM processing).
* fix(memory): thread remaining sync_turns, fix holographic, add config key
Plugin fixes:
- Hindsight: thread sync_turn (was blocking up to 30s via _run_in_thread)
- RetainDB: thread sync_turn (was blocking on HTTP POST)
- Both: shutdown now joins sync threads alongside prefetch threads
Holographic retrieval fixes:
- reason(): removed dead intersection_key computation (bundled but never
used in scoring). Now reuses pre-computed entity_residuals directly,
moved role_content encoding outside the inner loop.
- contradict(): added _MAX_CONTRADICT_FACTS=500 scaling guard. Above
500 facts, only checks the most recently updated ones to avoid O(n^2)
explosion (~125K comparisons at 500 is acceptable).
Config:
- Added memory.provider key to DEFAULT_CONFIG ("" = builtin only).
No version bump needed (deep_merge handles new keys automatically).
* feat(memory): extract Honcho as a MemoryProvider plugin
Creates plugins/honcho-memory/ as a thin adapter over the existing
honcho_integration/ package. All 4 Honcho tools (profile, search,
context, conclude) move from the normal tool registry to the
MemoryProvider interface.
The plugin delegates all work to HonchoSessionManager — no Honcho
logic is reimplemented. It uses the existing config chain:
$HERMES_HOME/honcho.json -> ~/.honcho/config.json -> env vars.
Lifecycle hooks:
- initialize: creates HonchoSessionManager via existing client factory
- prefetch: background dialectic query
- sync_turn: records messages + flushes to API (threaded)
- on_memory_write: mirrors user profile writes as conclusions
- on_session_end: flushes all pending messages
This is a prerequisite for the MemoryManager wiring in run_agent.py.
Once wired, Honcho goes through the same provider interface as all
other memory plugins, and the scattered Honcho code in run_agent.py
can be consolidated into the single MemoryManager integration point.
* feat(memory): wire MemoryManager into run_agent.py
Adds 8 integration points for the external memory provider plugin,
all purely additive (zero existing code modified):
1. Init (~L1130): Create MemoryManager, find matching plugin provider
from memory.provider config, initialize with session context
2. Tool injection (~L1160): Append provider tool schemas to self.tools
and self.valid_tool_names after memory_manager init
3. System prompt (~L2705): Add external provider's system_prompt_block
alongside existing MEMORY.md/USER.md blocks
4. Tool routing (~L5362): Route provider tool calls through
memory_manager.handle_tool_call() before the catchall handler
5. Memory write bridge (~L5353): Notify external provider via
on_memory_write() when the built-in memory tool writes
6. Pre-compress (~L5233): Call on_pre_compress() before context
compression discards messages
7. Prefetch (~L6421): Inject provider prefetch results into the
current-turn user message (same pattern as Honcho turn context)
8. Turn sync + session end (~L8161, ~L8172): sync_all() after each
completed turn, queue_prefetch_all() for next turn, on_session_end()
+ shutdown_all() at conversation end
All hooks are wrapped in try/except — a failing provider never breaks
the agent. The existing memory system, Honcho integration, and all
other code paths are completely untouched.
Full suite: 7222 passed, 4 pre-existing failures.
* refactor(memory): remove legacy Honcho integration from core
Extracts all Honcho-specific code from run_agent.py, model_tools.py,
toolsets.py, and gateway/run.py. Honcho is now exclusively available
as a memory provider plugin (plugins/honcho-memory/).
Removed from run_agent.py (-457 lines):
- Honcho init block (session manager creation, activation, config)
- 8 Honcho methods: _honcho_should_activate, _strip_honcho_tools,
_activate_honcho, _register_honcho_exit_hook, _queue_honcho_prefetch,
_honcho_prefetch, _honcho_save_user_observation, _honcho_sync
- _inject_honcho_turn_context module-level function
- Honcho system prompt block (tool descriptions, CLI commands)
- Honcho context injection in api_messages building
- Honcho params from __init__ (honcho_session_key, honcho_manager,
honcho_config)
- HONCHO_TOOL_NAMES constant
- All honcho-specific tool dispatch forwarding
Removed from other files:
- model_tools.py: honcho_tools import, honcho params from handle_function_call
- toolsets.py: honcho toolset definition, honcho tools from core tools list
- gateway/run.py: honcho params from AIAgent constructor calls
Removed tests (-339 lines):
- 9 Honcho-specific test methods from test_run_agent.py
- TestHonchoAtexitFlush class from test_exit_cleanup_interrupt.py
Restored two regex constants (_SURROGATE_RE, _BUDGET_WARNING_RE) that
were accidentally removed during the honcho function extraction.
The honcho_integration/ package is kept intact — the plugin delegates
to it. tools/honcho_tools.py registry entries are now dead code (import
commented out in model_tools.py) but the file is preserved for reference.
Full suite: 7207 passed, 4 pre-existing failures. Zero regressions.
* refactor(memory): restructure plugins, add CLI, clean gateway, migration notice
Plugin restructure:
- Move all memory plugins from plugins/<name>-memory/ to plugins/memory/<name>/
(byterover, hindsight, holographic, honcho, mem0, openviking, retaindb)
- New plugins/memory/__init__.py discovery module that scans the directory
directly, loading providers by name without the general plugin system
- run_agent.py uses load_memory_provider() instead of get_plugin_memory_providers()
CLI wiring:
- hermes memory setup — interactive curses picker + config wizard
- hermes memory status — show active provider, config, availability
- hermes memory off — disable external provider (built-in only)
- hermes honcho — now shows migration notice pointing to hermes memory setup
Gateway cleanup:
- Remove _get_or_create_gateway_honcho (already removed in prev commit)
- Remove _shutdown_gateway_honcho and _shutdown_all_gateway_honcho methods
- Remove all calls to shutdown methods (4 call sites)
- Remove _honcho_managers/_honcho_configs dict references
Dead code removal:
- Delete tools/honcho_tools.py (279 lines, import was already commented out)
- Delete tests/gateway/test_honcho_lifecycle.py (131 lines, tested removed methods)
- Remove if False placeholder from run_agent.py
Migration:
- Honcho migration notice on startup: detects existing honcho.json or
~/.honcho/config.json, prints guidance to run hermes memory setup.
Only fires when memory.provider is not set and not in quiet mode.
Full suite: 7203 passed, 4 pre-existing failures. Zero regressions.
* feat(memory): standardize plugin config + add per-plugin documentation
Config architecture:
- Add save_config(values, hermes_home) to MemoryProvider ABC
- Honcho: writes to $HERMES_HOME/honcho.json (SDK native)
- Mem0: writes to $HERMES_HOME/mem0.json
- Hindsight: writes to $HERMES_HOME/hindsight/config.json
- Holographic: writes to config.yaml under plugins.hermes-memory-store
- OpenViking/RetainDB/ByteRover: env-var only (default no-op)
Setup wizard (hermes memory setup):
- Now calls provider.save_config() for non-secret config
- Secrets still go to .env via env vars
- Only memory.provider activation key goes to config.yaml
Documentation:
- README.md for each of the 7 providers in plugins/memory/<name>/
- Requirements, setup (wizard + manual), config reference, tools table
- Consistent format across all providers
The contract for new memory plugins:
- get_config_schema() declares all fields (REQUIRED)
- save_config() writes native config (REQUIRED if not env-var-only)
- Secrets use env_var field in schema, written to .env by wizard
- README.md in the plugin directory
* docs: add memory providers user guide + developer guide
New pages:
- user-guide/features/memory-providers.md — comprehensive guide covering
all 7 shipped providers (Honcho, OpenViking, Mem0, Hindsight,
Holographic, RetainDB, ByteRover). Each with setup, config, tools,
cost, and unique features. Includes comparison table and profile
isolation notes.
- developer-guide/memory-provider-plugin.md — how to build a new memory
provider plugin. Covers ABC, required methods, config schema,
save_config, threading contract, profile isolation, testing.
Updated pages:
- user-guide/features/memory.md — replaced Honcho section with link to
new Memory Providers page
- user-guide/features/honcho.md — replaced with migration redirect to
the new Memory Providers page
- sidebars.ts — added both new pages to navigation
* fix(memory): auto-migrate Honcho users to memory provider plugin
When honcho.json or ~/.honcho/config.json exists but memory.provider
is not set, automatically set memory.provider: honcho in config.yaml
and activate the plugin. The plugin reads the same config files, so
all data and credentials are preserved. Zero user action needed.
Persists the migration to config.yaml so it only fires once. Prints
a one-line confirmation in non-quiet mode.
* fix(memory): only auto-migrate Honcho when enabled + credentialed
Check HonchoClientConfig.enabled AND (api_key OR base_url) before
auto-migrating — not just file existence. Prevents false activation
for users who disabled Honcho, stopped using it (config lingers),
or have ~/.honcho/ from a different tool.
* feat(memory): auto-install pip dependencies during hermes memory setup
Reads pip_dependencies from plugin.yaml, checks which are missing,
installs them via pip before config walkthrough. Also shows install
guidance for external_dependencies (e.g. brv CLI for ByteRover).
Updated all 7 plugin.yaml files with pip_dependencies:
- honcho: honcho-ai
- mem0: mem0ai
- openviking: httpx
- hindsight: hindsight-client
- holographic: (none)
- retaindb: requests
- byterover: (external_dependencies for brv CLI)
* fix: remove remaining Honcho crash risks from cli.py and gateway
cli.py: removed Honcho session re-mapping block (would crash importing
deleted tools/honcho_tools.py), Honcho flush on compress, Honcho
session display on startup, Honcho shutdown on exit, honcho_session_key
AIAgent param.
gateway/run.py: removed honcho_session_key params from helper methods,
sync_honcho param, _honcho.shutdown() block.
tests: fixed test_cron_session_with_honcho_key_skipped (was passing
removed honcho_key param to _flush_memories_for_session).
* fix: include plugins/ in pyproject.toml package list
Without this, plugins/memory/ wouldn't be included in non-editable
installs. Hermes always runs from the repo checkout so this is belt-
and-suspenders, but prevents breakage if the install method changes.
* fix(memory): correct pip-to-import name mapping for dep checks
The heuristic dep.replace('-', '_') fails for packages where the pip
name differs from the import name: honcho-ai→honcho, mem0ai→mem0,
hindsight-client→hindsight_client. Added explicit mapping table so
hermes memory setup doesn't try to reinstall already-installed packages.
* chore: remove dead code from old plugin memory registration path
- hermes_cli/plugins.py: removed register_memory_provider(),
_memory_providers list, get_plugin_memory_providers() — memory
providers now use plugins/memory/ discovery, not the general plugin system
- hermes_cli/main.py: stripped 74 lines of dead honcho argparse
subparsers (setup, status, sessions, map, peer, mode, tokens,
identity, migrate) — kept only the migration redirect
- agent/memory_provider.py: updated docstring to reflect new
registration path
- tests: replaced TestPluginMemoryProviderRegistration with
TestPluginMemoryDiscovery that tests the actual plugins/memory/
discovery system. Added 3 new tests (discover, load, nonexistent).
* chore: delete dead honcho_integration/cli.py and its tests
cli.py (794 lines) was the old 'hermes honcho' command handler — nobody
calls it since cmd_honcho was replaced with a migration redirect.
Deleted tests that imported from removed code:
- tests/honcho_integration/test_cli.py (tested _resolve_api_key)
- tests/honcho_integration/test_config_isolation.py (tested CLI config paths)
- tests/tools/test_honcho_tools.py (tested the deleted tools/honcho_tools.py)
Remaining honcho_integration/ files (actively used by the plugin):
- client.py (445 lines) — config loading, SDK client creation
- session.py (991 lines) — session management, queries, flush
* refactor: move honcho_integration/ into the honcho plugin
Moves client.py (445 lines) and session.py (991 lines) from the
top-level honcho_integration/ package into plugins/memory/honcho/.
No Honcho code remains in the main codebase.
- plugins/memory/honcho/client.py — config loading, SDK client creation
- plugins/memory/honcho/session.py — session management, queries, flush
- Updated all imports: run_agent.py (auto-migration), hermes_cli/doctor.py,
plugin __init__.py, session.py cross-import, all tests
- Removed honcho_integration/ package and pyproject.toml entry
- Renamed tests/honcho_integration/ → tests/honcho_plugin/
* docs: update architecture + gateway-internals for memory provider system
- architecture.md: replaced honcho_integration/ with plugins/memory/
- gateway-internals.md: replaced Honcho-specific session routing and
flush lifecycle docs with generic memory provider interface docs
* fix: update stale mock path for resolve_active_host after honcho plugin migration
* fix(memory): address review feedback — P0 lifecycle, ABC contract, honcho CLI restore
Review feedback from Honcho devs (erosika):
P0 — Provider lifecycle:
- Remove on_session_end() + shutdown_all() from run_conversation() tail
(was killing providers after every turn in multi-turn sessions)
- Add shutdown_memory_provider() method on AIAgent for callers
- Wire shutdown into CLI atexit, reset_conversation, gateway stop/expiry
Bug fixes:
- Remove sync_honcho=False kwarg from /btw callsites (TypeError crash)
- Fix doctor.py references to dead 'hermes honcho setup' command
- Cache prefetch_all() before tool loop (was re-calling every iteration)
ABC contract hardening (all backwards-compatible):
- Add session_id kwarg to prefetch/sync_turn/queue_prefetch
- Make on_pre_compress() return str (provider insights in compression)
- Add **kwargs to on_turn_start() for runtime context
- Add on_delegation() hook for parent-side subagent observation
- Document agent_context/agent_identity/agent_workspace kwargs on
initialize() (prevents cron corruption, enables profile scoping)
- Fix docstring: single external provider, not multiple
Honcho CLI restoration:
- Add plugins/memory/honcho/cli.py (from main's honcho_integration/cli.py
with imports adapted to plugin path)
- Restore full hermes honcho command with all subcommands (status, peer,
mode, tokens, identity, enable/disable, sync, peers, --target-profile)
- Restore auto-clone on profile creation + sync on hermes update
- hermes honcho setup now redirects to hermes memory setup
* fix(memory): wire on_delegation, skip_memory for cron/flush, fix ByteRover return type
- Wire on_delegation() in delegate_tool.py — parent's memory provider
is notified with task+result after each subagent completes
- Add skip_memory=True to cron scheduler (prevents cron system prompts
from corrupting user representations — closes#4052)
- Add skip_memory=True to gateway flush agent (throwaway agent shouldn't
activate memory provider)
- Fix ByteRover on_pre_compress() return type: None -> str
* fix(honcho): port profile isolation fixes from PR #4632
Ports 5 bug fixes found during profile testing (erosika's PR #4632):
1. 3-tier config resolution — resolve_config_path() now checks
$HERMES_HOME/honcho.json → ~/.hermes/honcho.json → ~/.honcho/config.json
(non-default profiles couldn't find shared host blocks)
2. Thread host=_host_key() through from_global_config() in cmd_setup,
cmd_status, cmd_identity (--target-profile was being ignored)
3. Use bare profile name as aiPeer (not host key with dots) — Honcho's
peer ID pattern is ^[a-zA-Z0-9_-]+$, dots are invalid
4. Wrap add_peers() in try/except — was fatal on new AI peers, killed
all message uploads for the session
5. Gate Honcho clone behind --clone/--clone-all on profile create
(bare create should be blank-slate)
Also: sanitize assistant_peer_id via _sanitize_id()
* fix(tests): add module cleanup fixture to test_cli_provider_resolution
test_cli_provider_resolution._import_cli() wipes tools.*, cli, and
run_agent from sys.modules to force fresh imports, but had no cleanup.
This poisoned all subsequent tests on the same xdist worker — mocks
targeting tools.file_tools, tools.send_message_tool, etc. patched the
NEW module object while already-imported functions still referenced
the OLD one. Caused ~25 cascade failures: send_message KeyError,
process_registry FileNotFoundError, file_read_guards timeouts,
read_loop_detection file-not-found, mcp_oauth None port, and
provider_parity/codex_execution stale tool lists.
Fix: autouse fixture saves all affected modules before each test and
restores them after, matching the pattern in
test_managed_browserbase_and_modal.py.
The gateway's dangerous command approval system was fundamentally broken:
the agent loop continued running after a command was flagged, and the
approval request only reached the user after the agent finished its
entire conversation loop. By then the context was lost.
This change makes the gateway approval mirror the CLI's synchronous
behavior. When a dangerous command is detected:
1. The agent thread blocks on a threading.Event
2. The approval request is sent to the user immediately
3. The user responds with /approve or /deny
4. The event is signaled and the agent resumes with the real result
The agent never sees 'approval_required' as a tool result. It either
gets the command output (approved) or a definitive BLOCKED message
(denied/timed out) — same as CLI mode.
Queue-based design supports multiple concurrent approvals (parallel
subagents via delegate_task, execute_code RPC handlers). Each approval
gets its own _ApprovalEntry with its own threading.Event. /approve
resolves the oldest (FIFO); /approve all resolves all at once.
Changes:
- tools/approval.py: Queue-based per-session blocking gateway approval
(register/unregister callbacks, resolve with FIFO or all-at-once)
- gateway/run.py: Register approval callback in run_sync(), remove
post-loop pop_pending hack, /approve and /deny support 'all' flag
- tests: 21 tests including parallel subagent E2E scenarios
- Add logger + debug log to read_nous_access_token() catch-all so token
refresh failures are observable instead of silently swallowed
- Tighten _is_nous_auxiliary_client() domain check to use proper URL
hostname parsing instead of substring match, preventing false-positives
on domains like not-nousresearch.com or nousresearch.com.evil.com
LLM responses from browser snapshot extraction and vision analysis
could echo back secrets that appeared on screen or in page content.
Input redaction alone is insufficient — the LLM may reproduce secrets
it read from screenshots (which cannot be text-redacted).
Now redact outputs from:
- _extract_relevant_content (auxiliary LLM response)
- browser_vision (vision LLM response)
- camofox_vision (vision LLM response)
Three exfiltration vectors closed:
1. Browser URL exfil — agent could embed secrets in URL params and
navigate to attacker-controlled server. Now scans URLs for known
API key patterns before navigating (browser_navigate, web_extract).
2. Browser snapshot leak — page displaying env vars or API keys would
send secrets to auxiliary LLM via _extract_relevant_content before
run_agent.py's redaction layer sees the result. Now redacts snapshot
text before the auxiliary call.
3. Camofox annotation leak — accessibility tree text sent to vision
LLM could contain secrets visible on screen. Now redacts annotation
context before the vision call.
10 new tests covering URL blocking, snapshot redaction, and annotation
redaction for both browser and camofox backends.
* feat(skills): add content size limits for agent-created skills
Agent writes via skill_manage (create/edit/patch/write_file) are now
constrained to prevent unbounded growth:
- SKILL.md and supporting files: 100,000 character limit
- Supporting files: additional 1 MiB byte limit
- Patches on oversized hand-placed skills that reduce the size are
allowed (shrink path), but patches that grow beyond the limit are
rejected
Hand-placed skills and hub-installed skills have NO hard limit —
they load and function normally regardless of size. Hub installs
get a warning in the log if SKILL.md exceeds 100k chars.
This mirrors the memory system's char_limit pattern. Without this,
the agent auto-grows skills indefinitely through iterative patches
(hermes-agent-dev reached 197k chars / 72k tokens — 40x larger than
the largest skill in the entire skills.sh ecosystem).
Constants: MAX_SKILL_CONTENT_CHARS (100k), MAX_SKILL_FILE_BYTES (1MiB)
Tests: 14 new tests covering all write paths and edge cases
* feat(skills): add fuzzy matching to skill patch
_patch_skill now uses the same 8-strategy fuzzy matching engine
(tools/fuzzy_match.py) as the file patch tool. Handles whitespace
normalization, indentation differences, escape sequences, and
block-anchor matching. Eliminates exact-match failures when agents
patch skills with minor formatting mismatches.
Adds two Camofox features:
1. Persistent browser sessions: new `browser.camofox.managed_persistence`
config option. When enabled, Hermes sends a deterministic profile-scoped
userId to Camofox so the server maps it to a persistent browser profile
directory. Cookies, logins, and browser state survive across restarts.
Default remains ephemeral (random userId per session).
2. VNC URL discovery: Camofox /health endpoint returns vncPort when running
in headed mode. Hermes constructs the VNC URL and includes it in navigate
responses so the agent can share it with users.
Also fixes camofox_vision bug where call_llm response object was passed
directly to json.dumps instead of extracting .choices[0].message.content.
Changes from original PR:
- Removed browser_evaluate tool (separate feature, needs own PR)
- Removed snapshot truncation limit change (unrelated)
- Config.yaml only for managed_persistence (no env var, no version bump)
- Rewrote tests to use config mock instead of env var
- Reverted package-lock.json churn
Co-authored-by: analista <psikonetik@gmail.com.com>
After a successful write_file or patch, update the stored read
timestamp to match the file's new modification time. Without this,
consecutive edits by the same task (read → write → write) would
false-warn on the second write because the stored timestamp still
reflected the original read, not the first write.
Also renames the internal tracker key from 'file_mtimes' to
'read_timestamps' for clarity.
* fix: root-level provider in config.yaml no longer overrides model.provider
load_cli_config() had a priority inversion: a stale root-level
'provider' key in config.yaml would OVERRIDE the canonical
'model.provider' set by 'hermes model'. The gateway reads
model.provider directly from YAML and worked correctly, but
'hermes chat -q' and the interactive CLI went through the merge
logic and picked up the stale root-level key.
Fix: root-level provider/base_url are now only used as a fallback
when model.provider/model.base_url is not set (never as an override).
Also added _normalize_root_model_keys() to config.py load_config()
and save_config() — migrates root-level provider/base_url into the
model section and removes the root-level keys permanently.
Reported by (≧▽≦) in Discord: opencode-go provider persisted as a
root-level key and overrode the correct model.provider=openrouter,
causing 401 errors.
* fix(security): redact secrets from execute_code sandbox output
The execute_code sandbox stripped env vars with secret-like names from
the child process (preventing os.environ access), but scripts could
still read secrets from disk (e.g. open('~/.hermes/.env')) and print
them to stdout. The raw values entered the model context unredacted.
terminal_tool and file_tools already applied redact_sensitive_text()
to their output — execute_code was the only tool that skipped this
step. Now the same redaction runs on both stdout and stderr after
ANSI stripping.
Reported via Discord (not filed on GitHub to avoid public disclosure
of the reproduction steps).
Track file mtime when read_file is called. When write_file or patch
subsequently targets the same file, compare the current mtime against
the recorded one. If they differ (external edit, concurrent agent,
user change), include a _warning in the result advising the agent to
re-read. The write still proceeds — this is a soft signal, not a
hard block.
Key design points:
- Per-task isolation: task A's reads don't affect task B's writes.
- Files never read produce no warning (not enforcing read-before-write).
- mtime naturally updates after the agent's own writes, so the warning
only fires on external changes, not the agent's own edits.
- V4A multi-file patches check all target paths.
Tests: 10 new tests covering write staleness, patch staleness,
never-read files, cross-task isolation, and the helper function.
The camofox-browser image defaults to port 3000 internally, not 9377.
Without -e CAMOFOX_PORT=9377, the -p 9377:9377 mapping silently fails
because nothing listens on 9377 inside the container.
E2E verified: -p 9377:9377 alone → connection reset,
-p 9377:9377 -e CAMOFOX_PORT=9377 → healthy and functional.
* feat(file_tools): harden read_file with size guard, dedup, and device blocking
Three improvements to read_file_tool to reduce wasted context tokens and
prevent process hangs:
1. Character-count guard: reads that produce more than 100K characters
(≈25-35K tokens across tokenisers) are rejected with an error that
tells the model to use offset+limit for a smaller range. The
effective cap is min(file_size, 100K) so small files that happen to
have long lines aren't over-penalised. Large truncated files also
get a hint nudging toward targeted reads.
2. File-read deduplication: when the same (path, offset, limit) is read
a second time and the file hasn't been modified (mtime unchanged),
return a lightweight stub instead of re-sending the full content.
Writes and patches naturally change mtime, so post-edit reads always
return fresh content. The dedup cache is cleared on context
compression — after compression the original read content is
summarised away, so the model needs the full content again.
3. Device path blocking: paths like /dev/zero, /dev/random, /dev/stdin
etc. are rejected before any I/O to prevent process hangs from
infinite-output or blocking-input devices.
Tests: 17 new tests covering all three features plus the dedup-reset-
on-compression integration. All 52 file-read tests pass (35 existing +
17 new). Full tool suite (2124 tests) passes with 0 failures.
* feat: make file_read_max_chars configurable, add docs
Add file_read_max_chars to DEFAULT_CONFIG (default 100K). read_file_tool
reads this on first call and caches for the process lifetime. Users on
large-context models can raise it; users on small local models can lower it.
Also adds a 'File Read Safety' section to the configuration docs
explaining the char limit, dedup behavior, and example values.
_load_config_files() had the same hermes_home / item pattern without
containment checks. While config.yaml is user-controlled (lower threat
than skill frontmatter), defense in depth prevents exploitation via
config injection or copy-paste mistakes.
WSL detection was treated as a hard fail, blocking voice mode even when
audio worked via PulseAudio bridge. Now PULSE_SERVER env var presence
makes WSL a soft notice instead of a blocking warning. Device query
failures in WSL with PULSE_SERVER are also treated as non-blocking.
The SSRF protection added in #3041 blocks all private/internal addresses
unconditionally in browser_navigate(). This prevents legitimate local use
cases (localhost apps, LAN devices) when using Camofox or the built-in
headless Chromium without a cloud provider.
The check is only meaningful for cloud backends (Browserbase, BrowserUse)
where the agent could reach internal resources on a remote machine. Local
backends give the user full terminal and network access already — the
SSRF check adds zero security value.
Add _is_local_backend() helper that returns True when Camofox is active
or no cloud provider is configured. Both the pre-navigation and
post-redirect SSRF checks now skip when running locally. The
browser.allow_private_urls config option remains available as an
explicit opt-out for cloud mode.
1. browser_tool.py: _allow_private_urls() used 'global _allow_private_urls'
then assigned a bool to it, replacing the function in the module namespace.
After first call, subsequent calls hit TypeError: 'bool' object is not
callable. Renamed cache variable to _cached_allow_private_urls.
2. test_provider_parity.py: test_custom_endpoint_when_no_nous relied on
OPENAI_BASE_URL env var (removed in config refactor). Mock
_resolve_custom_runtime directly instead.
* fix(tools): skip SSRF check in local browser mode
The SSRF protection added in #3041 blocks all private/internal
addresses unconditionally in browser_navigate(). This prevents
legitimate local development use cases (localhost testing, LAN
device access) when using the local Chromium backend.
The SSRF check is only meaningful for cloud browsers (Browserbase,
BrowserUse) where the agent could reach internal resources on a
remote machine. In local mode, the user already has full terminal
and network access, so the check adds no security value.
This change makes the SSRF check conditional on _get_cloud_provider(),
keeping full protection in cloud mode while allowing private addresses
in local mode.
* fix(tools): make SSRF check configurable via browser.allow_private_urls
Replace unconditional SSRF check with a configurable setting.
Default (False) keeps existing security behavior. Setting to True
allows navigating to private/internal IPs for local dev and LAN use cases.
---------
Co-authored-by: Nils (Norya) <nils@begou.dev>
* fix(gateway): honor default for invalid bool-like config values
* refactor: simplify web backend priority detection
Replace cascading boolean conditions with a priority-ordered loop.
Same behavior (verified against all 16 env var combinations),
half the lines, trivially extensible for new backends.
---------
Co-authored-by: aydnOktay <xaydinoktay@gmail.com>
Camofox-browser is a self-hosted Node.js server wrapping Camoufox
(Firefox fork with C++ fingerprint spoofing). When CAMOFOX_URL is set,
all 11 browser tools route through the Camofox REST API instead of
the agent-browser CLI.
Maps 1:1 to the existing browser tool interface:
- Navigate, snapshot, click, type, scroll, back, press, close
- Get images, vision (screenshot + LLM analysis)
- Console (returns empty with note — camofox limitation)
Setup: npm start in camofox-browser dir, or docker run -p 9377:9377
Then: CAMOFOX_URL=http://localhost:9377 in ~/.hermes/.env
Advantages over Browserbase (cloud):
- Free (no per-session API costs)
- Local (zero network latency for browser ops)
- Anti-detection at C++ level (bypasses Cloudflare/Google bot detection)
- Works offline, Docker-ready
Files:
- tools/browser_camofox.py: Full REST backend (~400 lines)
- tools/browser_tool.py: Routing at each tool function
- hermes_cli/config.py: CAMOFOX_URL env var entry
- tests/tools/test_browser_camofox.py: 20 tests
* feat(telegram): add webhook mode as alternative to polling
When TELEGRAM_WEBHOOK_URL is set, the adapter starts an HTTP webhook
server (via python-telegram-bot's start_webhook()) instead of long
polling. This enables cloud platforms like Fly.io and Railway to
auto-wake suspended machines on inbound HTTP traffic.
Polling remains the default — no behavior change unless the env var
is set.
Env vars:
TELEGRAM_WEBHOOK_URL Public HTTPS URL for Telegram to push to
TELEGRAM_WEBHOOK_PORT Local listen port (default 8443)
TELEGRAM_WEBHOOK_SECRET Secret token for update verification
Cherry-picked and adapted from PR #2022 by SHL0MS. Preserved all
current main enhancements (network error recovery, polling conflict
detection, DM topics setup).
Co-authored-by: SHL0MS <SHL0MS@users.noreply.github.com>
* fix: send_document call in background task delivery + vision download timeout
Two fixes salvaged from PR #2269 by amethystani:
1. gateway/run.py: adapter.send_file() → adapter.send_document()
send_file() doesn't exist on BasePlatformAdapter. Background task
media files were silently never delivered (AttributeError swallowed
by except Exception: pass).
2. tools/vision_tools.py: configurable image download timeout via
HERMES_VISION_DOWNLOAD_TIMEOUT env var (default 30s), plus guard
against raise None when max_retries=0.
The third fix in #2269 (opencode-go auth config) was already resolved
on main.
Co-authored-by: amethystani <amethystani@users.noreply.github.com>
---------
Co-authored-by: SHL0MS <SHL0MS@users.noreply.github.com>
Co-authored-by: amethystani <amethystani@users.noreply.github.com>
Skills with scripts/, templates/, and references/ subdirectories need
those files available inside sandboxed execution environments. Previously
the skills directory was missing entirely from remote backends.
Live sync — files stay current as credentials refresh and skills update:
- Docker/Singularity: bind mounts are inherently live (host changes
visible immediately)
- Modal: _sync_files() runs before each command with mtime+size caching,
pushing only changed credential and skill files (~13μs no-op overhead)
- SSH: rsync --safe-links before each command (naturally incremental)
- Daytona: _upload_if_changed() with mtime+size caching before each command
Security — symlink filtering:
- Docker/Singularity: sanitized temp copy when symlinks detected
- Modal/Daytona: iter_skills_files() skips symlinks
- SSH: rsync --safe-links skips symlinks pointing outside source tree
- Temp dir cleanup via atexit + reuse across calls
Non-root user support:
- SSH: detects remote home via echo $HOME, syncs to $HOME/.hermes/
- Daytona: detects sandbox home before sync, uploads to $HOME/.hermes/
- Docker/Modal/Singularity: run as root, /root/.hermes/ is correct
Also:
- credential_files.py: fix name/path key fallback in required_credential_files
- Singularity, SSH, Daytona: gained credential file support
- 14 tests covering symlink filtering, name/path fallback, iter_skills_files
* feat(approvals): make dangerous command approval timeout configurable
Read `approvals.timeout` from config.yaml (default 60s) instead of
hardcoding 60 seconds in both the fallback CLI prompt and the TUI
prompt_toolkit callback.
Follows the same pattern as `clarify.timeout` which is already
configurable via CLI_CONFIG.
Closes#3765
* fix: add timeout default to approvals section in DEFAULT_CONFIG
---------
Co-authored-by: acsezen <asezen@icloud.com>
Closes gaps that allowed an agent to expose Docker's Remote API to the
internet by writing to /etc/docker/daemon.json.
Terminal tool (approval.py):
- chmod: now catches 666 and symbolic modes (o+w, a+w), not just 777
- cp/mv/install: detected when targeting /etc/
- sed -i/--in-place: detected when targeting /etc/
File tools (file_tools.py):
- write_file and patch now refuse to write to sensitive system paths
(/etc/, /boot/, /usr/lib/systemd/, docker.sock)
- Directs users to the terminal tool (which has approval prompts) for
system file modifications
When a command timed out, all captured output was discarded — the agent
only saw 'Command timed out after Xs' with zero context. Now returns
the buffered output followed by a timeout marker, matching the existing
interrupt path behavior.
Salvaged from PR #3286 by @binhnt92.
Co-authored-by: nguyen binh <binhnt92@users.noreply.github.com>
Adds WeCom as a gateway platform adapter using the AI Bot WebSocket
gateway for real-time bidirectional communication. No public endpoint
or new pip dependencies needed (uses existing aiohttp + httpx).
Features:
- WebSocket persistent connection with auto-reconnect (exponential backoff)
- DM and group messaging with configurable access policies
- Media upload/download with AES decryption for encrypted attachments
- Markdown rendering, quote context preservation
- Proactive + passive reply message modes
- Chunked media upload pipeline (512KB chunks)
Cherry-picked from PR #1898 by EvilRan with:
- Moved to current main (PR was 300 commits behind)
- Skipped base.py regressions (reply_to additions are good but belong
in a separate PR since they affect all platforms)
- Fixed test assertions to match current base class send() signature
(reply_to=None kwarg now explicit)
- All 16 integration points added surgically to current main
- No new pip dependencies (aiohttp + httpx already installed)
Fixes#1898
Co-authored-by: EvilRan <EvilRan@users.noreply.github.com>
Three safety gaps in vision_analyze_tool:
1. Local files accepted without checking if they're actually images —
a renamed text file would get base64-encoded and sent to the model.
Now validates magic bytes (PNG, JPEG, GIF, BMP, WebP, SVG).
2. No website policy enforcement on image URLs — blocked domains could
be fetched via the vision tool. Now checks before download.
3. No redirect check — if an allowed URL redirected to a blocked domain,
the download would proceed. Now re-checks the final URL.
Fixed one test that needed _validate_image_url mocked to bypass DNS
resolution on the fake blocked.test domain (is_safe_url does DNS
checks that were added after the original PR).
Co-authored-by: GutSlabs <GutSlabs@users.noreply.github.com>
Validate category names in _create_skill() before using them as
filesystem path segments. Previously, categories like '../escape' or
'/tmp/pwned' could write skill files outside ~/.hermes/skills/.
Adds _validate_category() that rejects slashes, backslashes, absolute
paths, and non-alphanumeric characters (reuses existing VALID_NAME_RE).
Tests: 5 new tests for traversal, absolute paths, and valid categories.
Salvaged from PR #1939 by Gutslabs.
When a subagent hit max_iterations, status was always 'failed' even
if it produced a usable summary via _handle_max_iterations(). This
happened because the status check required both completed=True AND
a summary, but completed is False whenever max_iterations is reached
(run_agent.py line 7969).
Now gates status on whether a summary was produced — if the subagent
returned a final_response, the parent has usable output regardless of
iteration budget. The exit_reason field already distinguishes
'completed' vs 'max_iterations' for anything that needs to know how
the task ended.
Closes#1899.
Adds Feishu (ByteDance's enterprise messaging platform) as a gateway
platform adapter with full feature parity: WebSocket + webhook transports,
message batching, dedup, rate limiting, rich post/card content parsing,
media handling (images/audio/files/video), group @mention gating,
reaction routing, and interactive card button support.
Cherry-picked from PR #1793 by penwyp with:
- Moved to current main (PR was 458 commits behind)
- Fixed _send_with_retry shadowing BasePlatformAdapter method (renamed to
_feishu_send_with_retry to avoid signature mismatch crash)
- Fixed import structure: aiohttp/websockets imported independently of
lark_oapi so they remain available when SDK is missing
- Fixed get_hermes_home import (hermes_constants, not hermes_cli.config)
- Added skip decorators for tests requiring lark_oapi SDK
- All 16 integration points added surgically to current main
New dependency: lark-oapi>=1.5.3,<2 (optional, pip install hermes-agent[feishu])
Fixes#1788
Co-authored-by: penwyp <penwyp@users.noreply.github.com>
The write-deny list in file_operations.py hardcoded ~/.hermes/.env,
which misses the actual .env in custom HERMES_HOME or profile setups.
Use get_hermes_home() for profile-safe path resolution.
Salvaged from PR #3232 by @erhnysr.
Co-authored-by: Erhnysr <erhnysr@users.noreply.github.com>
The honcho check_fn only checked runtime session state, which isn't
set until the agent initializes. At banner time, honcho tools showed
as red/disabled even when properly configured.
Now checks configuration (enabled + api_key/base_url) as a fallback
when the session context isn't active yet. Fast path (session active)
unchanged; slow path (config check) only runs at banner time.
Adds 4 tests covering: session active, configured but no session,
not configured, and import failure graceful fallback.
Closes#1843.
When a connected MCP server sends a ToolListChangedNotification (per the
MCP spec), Hermes now automatically re-fetches the tool list, deregisters
removed tools, and registers new ones — without requiring a restart.
This enables MCP servers with dynamic toolsets (e.g. GitHub MCP with
GITHUB_DYNAMIC_TOOLSETS=1) to add/remove tools at runtime.
Changes:
- registry.py: add ToolRegistry.deregister() for nuke-and-repave refresh
- mcp_tool.py: extract _register_server_tools() from
_discover_and_register_server() as a shared helper for both initial
discovery and dynamic refresh
- mcp_tool.py: add _make_message_handler() and _refresh_tools() on
MCPServerTask, wired into all 3 ClientSession sites (stdio, new HTTP,
deprecated HTTP)
- Graceful degradation: silently falls back to static discovery when the
MCP SDK lacks notification types or message_handler support
- 8 new tests covering registration, refresh, handler dispatch, and
deregister
Salvaged from PR #1794 by shivvor2.
When a tool plugin registers a schema without an explicit 'name' key,
get_definitions() crashes with KeyError:
available_tool_names = {t["function"]["name"] for t in filtered_tools}
Fix: always merge entry.name into schema so 'name' is never missing.
Refs: #3729
Co-authored-by: ekkoitac <ekko.itac@gmail.com>
* feat: show estimated tool token context in hermes tools checklist
Adds a live token estimate indicator to the bottom of the interactive
tool configuration checklist (hermes tools / hermes setup). As users
toggle toolsets on/off, the total estimated context cost updates in
real time.
Implementation:
- tools/registry.py: Add get_schema() for check_fn-free schema access
- hermes_cli/curses_ui.py: Add optional status_fn callback to
curses_checklist — renders at bottom-right of terminal, stays fixed
while items scroll
- hermes_cli/tools_config.py: Add _estimate_tool_tokens() using
tiktoken (cl100k_base, already installed) to count tokens in the
JSON-serialised OpenAI-format tool schemas. Results are cached
per-process. The status function deduplicates overlapping tools
(e.g. browser includes web_search) for accurate totals.
- 12 new tests covering estimation, caching, graceful degradation
when tiktoken is unavailable, status_fn wiring, deduplication,
and the numbered fallback display
* fix: use effective toolsets (includes plugins) for token estimation index mapping
The status_fn closure built ts_keys from CONFIGURABLE_TOOLSETS but the
checklist uses _get_effective_configurable_toolsets() which appends plugin
toolsets. With plugins present, the indices would mismatch, causing
IndexError when selecting a plugin toolset.
* fix(tools): implement send_message routing for Matrix, Mattermost, HomeAssistant, DingTalk
Matrix, Mattermost, HomeAssistant, and DingTalk were present in
platform_map but fell through to the "not yet implemented" else branch,
causing send_message tool calls to silently fail on these platforms.
Add four async sender functions:
- _send_mattermost: POST /api/v4/posts via Mattermost REST API
- _send_matrix: PUT /_matrix/client/v3/rooms/.../send via Matrix CS API
- _send_homeassistant: POST /api/services/notify/notify via HA REST API
- _send_dingtalk: POST to session webhook URL
Add routing in _send_to_platform() and 17 unit tests covering success,
HTTP errors, missing config, env var fallback, and Matrix txn_id uniqueness.
* fix: pass platform tokens explicitly to Mattermost/Matrix/HA senders
The original PR passed pconfig.extra to sender functions, but tokens
live at pconfig.token (not in extra). This caused the senders to always
fall through to env var lookup instead of using the gateway-resolved
token.
Changes:
- Mattermost/Matrix/HA: accept token as first arg, matching the
Telegram/Discord/Slack sender pattern
- DingTalk: add DINGTALK_WEBHOOK_URL env var fallback + docstring
explaining the session-webhook vs robot-webhook difference
- Tests updated for new signatures + new DingTalk env var test
---------
Co-authored-by: sprmn24 <oncuevtv@gmail.com>
When a user runs 'hermes update', the Python process caches old modules
in sys.modules. After git pull updates files on disk, lazy imports of
newly-updated modules fail because they try to import display_hermes_home
from the cached (old) hermes_constants which doesn't have the function.
This specifically broke the gateway auto-restart in cmd_update — importing
hermes_cli/gateway.py triggered the top-level 'from hermes_constants
import display_hermes_home' against the cached old module. The ImportError
was silently caught, so the gateway was never restarted after update.
Users with a running gateway then hit the ImportError on their next
Telegram/Discord message when the stale gateway process lazily loaded
run_agent.py (new version) which also had the top-level import.
Fixes:
- hermes_cli/gateway.py: lazy import at call site (line 940)
- run_agent.py: lazy import at call site (line 6927)
- tools/terminal_tool.py: lazy imports at 3 call sites
- tools/tts_tool.py: static schema string (no module-level call)
- hermes_cli/auth.py: lazy import at call site (line 2024)
- hermes_cli/main.py: reload hermes_constants after git pull in cmd_update
Also fixes 4 pre-existing test failures in test_parse_env_var caused by
NameError on display_hermes_home in terminal_tool.py.
Add skills.external_dirs config option — a list of additional directories
to scan for skills alongside ~/.hermes/skills/. External dirs are read-only:
skill creation/editing always writes to the local dir. Local skills take
precedence when names collide.
This lets users share skills across tools/agents without copying them into
Hermes's own directory (e.g. ~/.agents/skills, /shared/team-skills).
Changes:
- agent/skill_utils.py: add get_external_skills_dirs() and get_all_skills_dirs()
- agent/prompt_builder.py: scan external dirs in build_skills_system_prompt()
- tools/skills_tool.py: _find_all_skills() and skill_view() search external dirs;
security check recognizes configured external dirs as trusted
- agent/skill_commands.py: /skill slash commands discover external skills
- hermes_cli/config.py: add skills.external_dirs to DEFAULT_CONFIG
- cli-config.yaml.example: document the option
- tests/agent/test_external_skills.py: 11 tests covering discovery, precedence,
deduplication, and skill_view for external skills
Requested by community member primco.
Two related fixes for remote terminal backends (Modal/Docker):
1. NEW: Credential file mounting system
Skills declare required_credential_files in frontmatter. Files are
mounted into Docker (read-only bind mounts) and Modal (mounts at
creation + sync via exec on each command for mid-session changes).
Google Workspace skill updated with the new field.
2. FIX: Docker backend now includes env_passthrough vars
Skills that declare required_environment_variables (e.g. Notion with
NOTION_API_KEY) register vars in the env_passthrough system. The
local backend checked this, but Docker's forward_env was a separate
disconnected list. Now Docker exec merges both sources, so
skill-declared env vars are forwarded into containers automatically.
This fixes the reported issue where NOTION_API_KEY in ~/.hermes/.env
wasn't reaching the Docker container despite being registered via
the Notion skill's prerequisites.
Closes#3665
Prep for profiles: user-facing messages now use display_hermes_home() so
diagnostic output shows the correct path for each profile.
New helper: display_hermes_home() in hermes_constants.py
12 files swept, ~30 user-facing string replacements.
Includes dynamic TTS schema description.
Switch MCP HTTP transport from the deprecated streamablehttp_client()
(mcp < 1.24.0) to the new streamable_http_client() API that accepts a
pre-built httpx.AsyncClient.
Changes vs the original PR #3391:
- Separate try/except imports so mcp < 1.24.0 doesn't break (graceful
fallback to deprecated API instead of losing HTTP MCP entirely)
- Wrap httpx.AsyncClient in async-with for proper lifecycle management
(the new SDK API explicitly skips closing caller-provided clients)
- Match SDK's own create_mcp_http_client defaults: follow_redirects=True,
Timeout(connect_timeout, read=300.0)
- Keep deprecated code path as fallback for older SDK versions
Co-authored-by: HenkDz <HenkDz@users.noreply.github.com>
Salvage of PR #3452 (kentimsit). Fixes skill readiness checks on remote backends — persisted env vars are no longer incorrectly marked as missing.
Co-Authored-By: kentimsit <kentimsit@users.noreply.github.com>
Adds Exa (https://exa.ai) as a fourth web backend alongside Parallel,
Firecrawl, and Tavily. Follows the exact same integration pattern:
- Backend selection: config web.backend=exa or auto-detect from EXA_API_KEY
- Search: _exa_search() with highlights for result descriptions
- Extract: _exa_extract() with full text content extraction
- Lazy singleton client with x-exa-integration header
- Wired into web_search_tool and web_extract_tool dispatchers
- check_web_api_key() and requires_env updated
- CLI: hermes setup summary, hermes tools config, hermes config show
- config.py: EXA_API_KEY in OPTIONAL_ENV_VARS with metadata
- pyproject.toml: exa-py>=2.9.0,<3 in dependencies
Salvaged from PR #1850.
Co-authored-by: louiswalsh <louiswalsh@users.noreply.github.com>
New installs get a cleaner structure:
cache/images/ (was image_cache/)
cache/audio/ (was audio_cache/)
cache/documents/ (was document_cache/)
cache/screenshots/ (was browser_screenshots/)
platforms/whatsapp/session/ (was whatsapp/session/)
platforms/matrix/store/ (was matrix/store/)
platforms/pairing/ (was pairing/)
Existing installs are unaffected -- get_hermes_dir() checks for the
old path first and uses it if present. No migration needed.
Adds get_hermes_dir(new_subpath, old_name) helper to hermes_constants.py
for reuse by any future subsystem.
Add the gstack community skills repo to the default tap list and fix
skill_identifier construction for repos with an empty path prefix.
Co-authored-by: Tugrul Guner <tugrulguner@users.noreply.github.com>
Salvage of PR #3533 (binhnt92). Follow-up to #3480 — applies min(100, ...) to 5 remaining unclamped percentage display sites in context_compressor, cli /stats, gateway /stats, and memory tool. Defensive clamps now that the root cause (estimation heuristic) was already removed in #3480.
Co-Authored-By: binhnt92 <binhnt92@users.noreply.github.com>
One-shot local execution built `printf FENCE; <cmd>; __hermes_rc=...`, so a
command ending in a heredoc produced a closing line like `EOF; __hermes_rc=...`,
which is not a valid delimiter. Bash then treated the rest of the wrapper as
heredoc body, leaking it into tool output (e.g. gh issue/PR flows).
Use newline-separated wrapper lines so the delimiter stays alone and the
trailer runs after the heredoc completes.
Co-authored-by: kshitijk4poor <82637225+kshitijk4poor@users.noreply.github.com>
Prevent the agent from accidentally killing its own process with
pkill -f gateway, killall hermes, etc. Adds a dangerous command
pattern that triggers the approval flow.
Co-authored-by: arasovic <arasovic@users.noreply.github.com>
* feat: GPT tool-use steering + strip budget warnings from history
Two changes to improve tool reliability, especially for OpenAI GPT models:
1. GPT tool-use enforcement prompt: Adds GPT_TOOL_USE_GUIDANCE to the
system prompt when the model name contains 'gpt' and tools are loaded.
This addresses a known behavioral pattern where GPT models describe
intended actions ('I will run the tests') instead of actually making
tool calls. Inspired by similar steering in OpenCode (beast.txt) and
Cline (GPT-5.1 variant).
2. Budget warning history stripping: Budget pressure warnings injected by
_get_budget_warning() into tool results are now stripped when
conversation history is replayed via run_conversation(). Previously,
these turn-scoped signals persisted across turns, causing models to
avoid tool calls in all subsequent messages after any turn that hit
the 70-90% iteration threshold.
* fix: replace hardcoded ~/.hermes paths with get_hermes_home() for profile support
Prep for the upcoming profiles feature — each profile is a separate
HERMES_HOME directory, so all paths must respect the env var.
Fixes:
- gateway/platforms/matrix.py: Matrix E2EE store was hardcoded to
~/.hermes/matrix/store, ignoring HERMES_HOME. Now uses
get_hermes_home() so each profile gets its own Matrix state.
- gateway/platforms/telegram.py: Two locations reading config.yaml via
Path.home()/.hermes instead of get_hermes_home(). DM topic thread_id
persistence and hot-reload would read the wrong config in a profile.
- tools/file_tools.py: Security path for hub index blocking was
hardcoded to ~/.hermes, would miss the actual profile's hub cache.
- hermes_cli/gateway.py: Service naming now uses the profile name
(hermes-gateway-coder) instead of a cryptic hash suffix. Extracted
_profile_suffix() helper shared by systemd and launchd.
- hermes_cli/gateway.py: Launchd plist path and Label now scoped per
profile (ai.hermes.gateway-coder.plist). Previously all profiles
would collide on the same plist file on macOS.
- hermes_cli/gateway.py: Launchd plist now includes HERMES_HOME in
EnvironmentVariables — was missing entirely, making custom
HERMES_HOME broken on macOS launchd (pre-existing bug).
- All launchctl commands in gateway.py, main.py, status.py updated
to use get_launchd_label() instead of hardcoded string.
Test fixes: DM topic tests now set HERMES_HOME env var alongside
Path.home() mock. Launchd test uses get_launchd_label() for expected
commands.
Drop the swe-rex dependency for Modal terminal backend and use the
Modal SDK directly (Sandbox.create + Sandbox.exec). This fixes:
- AsyncUsageWarning from synchronous App.lookup() in async context
- DeprecationError from unencrypted_ports / .url on unencrypted tunnels
(deprecated 2026-03-05)
The new implementation:
- Uses modal.App.lookup.aio() for async-safe app creation
- Uses Sandbox.create.aio() with 'sleep infinity' entrypoint
- Uses Sandbox.exec.aio() for direct command execution (no HTTP server
or tunnel needed)
- Keeps all existing features: persistent filesystem snapshots,
configurable resources (CPU/memory/disk), sudo support, interrupt
handling, _AsyncWorker for event loop safety
Consistent with the Docker backend precedent (PR #2804) where we
removed mini-swe-agent in favor of direct docker run.
Files changed:
- tools/environments/modal.py - core rewrite
- tools/terminal_tool.py - health check: modal instead of swerex
- hermes_cli/setup.py - install modal instead of swe-rex[modal]
- pyproject.toml - modal extra: modal>=1.0.0 instead of swe-rex[modal]
- scripts/kill_modal.sh - grep for hermes-agent instead of swe-rex
- tests/ - updated for new implementation
- environments/README.md - updated patches section
- website/docs - updated install command
* Fix#3409: Add fallback to session_search to prevent false negatives on summarization failure
Fixes#3409. When the auxiliary summarizer fails or returns None, the tool now returns a raw fallback preview of the matched session instead of silently dropping it and returning an empty list
* fix: clean up fallback logic — separate exception handling from preview
Restructure the loop: handle exceptions first (log + nullify), build
entry dict once, then branch on result truthiness. Removes duplicated
field assignments and makes the control flow linear.
---------
Co-authored-by: devorun <130918800+devorun@users.noreply.github.com>
Salvage of #3389 by @binhnt92 with reasoning fallback and retry logic added on top.
All 7 auxiliary LLM call sites now use extract_content_or_reasoning() which mirrors the main agent loop's behavior: extract content, strip think blocks, fall back to structured reasoning fields, retry on empty.
Closes#3389.
Previously, tirith exit code 1 (block) immediately rejected the command
with no approval prompt — users saw 'BLOCKED: Command blocked by
security scan' and the agent moved on. This prevented gateway/CLI users
from approving pipe-to-shell installs like 'curl ... | sh' even when
they understood the risk.
Changes:
- Tirith 'block' and 'warn' now both go through the approval flow.
Users see the full tirith findings (severity, title, description,
safer alternatives) and can choose to approve or deny.
- New _format_tirith_description() builds rich descriptions from tirith
findings JSON so the approval prompt is informative.
- CLI startup now warns when tirith is enabled but not available, so
users know command scanning is degraded to pattern matching only.
The default approval choice is still deny, so the security posture is
unchanged for unattended/timeout scenarios.
Reported via Discord by pistrie — 'curl -fsSL https://mandex.dev/install.sh | sh'
was hard-blocked with no way to approve.
dict.get(key, default) returns None — not the default — when the key IS
present but explicitly set to null/~ in YAML. Calling .lower() on that
raises AttributeError.
Use (config.get(key) or fallback) so both missing keys and explicit nulls
coalesce to the intended default.
Files fixed:
- tools/tts_tool.py — _get_provider()
- tools/web_tools.py — _get_backend()
- tools/mcp_tool.py — MCPServerTask auth config
- trajectory_compressor.py — _detect_provider() and config loading
Co-authored-by: dieutx <dangtc94@gmail.com>
V4A patches with only + lines (no context or - lines) were silently
dropped because search_lines was empty and the 'if search_lines:' block
was the only code path. Addition-only hunks are common when the model
generates patches for new functions or blocks.
Adds an else branch that inserts at the context_hint position when
available, or appends at end of file.
Includes 2 regression tests for addition-only hunks with and without
context hints.
Salvaged from PR #3092 by thakoreh.
Co-authored-by: Hiren <hiren.thakore58@gmail.com>
- add managed modal and gateway-backed tool integrations\n- improve CLI setup, auth, and configuration for subscriber flows\n- expand tests and docs for managed tool support
The delegate_task tool accepts a toolsets parameter directly from the
LLM's function call arguments. When provided, these toolsets are passed
through _strip_blocked_tools but never intersected with the parent
agent's enabled_toolsets. A model can request toolsets the parent does
not have (e.g., web, browser, rl), granting the subagent tools that
were explicitly disabled for the parent.
Intersect LLM-requested toolsets with the parent's enabled set before
applying the blocked-tool filter, so subagents can only receive a
subset of the parent's tools.
Co-authored-by: dieutx <dangtc94@gmail.com>
When third-party tools (Paperclip orchestrator, etc.) spawn hermes chat
as a subprocess, their sessions pollute user session history and search.
- hermes chat --source <tag> (also HERMES_SESSION_SOURCE env var)
- exclude_sources parameter on list_sessions_rich() and search_messages()
- Sessions with source=tool hidden from sessions list/browse/search
- Third-party adapters pass --source tool to isolate agent sessions
Cherry-picked from PR #3208 by HenkDz.
Co-authored-by: Henkey <noonou7@gmail.com>
detect_dangerous_command() ran regex patterns against raw command strings
without normalization, allowing bypass via Unicode fullwidth chars,
ANSI escape codes, null bytes, and 8-bit C1 controls.
Adds _normalize_command_for_detection() that:
- Strips ANSI escapes using the full ECMA-48 strip_ansi() from
tools/ansi_strip (CSI, OSC, DCS, 8-bit C1, nF sequences)
- Removes null bytes
- Normalizes Unicode via NFKC (fullwidth Latin → ASCII, etc.)
Includes 12 regression tests covering fullwidth, ANSI, C1, null byte,
and combined obfuscation bypasses.
Salvaged from PR #3089 by thakoreh — improved ANSI stripping to use
existing comprehensive strip_ansi() instead of a weaker hand-rolled
regex, and added test coverage.
Co-authored-by: Hiren <hiren.thakore58@gmail.com>
* fix(skills): reduce skills.sh resolution churn and preserve trust for wrapped identifiers
- Accept common skills.sh prefix typos (skils-sh/, skils.sh/)
- Strip skills-sh/ prefix in _resolve_trust_level() so trusted repos
stay trusted when installed through skills.sh
- Use resolved identifier (from bundle/meta) for scan_skill source
- Prefer tree search before root scan in _discover_identifier()
- Add _resolve_github_meta() consolidation for inspect flow
Cherry-picked from PR #3001 by kshitijk4poor.
* fix: restore candidate loop in SkillsShSource.fetch() for consistency
The cherry-picked PR only tried the first candidate identifier in
fetch() while inspect() (via _resolve_github_meta) tried all four.
This meant skills at repo/skills/path would be found by inspect but
missed by fetch, forcing it through the heavier _discover_identifier
flow. Restore the candidate loop so both paths behave identically.
Updated the test assertion to match.
---------
Co-authored-by: kshitijk4poor <82637225+kshitijk4poor@users.noreply.github.com>
_send_discord(), _send_slack(), and _send_twilio() all created
aiohttp.ClientSession() without a timeout, leaving HTTP requests
able to hang indefinitely. _send_whatsapp() already used
aiohttp.ClientTimeout(total=30) — this fix applies the same
pattern consistently to all platform send functions.
- Add ClientTimeout(total=30) to _send_discord() ClientSession
- Add ClientTimeout(total=30) to _send_slack() ClientSession
- Add ClientTimeout(total=30) to _send_twilio() ClientSession
Three categories of cleanup, all zero-behavioral-change:
1. F-strings without placeholders (154 fixes across 29 files)
- Converted f'...' to '...' where no {expression} was present
- Heaviest files: run_agent.py (24), cli.py (20), honcho_integration/cli.py (34)
2. Simplify defensive patterns in run_agent.py
- Added explicit self._is_anthropic_oauth = False in __init__ (before
the api_mode branch that conditionally sets it)
- Replaced 7x getattr(self, '_is_anthropic_oauth', False) with direct
self._is_anthropic_oauth (attribute always initialized now)
- Added _is_openrouter_url() and _is_anthropic_url() helper methods
- Replaced 3 inline 'openrouter' in self._base_url_lower checks
3. Remove dead code in small files
- hermes_cli/claw.py: removed unused 'total' computation
- tools/fuzzy_match.py: removed unused strip_indent() function and
pattern_stripped variable
Full test suite: 6184 passed, 0 failures
E2E PTY: banner clean, tool calls work, zero garbled ANSI
In gateway mode, async tools (vision_analyze, web_extract, session_search)
deadlock because _run_async() spawns a thread with asyncio.run(), creating
a new event loop, but _get_cached_client() returns an AsyncOpenAI client
bound to a different loop. httpx.AsyncClient cannot work across event loop
boundaries, causing await client.chat.completions.create() to hang forever.
Fix: include the event loop identity in the async client cache key so each
loop gets its own AsyncOpenAI instance. Also fix session_search_tool.py
which had its own broken asyncio.run()-in-thread pattern — now uses the
centralized _run_async() bridge.
- Registry now warns when a tool name is overwritten by a different
toolset (silent dict overwrite was the previous behavior)
- MCP tool registration checks for collisions with non-MCP (built-in)
tools before registering. If an MCP tool's prefixed name matches an
existing built-in, the MCP tool is skipped and a warning is logged.
MCP-to-MCP collisions are allowed (last server wins).
- Both regular MCP tools and utility tools (resources/prompts) are
guarded.
- Adds 5 tests covering: registry overwrite warning, same-toolset
re-registration silence, built-in collision skip, normal registration,
and MCP-to-MCP collision pass-through.
Reported by k_sze (KONG) — MiniMax MCP server's web_search tool could
theoretically shadow Hermes's built-in web_search if prefixing failed.
Centralizes two widely-duplicated patterns into hermes_constants.py:
1. get_hermes_home() — Path resolution for ~/.hermes (HERMES_HOME env var)
- Was copy-pasted inline across 30+ files as:
Path(os.getenv("HERMES_HOME", Path.home() / ".hermes"))
- Now defined once in hermes_constants.py (zero-dependency module)
- hermes_cli/config.py re-exports it for backward compatibility
- Removed local wrapper functions in honcho_integration/client.py,
tools/website_policy.py, tools/tirith_security.py, hermes_cli/uninstall.py
2. parse_reasoning_effort() — Reasoning effort string validation
- Was copy-pasted in cli.py, gateway/run.py, cron/scheduler.py
- Same validation logic: check against (xhigh, high, medium, low, minimal, none)
- Now defined once in hermes_constants.py, called from all 3 locations
- Warning log for unknown values kept at call sites (context-specific)
31 files changed, net +31 lines (125 insertions, 94 deletions)
Full test suite: 6179 passed, 0 failed
* fix(security): add SSRF protection to browser_navigate
browser_navigate() only checked the website blocklist policy but did
not call is_safe_url() to block private/internal addresses. This
allowed the agent to navigate to localhost, cloud metadata endpoints
(169.254.169.254), and private network IPs via the browser.
web_tools and vision_tools already had this check. Added the same
is_safe_url() pre-flight validation before the blocklist check in
browser_navigate().
* fix: move SSRF import to module level, fix policy test mock
Move is_safe_url import to module level so it can be monkeypatched
in tests. Update test_browser_navigate_returns_policy_block to mock
_is_safe_url so the SSRF check passes and the policy check is reached.
* fix(security): harden browser SSRF protection
Follow-up to cherry-picked PR #3041:
1. Fail-closed fallback: if url_safety module can't import, block all
URLs instead of allowing all. Security guards should never fail-open.
2. Post-redirect SSRF check: after navigation, verify the final URL
isn't a private/internal address. If a public URL redirected to
169.254.169.254 or localhost, navigate to about:blank and return
an error — prevents the model from reading internal content via
subsequent browser_snapshot calls.
---------
Co-authored-by: 0xbyt4 <35742124+0xbyt4@users.noreply.github.com>
* feat: nix flake, uv2nix build, dev shell and home manager
* fixed nix run, updated docs for setup
* feat(nix): NixOS module with persistent container mode, managed guards, checks
- Replace homeModules.nix with nixosModules.nix (two deployment modes)
- Mode A (native): hardened systemd service with ProtectSystem=strict
- Mode B (container): persistent Ubuntu container with /nix/store bind-mount,
identity-hash-based recreation, GC root protection, symlink-based updates
- Add HERMES_MANAGED guards blocking CLI config mutation (config set, setup,
gateway install/uninstall) when running under NixOS module
- Add nix/checks.nix with build-time verification (binary, CLI, managed guard)
- Remove container.nix (no Nix-built OCI image; pulls ubuntu:24.04 at runtime)
- Simplify packages.nix (drop fetchFromGitHub submodules, PYTHONPATH wrappers)
- Rewrite docs/nixos-setup.md with full options reference, container
architecture, secrets management, and troubleshooting guide
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* Update config.py
* feat(nix): add CI workflow and enhanced build checks
- GitHub Actions workflow for nix flake check + build on linux/macOS
- Entry point sync check to catch pyproject.toml drift
- Expanded managed-guard check to cover config edit
- Wrap hermes-acp binary in Nix package
- Fix Path type mismatch in is_managed()
* Update MCP server package name; bundled skills support
* fix reading .env. instead have container user a common mounted .env file
* feat(nix): container entrypoint with privilege drop and sudo provisioning
Container was running as non-root via --user, which broke apt/pip installs
and caused crashes when $HOME didn't exist. Replace --user with a Nix-built
entrypoint script that provisions the hermes user, sudo (NOPASSWD), and
/home/hermes inside the container on first boot, then drops privileges via
setpriv. Writable layer persists so setup only runs once.
Also expands MCP server options to support HTTP transport and sampling.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix group and user creation in container mode
* feat(nix): persistent /home/hermes and MESSAGING_CWD in container mode
Container mode now bind-mounts ${stateDir}/home to /home/hermes so the
agent's home directory survives container recreation. Previously it lived
in the writable layer and was lost on image/volume/options changes.
Also passes MESSAGING_CWD to the container so the agent finds its
workspace and documents, matching native mode behavior.
Other changes:
- Extract containerDataDir/containerHomeDir bindings (no more magic strings)
- Fix entrypoint chown to run unconditionally (volume mounts always exist)
- Add schema field to container identity hash for auto-recreation
- Add idempotency test (Scenario G) to config-roundtrip check
* docs: add Nix & NixOS setup guide to docs site
Add comprehensive Nix documentation to the Docusaurus site at
website/docs/getting-started/nix-setup.md, covering nix run/profile
install, NixOS module (native + container modes), declarative settings,
secrets management, MCP servers, managed mode, container architecture,
dev shell, flake checks, and full options reference.
- Register nix-setup in sidebar after installation page
- Add Nix callout tip to installation.md linking to new guide
- Add canonical version pointer in docs/nixos-setup.md
* docs: remove docs/nixos-setup.md, consolidate into website docs
Backfill missing details (restart/restartSec in full example,
gateway.pid, 0750 permissions, docker inspect commands) into
the canonical website/docs/getting-started/nix-setup.md and
delete the old standalone file.
* fix(nix): add compression.protect_last_n and target_ratio to config-keys.json
New keys were added to DEFAULT_CONFIG on main, causing the
config-drift check to fail in CI.
* fix(nix): skip checks on aarch64-darwin (onnxruntime wheel missing)
The full Python venv includes onnxruntime (via faster-whisper/STT)
which lacks a compatible uv2nix wheel on aarch64-darwin. Gate all
checks behind stdenv.hostPlatform.isLinux. The package and devShell
still evaluate on macOS.
* fix(nix): skip flake check and build on macOS CI
onnxruntime (transitive dep via faster-whisper) lacks a compatible
uv2nix wheel on aarch64-darwin. Run full checks and build on Linux
only; macOS CI verifies the flake evaluates without building.
* fix(nix): preserve container writable layer across nixos-rebuild
The container identity hash included the entrypoint's Nix store path,
which changes on every nixpkgs update (due to runtimeShell/stdenv
input-addressing). This caused false-positive identity mismatches,
triggering container recreation and losing the persistent writable layer.
- Use stable symlink (current-entrypoint) like current-package already does
- Remove entrypoint from identity hash (only image/volumes/options matter)
- Add GC root for entrypoint so nix-collect-garbage doesn't break it
- Remove global HERMES_HOME env var from addToSystemPackages (conflicted
with interactive CLI use, service already sets its own)
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Each subagent now gets its own IterationBudget instead of sharing the
parent's. The per-subagent cap is controlled by delegation.max_iterations
in config.yaml (default 50). Total iterations across parent + subagents
can exceed the parent's max_iterations, but the user retains control via
the config setting.
Previously, subagents shared the parent's budget, so three parallel
subagents configured for max_iterations=50 racing against a parent that
already used 60 of 90 would each only get ~10 iterations.
Inspired by PR #2928 (Bartok9) which identified the issue (#2873).
* fix(skills): use Git Trees API to prevent silent subdirectory loss during install
Refactors _download_directory() to use the Git Trees API (single call
for the entire repo tree) as the primary path, falling back to the
recursive Contents API when the tree endpoint is unavailable or
truncated. Prevents silent subdirectory loss caused by per-directory
rate limiting or transient failures.
Cherry-picked from PR #2981 by tugrulguner.
Fixes#2940.
* fix: simplify tree API — use branch name directly as tree-ish
Eliminates an extra git/ref/heads API call by passing the branch name
directly to git/trees/{branch}?recursive=1, matching the pattern
already used by _find_skill_in_repo_tree.
---------
Co-authored-by: tugrulguner <tugrulguner@users.noreply.github.com>
* fix(run_agent): ensure _fire_first_delta() is called for tool generation events
Added calls to _fire_first_delta() in the AIAgent class to improve the handling of tool generation events, ensuring timely notifications during the processing of function calls and tool usage.
* fix(run_agent): improve timeout handling for chat completions
Enhanced the timeout configuration for chat completions in the AIAgent class by introducing customizable connection, read, and write timeouts using environment variables. This ensures more robust handling of API requests during streaming operations.
* fix(run_agent): reduce default stream read timeout for chat completions
Updated the default stream read timeout from 120 seconds to 60 seconds in the AIAgent class, enhancing the timeout configuration for chat completions. This change aims to improve responsiveness during streaming operations.
* fix(run_agent): enhance streaming error handling and retry logic
Improved the error handling and retry mechanism for streaming requests in the AIAgent class. Introduced a configurable maximum number of stream retries and refined the handling of transient network errors, allowing for retries with fresh connections. Non-transient errors now trigger a fallback to non-streaming only when appropriate, ensuring better resilience during API interactions.
* fix: skills-sh install fails for deeply nested repo structures
Skills in repos with deep directory nesting (e.g.
cli-tool/components/skills/development/senior-backend/) could not be
installed because the candidate path generation and shallow root-dir
scan never reached them.
Added GitHubSource._find_skill_in_repo_tree() which uses the GitHub
Trees API to recursively search the entire repo tree in a single API
call. This is used as a final fallback in
SkillsShSource._discover_identifier() when the standard candidate
paths and shallow scan both fail.
Fixes installation of skills from repos like davila7/claude-code-templates
where skills are nested 4+ levels deep.
Reported by user Samuraixheart.
_resolve_trust_level() didn't handle 'agent-created' source, so it
fell through to 'community' trust level. Community policy blocks on
any caution or dangerous findings, which meant common patterns like
curl with env vars, systemctl, crontab, cloudflared references etc.
would block skill creation/patching.
The agent-created policy row already existed in INSTALL_POLICY with
permissive settings (allow caution, ask on dangerous) but was never
reached. Now it is.
Fixes reports of skill_manage being blocked by security scanner.
* docs: unify hooks documentation — add plugin hooks to hooks page, add session:end event
The hooks page only documented gateway event hooks (HOOK.yaml system).
The plugins page listed plugin hooks (pre_tool_call, etc.) that weren't
referenced from the hooks page, which was confusing.
Changes:
- hooks.md: Add overview table showing both hook systems
- hooks.md: Add Plugin Hooks section with available hooks, callback
signatures, and example
- hooks.md: Add missing session:end gateway event (emitted but undocumented)
- hooks.md: Mark pre_llm_call, post_llm_call, on_session_start,
on_session_end as planned (defined in VALID_HOOKS but not yet invoked)
- hooks.md: Update info box to cross-reference plugin hooks
- hooks.md: Fix heading hierarchy (gateway content as subsections)
- plugins.md: Add cross-reference to hooks page for full details
- plugins.md: Mark planned hooks as (planned)
* fix: browser_vision ignores auxiliary.vision.timeout config
browser_vision called call_llm() without passing a timeout parameter,
so it always used the 30-second default in auxiliary_client.py. This
made vision analysis with local models (llama.cpp, ollama) impossible
since they typically need more than 30s for screenshot analysis.
Now browser_vision reads auxiliary.vision.timeout from config.yaml
(same config key that vision_analyze already uses) and passes it
through to call_llm().
Also bumped the default vision timeout from 30s to 120s in both
browser_vision and vision_analyze — 30s is too aggressive for local
models and the previous default silently failed for anyone running
vision locally.
Fixes user report from GamerGB1988.
* feat: env var passthrough for skills and user config
Skills that declare required_environment_variables now have those vars
passed through to sandboxed execution environments (execute_code and
terminal). Previously, execute_code stripped all vars containing KEY,
TOKEN, SECRET, etc. and the terminal blocklist removed Hermes
infrastructure vars — both blocked skill-declared env vars.
Two passthrough sources:
1. Skill-scoped (automatic): when a skill is loaded via skill_view and
declares required_environment_variables, vars that are present in
the environment are registered in a session-scoped passthrough set.
2. Config-based (manual): terminal.env_passthrough in config.yaml lets
users explicitly allowlist vars for non-skill use cases.
Changes:
- New module: tools/env_passthrough.py — shared passthrough registry
- hermes_cli/config.py: add terminal.env_passthrough to DEFAULT_CONFIG
- tools/skills_tool.py: register available skill env vars on load
- tools/code_execution_tool.py: check passthrough before filtering
- tools/environments/local.py: check passthrough in _sanitize_subprocess_env
and _make_run_env
- 19 new tests covering all layers
* docs: add environment variable passthrough documentation
Document the env var passthrough feature across four docs pages:
- security.md: new 'Environment Variable Passthrough' section with
full explanation, comparison table, and security considerations
- code-execution.md: update security section, add passthrough subsection,
fix comparison table
- creating-skills.md: add tip about automatic sandbox passthrough
- skills.md: add note about passthrough after secure setup docs
Live-tested: launched interactive CLI, loaded a skill with
required_environment_variables, verified TEST_SKILL_SECRET_KEY was
accessible inside execute_code sandbox (value: passthrough-test-value-42).
Complete cleanup after dropping the mini-swe-agent submodule (PR #2804):
- Remove MSWEA_SILENT_STARTUP and MSWEA_GLOBAL_CONFIG_DIR env var
settings from cli.py, run_agent.py, hermes_cli/main.py, doctor.py
- Remove mini-swe-agent health check from hermes doctor
- Remove 'minisweagent' from logger suppression lists
- Remove litellm/typer/platformdirs from requirements.txt
- Remove mini-swe-agent install steps from install.ps1 (Windows)
- Remove mini-swe-agent install steps from website docs
- Update all stale comments/docstrings referencing mini-swe-agent
in terminal_tool.py, tools/__init__.py, code_execution_tool.py,
environments/README.md, environments/agent_loop.py
- Remove mini_swe_runner from pyproject.toml py-modules
(still exists as standalone script for RL training use)
- Shrink test_minisweagent_path.py to empty stub
The orphaned mini-swe-agent/ directory on disk needs manual removal:
rm -rf mini-swe-agent/
When TERMINAL_CWD is set to '.' or any relative path (common when the
CLI config defaults to cwd='.'), container backends (docker, modal,
singularity, daytona) would pass it directly to the container where it's
meaningless. This caused 'docker run -d -w .' to fail.
Now relative paths are caught alongside host paths and replaced with
the default '/root' for container backends.
Drop the mini-swe-agent git submodule. All terminal backends now use
hermes-agent's own environment implementations directly.
Docker backend:
- Inline the `docker run -d` container startup (was 15 lines in
minisweagent's DockerEnvironment). Our wrapper already handled
execute(), cleanup(), security hardening, volumes, and resource limits.
Modal backend:
- Import swe-rex's ModalDeployment directly instead of going through
minisweagent's 90-line passthrough wrapper.
- Bake the _AsyncWorker pattern (from environments/patches.py) directly
into ModalEnvironment for Atropos compatibility without monkey-patching.
Cleanup:
- Remove minisweagent_path.py (submodule path resolution helper)
- Remove submodule init/install from install.sh and setup-hermes.sh
- Remove mini-swe-agent from .gitmodules
- environments/patches.py is now a no-op (kept for backward compat)
- terminal_tool.py no longer does sys.path hacking for minisweagent
- mini_swe_runner.py guards imports (optional, for RL training only)
- Update all affected tests to mock the new direct subprocess calls
- Update README.md, CONTRIBUTING.md
No functionality change — all Docker, Modal, local, SSH, Singularity,
and Daytona backends behave identically. 6093 tests pass.
browser_vision and other browser commands had a hardcoded 30-second
subprocess timeout that couldn't be overridden. Users with slower
machines (local Chromium without GPU) would hit timeouts on screenshot
capture even when setting browser.command_timeout in config.yaml,
because nothing read that value.
Changes:
- Add browser.command_timeout to DEFAULT_CONFIG (default: 30s)
- Add _get_command_timeout() helper that reads config, falls back to 30s
- _run_browser_command() now defaults to config value instead of constant
- browser_vision screenshot no longer hardcodes timeout=30
- browser_navigate uses max(config_timeout, 60) as floor for navigation
Reported by Gamer1988.
On macOS with Homebrew (Apple Silicon), Node.js and agent-browser
binaries live under /opt/homebrew/bin/ which is not included in the
_SANE_PATH fallback used by browser_tool.py and environments/local.py.
When Hermes runs with a filtered PATH (e.g. as a systemd service),
these binaries are invisible, causing 'env: node: No such file or
directory' errors when using browser tools.
Changes:
- Add /opt/homebrew/bin and /opt/homebrew/sbin to _SANE_PATH in both
browser_tool.py and environments/local.py
- Add _discover_homebrew_node_dirs() to find versioned Node installs
(e.g. brew install node@24) that aren't linked into /opt/homebrew/bin
- Extend _find_agent_browser() to search Homebrew and Hermes-managed
dirs when agent-browser isn't on the current PATH
- Include discovered Homebrew node dirs in subprocess PATH when
launching agent-browser
- Add 11 new tests covering all Homebrew path discovery logic
echo was called with the full unquoted path (~username/suffix), allowing
command substitution in the suffix (e.g. ~user/$(malicious)) to execute
arbitrary shell commands. The fix expands only the validated ~username
portion via the shell and concatenates the suffix as a plain string.
Co-authored-by: Gutslabs <gutslabsxyz@gmail.com>
httpx.AsyncClient awaits event hooks. The sync _ssrf_redirect_guard
returned None, causing 'object NoneType can't be used in await
expression' on any vision_analyze call that followed redirects.
Caught during live PTY testing of the merged SSRF protection.
* fix(security): add SSRF protection to vision_tools and web_tools
Both vision_analyze and web_extract/web_crawl accept arbitrary URLs
without checking if they target private/internal network addresses.
A prompt-injected or malicious skill could use this to access cloud
metadata endpoints (169.254.169.254), localhost services, or private
network hosts.
Adds a shared url_safety.is_safe_url() that resolves hostnames and
blocks private, loopback, link-local, and reserved IP ranges. Also
blocks known internal hostnames (metadata.google.internal).
Integrated at the URL validation layer in vision_tools and before
each website_policy check in web_tools (extract, crawl).
* test(vision): update localhost test to reflect SSRF protection
The existing test_valid_url_with_port asserted localhost URLs pass
validation. With SSRF protection, localhost is now correctly blocked.
Update the test to verify the block, and add a separate test for
valid URLs with ports using a public hostname.
* fix(security): harden SSRF protection — fail-closed, CGNAT, multicast, redirect guard
Follow-up hardening on top of dieutx's SSRF protection (PR #2630):
- Change fail-open to fail-closed: DNS errors and unexpected exceptions
now block the request instead of allowing it (OWASP best practice)
- Block CGNAT range (100.64.0.0/10): Python's ipaddress.is_private
does NOT cover this range (returns False for both is_private and
is_global). Used by Tailscale/WireGuard and carrier infrastructure.
- Add is_multicast and is_unspecified checks: multicast (224.0.0.0/4)
and unspecified (0.0.0.0) addresses were not caught by the original
four-check chain
- Add redirect guard for vision_tools: httpx event hook re-validates
each redirect target against SSRF checks, preventing the classic
redirect-based SSRF bypass (302 to internal IP)
- Move SSRF filtering before backend dispatch in web_extract: now
covers Parallel and Tavily backends, not just Firecrawl
- Extract _is_blocked_ip() helper for cleaner IP range checking
- Add 24 new tests (CGNAT, multicast, IPv4-mapped IPv6, fail-closed
behavior, parametrized blocked/allowed IP lists)
- Fix existing tests to mock DNS resolution for test hostnames
---------
Co-authored-by: dieutx <dangtc94@gmail.com>
Root cause: terminal_tool, execute_code, and process_registry returned raw
subprocess output with ANSI escape sequences intact. The model saw these
in tool results and copied them into file writes.
Previous fix (PR #2532) stripped ANSI at the write point in file_tools.py,
but this was a band-aid — regex on file content risks corrupting legitimate
content, and doesn't prevent ANSI from wasting tokens in the model context.
Source-level fix:
- New tools/ansi_strip.py with comprehensive ECMA-48 regex covering CSI
(incl. private-mode, colon-separated, intermediate bytes), OSC (both
terminators), DCS/SOS/PM/APC strings, Fp/Fe/Fs/nF escapes, 8-bit C1
- terminal_tool.py: strip output before returning to model
- code_execution_tool.py: strip stdout/stderr before returning
- process_registry.py: strip output in poll/read_log/wait
- file_tools.py: remove _strip_ansi band-aid (no longer needed)
Verified: `ls --color=always` output returned as clean text to model,
file written from that output contains zero ESC bytes.
An agent session killed the systemd-managed gateway (PID 1605) and restarted
it with '&disown', taking it outside systemd's Restart= management. When the
orphaned process later received SIGTERM, nothing restarted it.
Add dangerous command patterns to detect:
- 'gateway run' with & (background), disown, nohup, or setsid
- These should use 'systemctl --user restart hermes-gateway' instead
Also applied directly to main repo and fixed the systemd service:
- Changed Restart=on-failure to Restart=always (clean SIGTERM = exit 0 = not
a 'failure', so on-failure never triggered)
- RestartSec=10 for reasonable restart delay
Path('~/.hermes/image.png').is_file() returns False because Path
doesn't expand tilde. This caused the tool to fall through to URL
validation, which also failed, producing a confusing error:
'Invalid image source. Provide an HTTP/HTTPS URL or a valid local
file path.'
Fix: use os.path.expanduser() before constructing the Path object.
Added two tests for tilde expansion (success and nonexistent file).
* fix(mcp-oauth): port mismatch, path traversal, and shared state in OAuth flow
Three bugs in the new MCP OAuth 2.1 PKCE implementation:
1. CRITICAL: OAuth redirect port mismatch — build_oauth_auth() calls
_find_free_port() to register the redirect_uri, but _wait_for_callback()
calls _find_free_port() again getting a DIFFERENT port. Browser redirects
to port A, server listens on port B — callback never arrives, 120s timeout.
Fix: share the port via module-level _oauth_port variable.
2. MEDIUM: Path traversal via unsanitized server_name — HermesTokenStorage
uses server_name directly in filenames. A name like "../../.ssh/config"
writes token files outside ~/.hermes/mcp-tokens/.
Fix: sanitize server_name with the same regex pattern used elsewhere.
3. MEDIUM: Class-level auth_code/state on _CallbackHandler causes data
races if concurrent OAuth flows run. Second callback overwrites first.
Fix: factory function _make_callback_handler() returns a handler class
with a closure-scoped result dict, isolating each flow.
* test: add tests for MCP OAuth path traversal, handler isolation, and port sharing
7 new tests covering:
- Path traversal blocked (../../.ssh/config stays in mcp-tokens/)
- Dots/slashes sanitized and resolved within base dir
- Normal server names preserved
- Special characters sanitized (@, :, /)
- Concurrent handler result dicts are independent
- Handler writes to its own result dict, not class-level
- build_oauth_auth stores port in module-level _oauth_port
---------
Co-authored-by: 0xbyt4 <35742124+0xbyt4@users.noreply.github.com>
When session_search is called without a query (or with an empty query),
it now returns metadata for the most recent sessions instead of erroring.
This lets the agent quickly see what was worked on recently without
needing specific keywords.
Returns for each session: session_id, title, source, started_at,
last_active, message_count, preview (first user message).
Zero LLM cost — pure DB query. Current session lineage and child
delegation sessions are excluded.
The agent can then keyword-search specific sessions if it needs
deeper context from any of them.
Models occasionally copy ANSI escape sequences from terminal output
or display formatting into file content, breaking shebangs and
injecting binary characters into scripts.
Strip ANSI codes (CSI, OSC, simple escapes) from:
- write_file content
- patch old_string, new_string, and V4A patch content
The check is fast (skips entirely if no ESC byte present).
Reported by Andi Jaeger.
Reads auxiliary.vision.timeout from config.yaml (default: 30s) and
passes it to async_call_llm. Useful for slow local vision models
that need more than 30 seconds.
Setting is in config.yaml (not .env) since it's not a secret:
auxiliary:
vision:
timeout: 120
Based on PR #2306.
Co-authored-by: kshitijk4poor <kshitijk4poor@users.noreply.github.com>
Add hermes mcp add/remove/list/test/configure CLI for managing MCP
server connections interactively. Discovery-first 'add' flow connects,
discovers tools, and lets users select which to enable via curses checklist.
Add OAuth 2.1 PKCE authentication for MCP HTTP servers (RFC 7636).
Supports browser-based and manual (headless) authorization, token
caching with 0600 permissions, automatic refresh. Zero external deps.
Add ${ENV_VAR} interpolation in MCP server config values, resolved
from os.environ + ~/.hermes/.env at load time.
Core OAuth module from PR #2021 by @imnotdev25. CLI and mcp_tool
wiring rewritten against current main. Closes#497, #690.
When sounddevice is installed but libportaudio2 is not present on the
system, the OSError was caught together with ImportError and showed a
generic 'pip install sounddevice' message that sent users down the wrong
path.
Split the except clause to give a clear, actionable message for the
OSError case, including the correct apt/brew commands to install the
system library.
* fix: respect DashScope v1 runtime mode for alibaba
Remove the hardcoded Alibaba branch from resolve_runtime_provider()
that forced api_mode='anthropic_messages' regardless of the base URL.
Alibaba now goes through the generic API-key provider path, which
auto-detects the protocol from the URL:
- /apps/anthropic → anthropic_messages (via endswith check)
- /v1 → chat_completions (default)
This fixes Alibaba setup with OpenAI-compatible DashScope endpoints
(e.g. coding-intl.dashscope.aliyuncs.com/v1) that were broken because
runtime always forced Anthropic mode even when setup saved a /v1 URL.
Based on PR #2024 by @kshitijk4poor.
* docs(skill): add split, merge, search examples to ocr-and-documents skill
Adds pymupdf examples for PDF splitting, merging, and text search
to the existing ocr-and-documents skill. No new dependencies — pymupdf
already covers all three operations natively.
* fix: replace all production print() calls with logger in rl_training_tool
Replace all bare print() calls in production code paths with proper logger calls.
- Add `import logging` and module-level `logger = logging.getLogger(__name__)`
- Replace print() in _start_training_run() with logger.info()
- Replace print() in _stop_training_run() with logger.info()
- Replace print(Warning/Note) calls with logger.warning() and logger.info()
Using the logging framework allows log level filtering, proper formatting,
and log routing instead of always printing to stdout.
---------
Co-authored-by: kshitijk4poor <kshitijk4poor@users.noreply.github.com>
Co-authored-by: memosr.eth <96793918+memosr@users.noreply.github.com>
Parse thread_id from explicit deliver target (e.g. telegram:-1003724596514:17)
and forward it to _send_to_platform and mirror_to_session.
Previously _resolve_delivery_target() always set thread_id=None when
parsing the platform:chat_id format, breaking cron job delivery to
specific Telegram topics.
Added tests:
- test_explicit_telegram_topic_target_with_thread_id
- test_explicit_telegram_chat_id_without_thread_id
Also updated CRONJOB_SCHEMA deliver description to document the
platform:chat_id:thread_id format.
Co-authored-by: Alex Ferrari <alex@thealexferrari.com>
Cherry-picked from PR #2122 by @AtlasMeridia.
1. do_inspect bytes crash: bundle.files returns bytes for official
skills, .split() expected str. Added decode guard.
2. GitHub redirects: three httpx.get calls missing follow_redirects=True,
causing silent 301 failures on renamed orgs.
3. Skill discovery fallback: scan repo root directories when standard
paths (skills/, .agents/skills/, .claude/skills/) miss.
4. tap list KeyError: t['repo'] crashes for local taps. Use safe .get().
Changes the policy for agent-created skills with critical security
findings from 'block' (silently rejected) to 'ask' (allowed with
warning logged). The agent created the skill, so blocking it entirely
is too aggressive — let it through but log the findings.
- Policy: agent-created dangerous changed from block to ask
- should_allow_install returns None for 'ask' (vs True/False)
- format_scan_report shows 'NEEDS CONFIRMATION' for ask
- skill_manager_tool.py caller handles None (allows with warning)
- force=True still overrides as before
Based on PR #2271 by redhelix (closed — 3200 lines of unrelated
Mission Control code excluded).
Two fixes:
1. Use a single open(os.devnull) handle for both stdout and stderr
suppression, preventing a file handle leak if the second open() fails.
2. Set server_sock = None after closing it in the try block to prevent
the finally block from closing it again (causing an OSError).
Closes#2136
Co-authored-by: dieutx <dangtc94@gmail.com>
Cherry-picked from PR #2201 by @Gutslabs.
session_search resolved hits to parent/root sessions but only excluded
the exact current_session_id. If the active session was a child
continuation (compression/delegation), its parent could still appear
as a 'past' conversation result.
Fix: resolve current_session_id to its lineage root before filtering,
so the entire active lineage (parent and children) is excluded.
Cherry-picked from PR #2169 by @0xbyt4.
1. _strip_provider_prefix: skip Ollama model:tag names (qwen:0.5b)
2. Fuzzy match: remove reverse direction that made claude-sonnet-4
resolve to 1M instead of 200K
3. _has_content_after_think_block: reuse _strip_think_blocks() to
handle all tag variants (thinking, reasoning, REASONING_SCRATCHPAD)
4. models.dev lookup: elif→if so nous provider also queries models.dev
5. Disk cache fallback: use 5-min TTL instead of full hour so network
is retried soon
6. Delegate build: wrap child construction in try/finally so
_last_resolved_tool_names is always restored on exception
Matrix, Mattermost, Home Assistant, and DingTalk were missing from the
platform_map in both cron/scheduler.py and tools/send_message_tool.py,
causing delivery to those platforms to silently fail.
Also updates the cronjob tool schema description to list all available
delivery targets so the model knows its options.
Cron jobs run unattended with no user present. Previously the agent had
send_message and clarify tools available, which makes no sense — the
final response is auto-delivered, and there's nobody to ask questions to.
Changes:
- Disable messaging and clarify toolsets for cron agent sessions
- Update cron platform hint to emphasize autonomous execution: no user
present, cannot ask questions, must execute fully and make decisions
- Update cronjob tool schema description to match (remove stale
send_message guidance)
Authored by Hanai. Allows overriding the OpenAI TTS endpoint via
tts.openai.base_url in config.yaml for self-hosted or OpenAI-compatible
TTS services. Falls back to api.openai.com when not set.
find_one is being deprecated. Primary lookup now uses get() with a
deterministic sandbox name (hermes-{task_id}). A legacy fallback via
list(labels=...) ensures sandboxes created before this migration are
still resumable.
The merge at e7844e9c re-introduced a line in _build_child_agent() that
references _saved_tool_names — a variable only defined in _run_single_child().
This caused NameError on every delegate_task call, completely breaking
subagent delegation.
Moves the child._delegate_saved_tool_names assignment to _run_single_child()
where _saved_tool_names is actually defined, keeping the save/restore in the
same scope as the try/finally block.
Adds two regression tests from PR #2038 (YanSte).
Also fixes the same issue reported in PR #2048 (Gutslabs).
Co-authored-by: Yannick Stephan <yannick.stephan@gmail.com>
Co-authored-by: Guts <gutslabs@users.noreply.github.com>
Allow users to configure a custom base_url for the OpenAI TTS provider
in ~/.hermes/config.yaml under tts.openai.base_url. Defaults to the
official OpenAI endpoint. Enables use of self-hosted or OpenAI-compatible
TTS services (e.g. http://localhost:8000/v1).
Also adds a TTS configuration example block to cli-config.yaml.example.
* fix: banner skill count now respects disabled skills and platform filtering
The banner's get_available_skills() was doing a raw rglob scan of
~/.hermes/skills/ without checking:
- Whether skills are disabled (skills.disabled config)
- Whether skills match the current platform (platforms: frontmatter)
This caused the banner to show inflated skill counts (e.g. '100 skills'
when many are disabled) and list macOS-only skills on Linux.
Fix: delegate to _find_all_skills() from tools/skills_tool which already
handles both platform gating and disabled-skill filtering.
* fix: system prompt and slash commands now respect disabled skills
Two more places where disabled skills were still surfaced:
1. build_skills_system_prompt() in prompt_builder.py — disabled skills
appeared in the <available_skills> system prompt section, causing
the agent to suggest/load them despite being disabled.
2. scan_skill_commands() in skill_commands.py — disabled skills still
registered as /skill-name slash commands in CLI help and could be
invoked.
Both now load _get_disabled_skill_names() and filter accordingly.
* fix: skill_view blocks disabled skills
skill_view() checked platform compatibility but not disabled state,
so the agent could still load and read disabled skills directly.
Now returns a clear error when a disabled skill is requested, telling
the user to enable it via hermes skills or inspect the files manually.
---------
Co-authored-by: Test <test@test.com>
Each configured MCP server now registers as its own toolset in TOOLSETS
(e.g. TOOLSETS['github'] = {tools: ['mcp_github_list_files', ...]}),
making raw server names resolvable in platform_toolsets overrides.
Previously MCP tools were only injected into hermes-* umbrella toolsets,
so gateway sessions using raw toolset names like ['terminal', 'github']
in platform_toolsets couldn't resolve MCP tools.
Skips server names that collide with built-in toolsets. Also handles
idempotent reloads (syncs toolsets even when no new servers connect).
Inspired by PR #1876 by @kshitijk4poor.
Adds 2 tests (standalone toolset creation + built-in collision guard).
MiniMax: Add M2.7 and M2.7-highspeed as new defaults across provider
model lists, auxiliary client, metadata, setup wizard, RL training tool,
fallback tests, and docs. Retain M2.5/M2.1 as alternatives.
OpenRouter: Add grok-4.20-beta, nemotron-3-super-120b-a12b:free,
trinity-large-preview:free, glm-5-turbo, and hunter-alpha to the
model catalog.
MiniMax changes based on PR #1882 by @octo-patch (applied manually
due to stale conflicts in refactored pricing module).
Fixes#1802
The v0.3.0 refactor split child agent construction (_build_child_agent)
and execution (_run_single_child) into separate functions. This created
a scope bug where _saved_tool_names was defined in _build_child_agent
but referenced in _run_single_child's finally block, causing a NameError
on every delegate_task call.
Solution: Move the save/restore logic entirely into _run_single_child,
keeping the save and restore in the same scope as the try/finally block.
This is cleaner than passing the variable through and removes the dead
save from _build_child_agent.
Add first-class GitHub Copilot and Copilot ACP provider support across
model selection, runtime provider resolution, CLI sessions, delegated
subagents, cron jobs, and the Telegram gateway.
This also normalizes Copilot model catalogs and API modes, introduces a
Copilot ACP OpenAI-compatible shim, and fixes service-mode auth by
resolving Homebrew-installed gh binaries under launchd.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Agent-created skills were using the same policy as community hub
installs, blocking any skill with medium/high severity findings
(e.g. docker pull, pip install, git clone). This meant the agent
couldn't create skills that reference Docker or other common tools.
Changed agent-created policy from (allow, block, block) to
(allow, allow, block) — matching the trusted policy. Caution-level
findings (medium/high severity) are now allowed through, while
dangerous findings (critical severity like exfiltration, prompt
injection, reverse shells) remain blocked.
Added 4 tests covering the agent-created policy: safe allowed,
caution allowed, dangerous blocked, force override.
* fix: NameError in OpenCode provider setup (prompt_text -> prompt)
The OpenCode Zen and OpenCode Go setup sections used prompt_text()
which is undefined. All other providers correctly use the local
prompt() function defined in setup.py. Fixes crash during
'hermes setup' when selecting either OpenCode provider.
* fix: Telegram streaming — config bridge, not-modified, flood control
Three fixes for gateway streaming:
1. Bridge streaming config from config.yaml into gateway runtime.
load_gateway_config() now reads the 'streaming' key from config.yaml
(same pattern as session_reset, stt, etc.), matching the docs.
Previously only gateway.json was read.
2. Handle 'Message is not modified' in Telegram edit_message().
This Telegram API error fires when editing with identical content —
a no-op, not a real failure. Previously it returned success=False
which made the stream consumer disable streaming entirely.
3. Handle RetryAfter / flood control in Telegram edit_message().
Fast providers can hit Telegram rate limits during streaming.
Now waits the requested retry_after duration and retries once,
instead of treating it as a fatal edit failure.
Also fixed double-edit on stream finish: the consumer now tracks
last-sent text and skips redundant edits, preventing the not-modified
error at the source.
* refactor: make config.yaml the primary gateway config source
Eliminates the per-key bridge pattern in load_gateway_config().
Previously gateway.json was the primary source and each config.yaml
key needed an individual bridge — easy to forget (streaming was
missing, causing garl4546's bug).
Now config.yaml is read first and its keys are mapped directly into
the GatewayConfig.from_dict() schema. gateway.json is kept as a
legacy fallback layer (loaded first, then overwritten by config.yaml
keys). If gateway.json exists, a log message suggests migrating.
Also:
- Removed dead save_gateway_config() (never called anywhere)
- Updated CLI help text and send_message error to reference
config.yaml instead of gateway.json
---------
Co-authored-by: Test <test@test.com>
Save and restore the process-global _last_resolved_tool_names in
_run_single_child() so the parent's execute_code sandbox generates
correct tool imports after delegation completes.
The global was already mostly mitigated (run_agent.py passes
enabled_tools via self.valid_tool_names), but the global itself
remained corrupted — a footgun for any code that reads it directly.
Co-authored-by: shane9coy <shane9coy@users.noreply.github.com>
* fix(session): skip corrupt lines in load_transcript instead of crashing
Wrap json.loads() in load_transcript() with try/except JSONDecodeError
so that partial JSONL lines (from mid-write crashes like OOM/SIGKILL)
are skipped with a warning instead of crashing the entire transcript
load. The rest of the history loads fine.
Adds a logger.warning with the session ID and truncated corrupt line
content for debugging visibility.
Salvaged from PR #1193 by alireza78a.
Closes#1193
* fix(stt): respect explicit provider config instead of env-var fallback
Rework _get_provider() to separate explicit config from auto-detect.
When stt.provider is explicitly set in config.yaml, that choice is
authoritative — no silent cross-provider fallback based on which env
vars happen to be set. When no provider is configured, auto-detect
still tries: local > groq > openai.
This fixes the reported scenario where provider: local + a placeholder
OPENAI_API_KEY caused the system to silently select OpenAI and fail
with a 401.
Closes#1774
1. browser_tool.py: Replace **args spread on browser_click, browser_type,
and browser_scroll handlers with explicit parameter extraction. The
**args pattern passed all dict keys as keyword arguments, causing
TypeError if the LLM sent unexpected parameters. Now extracts only
the expected params (ref, text, direction) with safe defaults.
2. fuzzy_match.py: Update module docstring to match actual strategy
order in code. Block anchor was listed as #3 but is actually #7.
Multi-occurrence is not a separate strategy but a flag. Updated
count from 9 to 8.
Salvage of PR #1707 by @kshitijk4poor (cherry-picked with authorship preserved).
Adds Tavily as a third web backend alongside Firecrawl and Parallel, using the Tavily REST API via httpx.
- Backend selection via hermes tools → saved as web.backend in config.yaml
- All three tools supported: search, extract, crawl
- TAVILY_API_KEY in config registry, doctor, status, setup wizard
- 15 new Tavily tests + 9 backend selection tests + 5 config tests
- Backward compatible
Closes#1707
Two concurrent gateway sessions calling memory add/replace/remove
simultaneously could both read the old state, apply their changes
independently, and write — the last writer silently drops the first
writer's entry.
Fix: wrap each mutation in a file lock (fcntl.flock on a .lock file).
Under the lock, re-read entries from disk to get the latest state,
apply the mutation, then write. This ensures concurrent writers
serialize properly.
The lock uses a separate .lock file since the memory file itself is
atomically replaced via os.replace() (can't flock a replaced file).
Readers remain lock-free since atomic rename ensures they always see
a complete file.
Two concurrent threads (e.g. parallel subagents) could both pass the
'task_id in _active_sessions' check, both create cloud sessions via
network calls, and then one would overwrite the other — leaking the
first cloud session.
Add double-check after the lock is re-acquired: if another thread
already created a session while we were doing the network call, use
the existing one instead of orphaning it.
* feat(web): add Parallel as alternative web search/extract backend
Adds Parallel (parallel.ai) as a drop-in alternative to Firecrawl for
web_search and web_extract tools using the official parallel-web SDK.
- Backend selection via WEB_SEARCH_BACKEND env var (auto/parallel/firecrawl)
- Auto mode prefers Firecrawl when both keys present; Parallel when sole backend
- web_crawl remains Firecrawl-only with clear error when unavailable
- Lazy SDK imports, interrupt support, singleton clients
- 16 new unit tests for backend selection and client config
Co-authored-by: s-jag <s-jag@users.noreply.github.com>
* fix: add PARALLEL_API_KEY to config registry and fix web_crawl policy tests
Follow-up for Parallel backend integration:
- Add PARALLEL_API_KEY to OPTIONAL_ENV_VARS (hermes doctor, env blocklist)
- Add to set_config_value api_keys list (hermes config set)
- Add to doctor keys display
- Fix 2 web_crawl policy tests that didn't set FIRECRAWL_API_KEY
(needed now that web_crawl has a Firecrawl availability guard)
* refactor: explicit backend selection via hermes tools, not auto-detect
Replace the auto-detect backend selection with explicit user choice:
- hermes tools saves WEB_SEARCH_BACKEND to .env when user picks a provider
- _get_backend() reads the explicit choice first
- Fallback only for manual/legacy config (uses whichever key is present)
- _is_provider_active() shows [active] for the selected web backend
- Updated tests, docs, and .env.example to remove 'auto' mode language
* refactor: use config.yaml for web backend, not env var
Match the TTS/browser pattern — web.backend is stored in config.yaml
(set by hermes tools), not as a WEB_SEARCH_BACKEND env var.
- _load_web_config() reads web: section from config.yaml
- _get_backend() reads web.backend from config, falls back to key detection
- _configure_provider() saves to config dict (saved to config.yaml)
- _is_provider_active() reads from config dict
- Removed WEB_SEARCH_BACKEND from .env.example, set_config_value, docs
- Updated all tests to mock _load_web_config instead of env vars
---------
Co-authored-by: s-jag <s-jag@users.noreply.github.com>
When container_persistent=false, the inner mini-swe-agent cleanup only
runs 'docker stop' in the background, leaving containers in Exited state.
Now cleanup() also runs 'docker rm -f' to fully remove the container.
Also fixes pre-existing test failures in model_metadata (gpt-4.1 1M context),
setup tests (TTS provider step), and adds MockInnerDocker.cleanup().
Original fix by crazywriter1. Cherry-picked and adapted for current main.
Fixes#1679
* feat: interactive MCP tool configuration in hermes tools
Add the ability to selectively enable/disable individual MCP server
tools through the interactive 'hermes tools' TUI.
Changes:
- tools/mcp_tool.py: Add probe_mcp_server_tools() — lightweight function
that temporarily connects to configured MCP servers, discovers their
tools (names + descriptions), and disconnects. No registry side effects.
- hermes_cli/tools_config.py: Add 'Configure MCP tools' option to the
interactive menu. When selected:
1. Probes all enabled MCP servers for their available tools
2. Shows a per-server curses checklist with tool descriptions
3. Pre-selects tools based on existing include/exclude config
4. Writes changes back as tools.exclude entries in config.yaml
5. Reports which servers failed to connect
The existing CLI commands (hermes tools enable/disable server:tool)
continue to work unchanged. This adds the interactive TUI counterpart
so users can browse and toggle MCP tools visually.
Tests: 22 new tests covering probe function edge cases and interactive
flow (pre-selection, exclude/include modes, description truncation,
multi-server handling, error paths).
* feat(telegram): auto-detect HTML tags and use parse_mode=HTML in send_message
When _send_telegram detects HTML tags in the message body, it now sends
with parse_mode='HTML' instead of converting to MarkdownV2. This allows
cron jobs and agents to send rich HTML-formatted Telegram messages with
bold, italic, code blocks, etc. that render correctly.
Detection uses the same regex from PR #1568 by @ashaney:
re.search(r'<[a-zA-Z/][^>]*>', message)
Plain-text and markdown messages continue through the existing
MarkdownV2 pipeline. The HTML fallback path also catches HTML parse
errors and falls back to plain text, matching the existing MarkdownV2
error handling.
Inspired by: github.com/ashaney — PR #1568
Salvaged from PR #1573 by @eren-karakus0. Cherry-picked with authorship preserved.
Fixes#1143 — background process notifications resume after gateway restart.
Co-authored-by: Muhammet Eren Karakuş <erenkar950@gmail.com>
Add the ability to selectively enable/disable individual MCP server
tools through the interactive 'hermes tools' TUI.
Changes:
- tools/mcp_tool.py: Add probe_mcp_server_tools() — lightweight function
that temporarily connects to configured MCP servers, discovers their
tools (names + descriptions), and disconnects. No registry side effects.
- hermes_cli/tools_config.py: Add 'Configure MCP tools' option to the
interactive menu. When selected:
1. Probes all enabled MCP servers for their available tools
2. Shows a per-server curses checklist with tool descriptions
3. Pre-selects tools based on existing include/exclude config
4. Writes changes back as tools.exclude entries in config.yaml
5. Reports which servers failed to connect
The existing CLI commands (hermes tools enable/disable server:tool)
continue to work unchanged. This adds the interactive TUI counterpart
so users can browse and toggle MCP tools visually.
Tests: 22 new tests covering probe function edge cases and interactive
flow (pre-selection, exclude/include modes, description truncation,
multi-server handling, error paths).
- Default enabled: false (zero overhead when not configured)
- Fast path: cached disabled state skips all work immediately
- TTL cache (30s) for parsed policy — avoids re-reading config.yaml
on every URL check
- Missing shared files warn + skip instead of crashing all web tools
- Lazy yaml import — missing PyYAML doesn't break browser toolset
- Guarded browser_tool import — fail-open lambda fallback
- check_website_access never raises for default path (fail-open with
warning log); only raises with explicit config_path (test mode)
- Simplified enforcement code in web_tools/browser_tool — no more
try/except wrappers since errors are handled internally
Extract the repeated line-position calculation pattern into a
_calculate_line_positions() helper. The same 4-line pattern was
duplicated across _strategy_trimmed_boundary, _strategy_block_anchor,
_strategy_context_aware, and _find_normalized_matches. Also
standardizes the end_pos clamping (some sites used min(), some used
an if-guard).
Based on PR #1604 by aydnOktay.
Co-authored-by: aydnOktay <aydnOktay@users.noreply.github.com>
Add inference.sh CLI (infsh) as a tool integration, giving agents
access to 150+ AI apps through a single CLI — image gen (FLUX, Reve,
Seedream), video (Veo, Wan, Seedance), LLMs, search (Tavily, Exa),
3D, avatar/lipsync, and more. One API key manages all services.
Tools:
- infsh: run any infsh CLI command (app list, app run, etc.)
- infsh_install: install the CLI if not present
Registered as an 'inference' toolset (opt-in, not in core tools).
Includes comprehensive skill docs with examples for all app categories.
Changes from original PR:
- NOT added to _HERMES_CORE_TOOLS (available via --toolsets inference)
- Added 12 tests covering tool registration, command execution,
error handling, timeout, JSON parsing, and install flow
Inspired by PR #1021 by @okaris.
Co-authored-by: okaris <okaris@users.noreply.github.com>
Cast path to str() before os.path.expanduser() to handle pathlib.Path
inputs safely.
Based on PR #1051 by JackTheGit.
Co-authored-by: JackTheGit <JackTheGit@users.noreply.github.com>
* fix: thread safety for concurrent subagent delegation
Four thread-safety fixes that prevent crashes and data races when
running multiple subagents concurrently via delegate_task:
1. Remove redirect_stdout/stderr from delegate_tool — mutating global
sys.stdout races with the spinner thread when multiple children start
concurrently, causing segfaults. Children already run with
quiet_mode=True so the redirect was redundant.
2. Split _run_single_child into _build_child_agent (main thread) +
_run_single_child (worker thread). AIAgent construction creates
httpx/SSL clients which are not thread-safe to initialize
concurrently.
3. Add threading.Lock to SessionDB — subagents share the parent's
SessionDB and call create_session/append_message from worker threads
with no synchronization.
4. Add _active_children_lock to AIAgent — interrupt() iterates
_active_children while worker threads append/remove children.
5. Add _client_cache_lock to auxiliary_client — multiple subagent
threads may resolve clients concurrently via call_llm().
Based on PR #1471 by peteromallet.
* feat: Honcho base_url override via config.yaml + quick command alias type
Two features salvaged from PR #1576:
1. Honcho base_url override: allows pointing Hermes at a remote
self-hosted Honcho deployment via config.yaml:
honcho:
base_url: "http://192.168.x.x:8000"
When set, this overrides the Honcho SDK's environment mapping
(production/local), enabling LAN/VPN Honcho deployments without
requiring the server to live on localhost. Uses config.yaml instead
of env var (HONCHO_URL) per project convention.
2. Quick command alias type: adds a new 'alias' quick command type
that rewrites to another slash command before normal dispatch:
quick_commands:
sc:
type: alias
target: /context
Supports both CLI and gateway. Arguments are forwarded to the
target command.
Based on PR #1576 by redhelix.
---------
Co-authored-by: peteromallet <peteromallet@users.noreply.github.com>
Co-authored-by: redhelix <redhelix@users.noreply.github.com>
Docker terminal sessions are secret-dark by default. This adds
terminal.docker_forward_env as an explicit allowlist for env vars
that may be forwarded into Docker containers.
Values resolve from the current shell first, then fall back to
~/.hermes/.env. Only variables the user explicitly lists are
forwarded — nothing is auto-exposed.
Cherry-picked from PR #1449 by @teknium1, conflict-resolved onto
current main.
Fixes#1436
Supersedes #1439
Remove the optional skill (redundant now that NeuTTS is a built-in TTS
provider). Replace neutts_cli dependency with a standalone synthesis
helper (tools/neutts_synth.py) that calls the neutts Python API directly
in a subprocess.
Add TTS provider selection to hermes setup:
- 'hermes setup' now prompts for TTS provider after model selection
- 'hermes setup tts' available as standalone section
- Selecting NeuTTS checks for deps and offers to install:
espeak-ng (system) + neutts[all] (pip)
- ElevenLabs/OpenAI selections prompt for API keys
- Tool status display shows NeuTTS install state
Changes:
- Remove optional-skills/mlops/models/neutts/ (skill + CLI scaffold)
- Add tools/neutts_synth.py (standalone synthesis subprocess helper)
- Move jo.wav/jo.txt to tools/neutts_samples/ (bundled default voice)
- Refactor _generate_neutts() — uses neutts API via subprocess, no
neutts_cli dependency, config-driven ref_audio/ref_text/model/device
- Add TTS setup to hermes_cli/setup.py (SETUP_SECTIONS, tool status)
- Update config.py defaults (ref_audio, ref_text, model, device)
search_files(target='files') now uses rg --files -g instead of find.
Ripgrep respects .gitignore, excludes hidden dirs by default, and has
parallel directory traversal — ~200x faster on wide trees (0.14s vs 34s
benchmarked on 164-repo tree).
Falls back to find when rg is unavailable, preserving hidden-dir
exclusion and BSD find compatibility.
Salvaged from PR #1464 by @light-merlin-dark (Merlin) — adapted to
preserve hidden-dir exclusion added since the original PR.
* fix(security): harden terminal safety and sandbox file writes
Two security improvements:
1. Dangerous command detection: expand shell -c pattern to catch
combined flags (bash -lc, bash -ic, ksh -c) that were previously
undetected. Pattern changed from matching only 'bash -c' to
matching any shell invocation with -c anywhere in the flags.
2. File write sandboxing: add HERMES_WRITE_SAFE_ROOT env var that
constrains all write_file/patch operations to a configured directory
tree. Opt-in — when unset, behavior is unchanged. Useful for
gateway/messaging deployments that should only touch a workspace.
Based on PR #1085 by ismoilh.
* fix: correct "POSIDEON" typo to "POSEIDON" in banner ASCII art
The poseidon skin's banner_logo had the E and I letters swapped,
spelling "POSIDEON-AGENT" instead of "POSEIDON-AGENT".
---------
Co-authored-by: ismoilh <ismoilh@users.noreply.github.com>
Co-authored-by: unmodeled-tyler <unmodeled.tyler@proton.me>
* fix: prevent infinite 400 failure loop on context overflow (#1630)
When a gateway session exceeds the model's context window, Anthropic may
return a generic 400 invalid_request_error with just 'Error' as the
message. This bypassed the phrase-based context-length detection,
causing the agent to treat it as a non-retryable client error. Worse,
the failed user message was still persisted to the transcript, making
the session even larger on each attempt — creating an infinite loop.
Three-layer fix:
1. run_agent.py — Fallback heuristic: when a 400 error has a very short
generic message AND the session is large (>40% of context or >80
messages), treat it as a probable context overflow and trigger
compression instead of aborting.
2. run_agent.py + gateway/run.py — Don't persist failed messages:
when the agent returns failed=True before generating any response,
skip writing the user's message to the transcript/DB. This prevents
the session from growing on each failure.
3. gateway/run.py — Smarter error messages: detect context-overflow
failures and suggest /compact or /reset specifically, instead of a
generic 'try again' that will fail identically.
* fix(skills): detect prompt injection patterns and block cache file reads
Adds two security layers to prevent prompt injection via skills hub
cache files (#1558):
1. read_file: blocks direct reads of ~/.hermes/skills/.hub/ directory
(index-cache, catalog files). The 3.5MB clawhub_catalog_v1.json
was the original injection vector — untrusted skill descriptions
in the catalog contained adversarial text that the model executed.
2. skill_view: warns when skills are loaded from outside the trusted
~/.hermes/skills/ directory, and detects common injection patterns
in skill content ("ignore previous instructions", "<system>", etc.).
Cherry-picked from PR #1562 by ygd58.
* fix(tools): chunk long messages in send_message_tool before dispatch (#1552)
Long messages sent via send_message tool or cron delivery silently
failed when exceeding platform limits. Gateway adapters handle this
via truncate_message(), but the standalone senders in send_message_tool
bypassed that entirely.
- Apply truncate_message() chunking in _send_to_platform() before
dispatching to individual platform senders
- Remove naive message[i:i+2000] character split in _send_discord()
in favor of centralized smart splitting
- Attach media files to last chunk only for Telegram
- Add regression tests for chunking and media placement
Cherry-picked from PR #1557 by llbn.
* fix(approval): show full command in dangerous command approval (#1553)
Previously the command was truncated to 80 chars in CLI (with a
[v]iew full option), 500 chars in Discord embeds, and missing entirely
in Telegram/Slack approval messages. Now the full command is always
displayed everywhere:
- CLI: removed 80-char truncation and [v]iew full menu option
- Gateway (TG/Slack): approval_required message includes full command
in a code block
- Discord: embed shows full command up to 4096-char limit
- Windows: skip SIGALRM-based test timeout (Unix-only)
- Updated tests: replaced view-flow tests with direct approval tests
Cherry-picked from PR #1566 by crazywriter1.
* fix(cli): flush stdout during agent loop to prevent macOS display freeze (#1624)
The interrupt polling loop in chat() waited on the queue without
invalidating the prompt_toolkit renderer. On macOS, the StdoutProxy
buffer only flushed on input events, causing the CLI to appear frozen
during tool execution until the user typed a key.
Fix: call _invalidate() on each queue timeout (every ~100ms, throttled
to 150ms) to force the renderer to flush buffered agent output.
* fix(claw): warn when API keys are skipped during OpenClaw migration (#1580)
When --migrate-secrets is not passed (the default), API keys like
OPENROUTER_API_KEY are silently skipped with no warning. Users don't
realize their keys weren't migrated until the agent fails to connect.
Add a post-migration warning with actionable instructions: either
re-run with --migrate-secrets or add the key manually via
hermes config set.
Cherry-picked from PR #1593 by ygd58.
* fix(security): block sandbox backend creds from subprocess env (#1264)
Add Modal and Daytona sandbox credentials to the subprocess env
blocklist so they're not leaked to agent terminal sessions via
printenv/env.
Cherry-picked from PR #1571 by ygd58.
---------
Co-authored-by: buray <ygd58@users.noreply.github.com>
Co-authored-by: lbn <llbn@users.noreply.github.com>
Co-authored-by: crazywriter1 <53251494+crazywriter1@users.noreply.github.com>
* feat(skills): add bundled neutts optional skill
Add NeuTTS optional skill with CLI scaffold, bootstrap helper, and
sample voice profile. Also fixes skills_hub.py to handle binary
assets (WAV files) during skill installation.
Changes:
- optional-skills/mlops/models/neutts/ — skill + CLI scaffold
- tools/skills_hub.py — binary asset support (read_bytes, write_bytes)
- tests/tools/test_skills_hub.py — regression tests for binary assets
* feat(tts): add NeuTTS as local TTS provider backend
Add NeuTTS as a fourth TTS provider option alongside Edge, ElevenLabs,
and OpenAI. NeuTTS runs fully on-device via neutts_cli — no API key
needed.
Provider behavior:
- Explicit: set tts.provider to 'neutts' in config.yaml
- Fallback: when Edge TTS is unavailable and neutts_cli is installed,
automatically falls back to NeuTTS instead of failing
- check_tts_requirements() now includes NeuTTS in availability checks
NeuTTS outputs WAV natively. For Telegram voice bubbles, ffmpeg
converts to Opus (same pattern as Edge TTS).
Changes:
- tools/tts_tool.py — _generate_neutts(), _check_neutts_available(),
provider dispatch, fallback logic, Opus conversion
- hermes_cli/config.py — tts.neutts config defaults
---------
Co-authored-by: unmodeled-tyler <unmodeled.tyler@proton.me>
The primary injection vector in #1558 was search_files discovering
catalog cache files in .hub/index-cache/ via find or grep, which
don't skip hidden directories like ripgrep does by default.
Three-layer fix:
1. _search_files (find): add -not -path '*/.*' to exclude hidden
directories, matching ripgrep's default behavior.
2. _search_with_grep: add --exclude-dir='.*' to skip hidden
directories in the grep fallback path.
3. _write_index_cache: write a .ignore file to .hub/ so ripgrep
also skips it even when invoked with --hidden (belt-and-suspenders).
This makes all three search backends (rg, grep, find) consistently
exclude hidden directories, preventing the agent from discovering
and reading unvetted community content in hub cache files.
* fix: prevent infinite 400 failure loop on context overflow (#1630)
When a gateway session exceeds the model's context window, Anthropic may
return a generic 400 invalid_request_error with just 'Error' as the
message. This bypassed the phrase-based context-length detection,
causing the agent to treat it as a non-retryable client error. Worse,
the failed user message was still persisted to the transcript, making
the session even larger on each attempt — creating an infinite loop.
Three-layer fix:
1. run_agent.py — Fallback heuristic: when a 400 error has a very short
generic message AND the session is large (>40% of context or >80
messages), treat it as a probable context overflow and trigger
compression instead of aborting.
2. run_agent.py + gateway/run.py — Don't persist failed messages:
when the agent returns failed=True before generating any response,
skip writing the user's message to the transcript/DB. This prevents
the session from growing on each failure.
3. gateway/run.py — Smarter error messages: detect context-overflow
failures and suggest /compact or /reset specifically, instead of a
generic 'try again' that will fail identically.
* fix(skills): detect prompt injection patterns and block cache file reads
Adds two security layers to prevent prompt injection via skills hub
cache files (#1558):
1. read_file: blocks direct reads of ~/.hermes/skills/.hub/ directory
(index-cache, catalog files). The 3.5MB clawhub_catalog_v1.json
was the original injection vector — untrusted skill descriptions
in the catalog contained adversarial text that the model executed.
2. skill_view: warns when skills are loaded from outside the trusted
~/.hermes/skills/ directory, and detects common injection patterns
in skill content ("ignore previous instructions", "<system>", etc.).
Cherry-picked from PR #1562 by ygd58.
* fix(tools): chunk long messages in send_message_tool before dispatch (#1552)
Long messages sent via send_message tool or cron delivery silently
failed when exceeding platform limits. Gateway adapters handle this
via truncate_message(), but the standalone senders in send_message_tool
bypassed that entirely.
- Apply truncate_message() chunking in _send_to_platform() before
dispatching to individual platform senders
- Remove naive message[i:i+2000] character split in _send_discord()
in favor of centralized smart splitting
- Attach media files to last chunk only for Telegram
- Add regression tests for chunking and media placement
Cherry-picked from PR #1557 by llbn.
* fix(approval): show full command in dangerous command approval (#1553)
Previously the command was truncated to 80 chars in CLI (with a
[v]iew full option), 500 chars in Discord embeds, and missing entirely
in Telegram/Slack approval messages. Now the full command is always
displayed everywhere:
- CLI: removed 80-char truncation and [v]iew full menu option
- Gateway (TG/Slack): approval_required message includes full command
in a code block
- Discord: embed shows full command up to 4096-char limit
- Windows: skip SIGALRM-based test timeout (Unix-only)
- Updated tests: replaced view-flow tests with direct approval tests
Cherry-picked from PR #1566 by crazywriter1.
---------
Co-authored-by: buray <ygd58@users.noreply.github.com>
Co-authored-by: lbn <llbn@users.noreply.github.com>
Co-authored-by: crazywriter1 <53251494+crazywriter1@users.noreply.github.com>
* fix: prevent infinite 400 failure loop on context overflow (#1630)
When a gateway session exceeds the model's context window, Anthropic may
return a generic 400 invalid_request_error with just 'Error' as the
message. This bypassed the phrase-based context-length detection,
causing the agent to treat it as a non-retryable client error. Worse,
the failed user message was still persisted to the transcript, making
the session even larger on each attempt — creating an infinite loop.
Three-layer fix:
1. run_agent.py — Fallback heuristic: when a 400 error has a very short
generic message AND the session is large (>40% of context or >80
messages), treat it as a probable context overflow and trigger
compression instead of aborting.
2. run_agent.py + gateway/run.py — Don't persist failed messages:
when the agent returns failed=True before generating any response,
skip writing the user's message to the transcript/DB. This prevents
the session from growing on each failure.
3. gateway/run.py — Smarter error messages: detect context-overflow
failures and suggest /compact or /reset specifically, instead of a
generic 'try again' that will fail identically.
* fix(skills): detect prompt injection patterns and block cache file reads
Adds two security layers to prevent prompt injection via skills hub
cache files (#1558):
1. read_file: blocks direct reads of ~/.hermes/skills/.hub/ directory
(index-cache, catalog files). The 3.5MB clawhub_catalog_v1.json
was the original injection vector — untrusted skill descriptions
in the catalog contained adversarial text that the model executed.
2. skill_view: warns when skills are loaded from outside the trusted
~/.hermes/skills/ directory, and detects common injection patterns
in skill content ("ignore previous instructions", "<system>", etc.).
Cherry-picked from PR #1562 by ygd58.
* fix(tools): chunk long messages in send_message_tool before dispatch (#1552)
Long messages sent via send_message tool or cron delivery silently
failed when exceeding platform limits. Gateway adapters handle this
via truncate_message(), but the standalone senders in send_message_tool
bypassed that entirely.
- Apply truncate_message() chunking in _send_to_platform() before
dispatching to individual platform senders
- Remove naive message[i:i+2000] character split in _send_discord()
in favor of centralized smart splitting
- Attach media files to last chunk only for Telegram
- Add regression tests for chunking and media placement
Cherry-picked from PR #1557 by llbn.
---------
Co-authored-by: buray <ygd58@users.noreply.github.com>
Co-authored-by: lbn <llbn@users.noreply.github.com>
* fix: prevent infinite 400 failure loop on context overflow (#1630)
When a gateway session exceeds the model's context window, Anthropic may
return a generic 400 invalid_request_error with just 'Error' as the
message. This bypassed the phrase-based context-length detection,
causing the agent to treat it as a non-retryable client error. Worse,
the failed user message was still persisted to the transcript, making
the session even larger on each attempt — creating an infinite loop.
Three-layer fix:
1. run_agent.py — Fallback heuristic: when a 400 error has a very short
generic message AND the session is large (>40% of context or >80
messages), treat it as a probable context overflow and trigger
compression instead of aborting.
2. run_agent.py + gateway/run.py — Don't persist failed messages:
when the agent returns failed=True before generating any response,
skip writing the user's message to the transcript/DB. This prevents
the session from growing on each failure.
3. gateway/run.py — Smarter error messages: detect context-overflow
failures and suggest /compact or /reset specifically, instead of a
generic 'try again' that will fail identically.
* fix(skills): detect prompt injection patterns and block cache file reads
Adds two security layers to prevent prompt injection via skills hub
cache files (#1558):
1. read_file: blocks direct reads of ~/.hermes/skills/.hub/ directory
(index-cache, catalog files). The 3.5MB clawhub_catalog_v1.json
was the original injection vector — untrusted skill descriptions
in the catalog contained adversarial text that the model executed.
2. skill_view: warns when skills are loaded from outside the trusted
~/.hermes/skills/ directory, and detects common injection patterns
in skill content ("ignore previous instructions", "<system>", etc.).
Cherry-picked from PR #1562 by ygd58.
---------
Co-authored-by: buray <ygd58@users.noreply.github.com>
The _send_telegram() function was sending raw markdown text without
parse_mode, causing bold, links, and headers to render as plain text.
This fix reuses the gateway adapter's format_message() to convert
markdown to Telegram's MarkdownV2 format, with a fallback to plain
text if parsing fails.
* fix(tools): remove unnecessary crontab requirement from cronjob tool
The hermes cron system is internal — it uses a JSON-based scheduler
ticked by the gateway (cron/scheduler.py), not system crontab.
The check for shutil.which('crontab') was preventing the cronjob tool
from being available in environments without crontab installed (e.g.
minimal Ubuntu containers).
Changes:
- Remove shutil.which('crontab') check from check_cronjob_requirements()
- Remove unused shutil import
- Update docstring to clarify internal scheduler is used
- Update tests to reflect new behavior and add coverage for all
session modes (interactive, gateway, exec_ask)
Fixes#1589
* test: add HERMES_EXEC_ASK coverage for cronjob requirements
Adds missing test for the exec_ask session mode, complementing
the cherry-picked fix from PR #1633.
---------
Co-authored-by: Bartok9 <bartokmagic@proton.me>
Introduce a cloud browser provider abstraction so users can switch
between Local Browser, Browserbase, and Browser Use (or future providers)
via hermes tools / hermes setup.
Cloud browser providers are behind an ABC (tools/browser_providers/base.py)
so adding a new provider is a single-file addition with no changes to
browser_tool.py internals.
Changes:
- tools/browser_providers/ package with ABC, Browserbase extraction,
and Browser Use provider
- browser_tool.py refactored to use _PROVIDER_REGISTRY + _get_cloud_provider()
(cached) instead of hardcoded _is_local_mode() / _create_browserbase_session()
- tools_config.py: generic _is_provider_active() / _detect_active_provider_index()
replace TTS-only logic; Browser Use added as third browser option
- config.py: BROWSER_USE_API_KEY added to OPTIONAL_ENV_VARS + show_config + allowlist
- subprocess pipe hang fix: agent-browser daemon inherits pipe fds,
communicate() blocks. Replaced with Popen + temp files.
Original PR: #1208
Co-authored-by: ShawnPana <shawnpana@users.noreply.github.com>
* fix: Anthropic OAuth compatibility — Claude Code identity fingerprinting
Anthropic routes OAuth/subscription requests based on Claude Code's
identity markers. Without them, requests get intermittent 500 errors
(~25% failure rate observed). This matches what pi-ai (clawdbot) and
OpenCode both implement for OAuth compatibility.
Changes (OAuth tokens only — API key users unaffected):
1. Headers: user-agent 'claude-cli/2.1.2 (external, cli)' + x-app 'cli'
2. System prompt: prepend 'You are Claude Code, Anthropic's official CLI'
3. System prompt sanitization: replace Hermes/Nous references
4. Tool names: prefix with 'mcp_' (Claude Code convention for non-native tools)
5. Tool name stripping: remove 'mcp_' prefix from response tool calls
Before: 9/12 OK, 1 hard fail, 4 needed retries (~25% error rate)
After: 16/16 OK, 0 failures, 0 retries (0% error rate)
* fix: three gateway issues from user error logs
1. send_animation missing metadata kwarg (base.py)
- Base class send_animation lacked the metadata parameter that the
call site in base.py line 917 passes. Telegram's override accepted
it, but any platform without an override (Discord, Slack, etc.)
hit TypeError. Added metadata to base class signature.
2. MarkdownV2 split-inside-inline-code (base.py truncate_message)
- truncate_message could split at a space inside an inline code span
(e.g. `function(arg1, arg2)`), leaving an unpaired backtick and
unescaped parentheses in the chunk. Telegram rejects with
'character ( is reserved'. Added inline code awareness to the
split-point finder — detects odd backtick counts and moves the
split before the code span.
3. tirith auto-install without cosign (tirith_security.py)
- Previously required cosign on PATH for auto-install, blocking
install entirely with a warning if missing. Now proceeds with
SHA-256 checksum verification only when cosign is unavailable.
Cosign is still used for full supply chain verification when
present. If cosign IS present but verification explicitly fails,
install is still aborted (tampered release).
* fix(tools): improve error logging in code_execution_tool
* fix: harden execute_code cleanup and reduce logging noise
Follow-up to cherry-picked PR #1588 (aydnOktay):
- Initialize server_sock = None before try block to prevent NameError
if exception occurs before socket creation (line 413 is inside the try)
- Guard server_sock.close() with None check
- Narrow cleanup exception handlers to OSError (the actual error type)
- Remove exc_info=True from cleanup debug logs — benign teardown
failures don't need stack traces, the message is sufficient
- Remove redundant try/except around shutil.rmtree(ignore_errors=True)
- Silence sock_path unlink with pass — expected when already cleaned up
---------
Co-authored-by: aydnOktay <xaydinoktay@gmail.com>
When neither apptainer nor singularity is installed, the Singularity
backend silently defaults to "singularity" and fails with a cryptic
FileNotFoundError inside _start_instance(). Add a preflight check
that resolves the executable and verifies it responds, raising a
clear RuntimeError with install instructions on failure.
Closes#1511
* feat: improve memory prioritization — user preferences over procedural knowledge
Inspired by OpenAI Codex's memory prompt improvements (openai/codex#14493)
which focus memory writes on user preferences and recurring patterns
rather than procedural task details.
Key insight: 'Optimize for reducing future user steering — the most
valuable memory prevents the user from having to repeat themselves.'
Changes:
- MEMORY_GUIDANCE (prompt_builder.py): added prioritization hierarchy
and the core principle about reducing user steering
- MEMORY_SCHEMA (memory_tool.py): reordered WHEN TO SAVE list to put
corrections first, added explicit PRIORITY guidance
- Memory nudge (run_agent.py): now asks specifically about preferences,
corrections, and workflow patterns instead of generic 'anything'
- Memory flush (run_agent.py): now instructs to prioritize user
preferences and corrections over task-specific details
* feat: more aggressive skill creation and update prompting
Press harder on skill updates — the agent should proactively patch
skills when it encounters issues during use, not wait to be asked.
Changes:
- SKILLS_GUIDANCE: 'consider saving' → 'save'; added explicit instruction
to patch skills immediately when found outdated/wrong
- Skills header: added instruction to update loaded skills before finishing
if they had missing steps or wrong commands
- Skill nudge: more assertive ('save the approach' not 'consider saving'),
now also prompts for updating existing skills used in the task
- Skill nudge interval: lowered default from 15 to 10 iterations
- skill_manage schema: added 'patch it immediately' to update triggers
Add /browser slash command for connecting browser tools to the user's
live Chrome instance via Chrome DevTools Protocol:
/browser connect — connect to Chrome on localhost:9222
/browser connect ws://host:port — custom CDP endpoint
/browser disconnect — revert to default (headless/Browserbase)
/browser status — show current browser mode + connectivity
When connected:
- All browser tools (navigate, snapshot, click, etc.) control the
user's real Chrome — logged-in sessions, cookies, open tabs
- Platform-specific Chrome launch instructions are shown
- Port connectivity is tested immediately
- A context message is injected so the model knows it's controlling
a live browser and should be mindful of user's open tabs
Implementation:
- BROWSER_CDP_URL env var drives the backend selection in browser_tool.py
- New _create_cdp_session() creates sessions using the CDP override
- _get_cdp_override() checked before local/Browserbase selection
- Existing agent-browser --cdp flag handles the actual CDP connection
Inspired by OpenClaw's browser profile system.
* feat: smart approvals — LLM-based risk assessment for dangerous commands
Adds a 'smart' approval mode that uses the auxiliary LLM to assess
whether a flagged command is genuinely dangerous or a false positive,
auto-approving low-risk commands without prompting the user.
Inspired by OpenAI Codex's Smart Approvals guardian subagent
(openai/codex#13860).
Config (config.yaml):
approvals:
mode: manual # manual (default), smart, off
Modes:
- manual — current behavior, always prompt the user
- smart — aux LLM evaluates risk: APPROVE (auto-allow), DENY (block),
or ESCALATE (fall through to manual prompt)
- off — skip all approval prompts (equivalent to --yolo)
When smart mode auto-approves, the pattern gets session-level approval
so subsequent uses of the same pattern don't trigger another LLM call.
When it denies, the command is blocked without user prompt. When
uncertain, it escalates to the normal manual approval flow.
The LLM prompt is carefully scoped: it sees only the command text and
the flagged reason, assesses actual risk vs false positive, and returns
a single-word verdict.
* feat: make smart approval model configurable via config.yaml
Adds auxiliary.approval section to config.yaml with the same
provider/model/base_url/api_key pattern as other aux tasks (vision,
web_extract, compression, etc.).
Config:
auxiliary:
approval:
provider: auto
model: '' # fast/cheap model recommended
base_url: ''
api_key: ''
Bridged to env vars in both CLI and gateway paths so the aux client
picks them up automatically.
* feat: add /stop command to kill all background processes
Adds a /stop slash command that kills all running background processes
at once. Currently users have to process(list) then process(kill) for
each one individually.
Inspired by OpenAI Codex's separation of interrupt (Ctrl+C stops current
turn) from /stop (cleans up background processes). See openai/codex#14602.
Ctrl+C continues to only interrupt the active agent turn — background
dev servers, watchers, etc. are preserved. /stop is the explicit way
to clean them all up.
Salvaged from PR #1528 by an420eth. Closes#517.
Improves _strategy_block_anchor in fuzzy_match.py:
- Add unicode normalization (smart quotes, em/en-dashes, ellipsis,
non-breaking spaces → ASCII) so LLM-produced unicode artifacts
don't break anchor line matching
- Lower thresholds: 0.10 for unique matches (was 0.70), 0.30 for
multiple candidates — if first/last lines match exactly, the
block is almost certainly correct
- Use original (non-normalized) content for offset calculation to
preserve correct character positions
Tested: 3 new scenarios fixed (em-dash anchors, non-breaking space
anchors, very-low-similarity unique matches), zero regressions on
all 9 existing fuzzy match tests.
Co-authored-by: an420eth <an420eth@users.noreply.github.com>
Keep Docker sandboxes isolated by default. Add an explicit terminal.docker_mount_cwd_to_workspace opt-in, thread it through terminal/file environment creation, and document the security tradeoff and config.yaml workflow clearly.
Fixes#1445 — When using Docker backend, the user's current working
directory is now automatically bind-mounted to /workspace inside the
container. This allows users to run `cd my-project && hermes` and have
their project files accessible to the agent without manual volume config.
Changes:
- Add host_cwd and auto_mount_cwd parameters to DockerEnvironment
- Capture original host CWD in _get_env_config() before container fallback
- Pass host_cwd through _create_environment() to Docker backend
- Add TERMINAL_DOCKER_NO_AUTO_MOUNT env var to disable if needed
- Skip auto-mount when /workspace is already explicitly mounted
- Add tests for auto-mount behavior
- Add documentation for the new feature
The auto-mount is skipped when:
1. TERMINAL_DOCKER_NO_AUTO_MOUNT=true is set
2. User configured docker_volumes with :/workspace
3. persistent_filesystem=true (persistent sandbox mode)
This makes the Docker backend behave more intuitively — the agent
operates on the user's actual project directory by default.
Checkpoint & rollback upgrades:
1. Enabled by default — checkpoints are now on for all new sessions.
Zero cost when no file-mutating tools fire. Disable with
checkpoints.enabled: false in config.yaml.
2. Diff preview — /rollback diff <N> shows a git diff between the
checkpoint and current working tree before committing to a restore.
3. File-level restore — /rollback <N> <file> restores a single file
from a checkpoint instead of the entire directory.
4. Conversation undo on rollback — when restoring files, the last
chat turn is automatically undone so the agent's context matches
the restored filesystem state.
5. Terminal command checkpoints — destructive terminal commands (rm,
mv, sed -i, truncate, git reset/clean, output redirects) now
trigger automatic checkpoints before execution. Previously only
write_file and patch were covered.
6. Change summary in listing — /rollback now shows file count and
+insertions/-deletions for each checkpoint.
7. Fixed dead code — removed duplicate _run_git call in
list_checkpoints with nonsensical --all if False condition.
8. Updated help text — /rollback with no args now shows available
subcommands (diff, file-level restore).
Salvaged from PR #1470 by adavyas.
Core fix: Honcho tool calls in a multi-session gateway could route to
the wrong session because honcho_tools.py relied on process-global
state. Now threads session context through the call chain:
AIAgent._invoke_tool() → handle_function_call() → registry.dispatch()
→ handler **kw → _resolve_session_context()
Changes:
- Add _resolve_session_context() to prefer per-call context over globals
- Plumb honcho_manager + honcho_session_key through handle_function_call
- Add sync_honcho=False to run_conversation() for synthetic flush turns
- Pass honcho_session_key through gateway memory flush lifecycle
- Harden gateway PID detection when /proc cmdline is unreadable
- Make interrupt test scripts import-safe for pytest-xdist
- Wrap BibTeX examples in Jekyll raw blocks for docs build
- Fix thread-order-dependent assertion in client lifecycle test
- Expand Honcho docs: session isolation, lifecycle, routing internals
Dropped from original PR:
- Indentation change in _create_request_openai_client that would move
client creation inside the lock (causes unnecessary contention)
Co-authored-by: adavyas <adavyas@users.noreply.github.com>
Salvaged from PR #1146 by spanishflu-est1918.
Background process progress/completion messages were sent with only
chat_id, landing in the general topic instead of the originating forum
topic. Thread the thread_id from HERMES_SESSION_THREAD_ID through the
watcher payload and pass it as metadata to adapter.send() so Telegram
routes notifications to the correct topic.
The env var export (HERMES_SESSION_THREAD_ID in _set_session_env /
_clear_session_env) already existed on main — this commit adds the
missing watcher plumbing.
Co-authored-by: spanishflu-est1918 <spanishflu-est1918@users.noreply.github.com>
Restore local STT command fallback for voice transcription, detect whisper and ffmpeg in common local install paths, and avoid bogus no-provider messaging when only a backend-specific key is missing.
- Add 'emoji' field to ToolEntry and 'get_emoji()' to ToolRegistry
- Add emoji= to all 50+ registry.register() calls across tool files
- Add get_tool_emoji() helper in agent/display.py with 3-tier resolution:
skin override → registry default → hardcoded fallback
- Replace hardcoded emoji maps in run_agent.py, delegate_tool.py, and
gateway/run.py with centralized get_tool_emoji() calls
- Add 'tool_emojis' field to SkinConfig so skins can override per-tool
emojis (e.g. ares skin could use swords instead of wrenches)
- Add 11 tests (5 registry emoji, 6 display/skin integration)
- Update AGENTS.md skin docs table
Based on the approach from PR #1061 by ForgingAlex (emoji centralization
in registry). This salvage fixes several issues from the original:
- Does NOT split the cronjob tool (which would crash on missing schemas)
- Does NOT change image_generate toolset/requires_env/is_async
- Does NOT delete existing tests
- Completes the centralization (gateway/run.py was missed)
- Hooks into the skin system for full customizability
SSH persistent shell now defaults to true — non-local backends benefit
most from state persistence across execute() calls. Local backend
remains opt-in via TERMINAL_LOCAL_PERSISTENT env var.
New config.yaml option: terminal.persistent_shell (default: true)
Controls the default for non-local backends. Users can disable with:
hermes config set terminal.persistent_shell false
Precedence: per-backend env var > TERMINAL_PERSISTENT_SHELL > default.
Wired through cli.py, gateway/run.py, and hermes_cli/config.py so the
config.yaml value reaches terminal_tool via env var bridge.
When a cronjob is created from within a Telegram or Slack thread,
deliver=origin was posting to the parent channel instead of the thread.
Root cause: the gateway never set HERMES_SESSION_THREAD_ID in the
session environment, so cronjob_tools.py could not capture thread_id
into the job's origin metadata — even though the scheduler already
reads origin.get('thread_id').
Fix:
- gateway/run.py: set HERMES_SESSION_THREAD_ID when thread_id is
present on the session context, and clear it in _clear_session_env
- tools/cronjob_tools.py: read HERMES_SESSION_THREAD_ID into origin
Closes#1219
Extend subprocess env sanitization beyond provider credentials by blocking Hermes-managed tool, messaging, and related gateway runtime vars. Reuse a shared sanitizer in LocalEnvironment and ProcessRegistry so background and PTY processes honor the same blocklist and _HERMES_FORCE_ escape hatch. Add regression coverage for local env execution and process_registry spawning.
Expanded the list of blocked environment variables to include Google, Groq, Mistral, and other major LLM providers. This ensures complete isolation and prevents conflicts with external CLI tools.
Salvaged from PR #1292 onto current main. Preserve per-job model,
provider, and base_url overrides in cron execution, persist them in
job records, expose them through the cronjob tool create/update paths,
and add regression coverage. Deliberately does not persist per-job
api_key values.
- add stt.enabled to the default user config
- make transcription_tools respect the disabled flag globally
- surface disabled state cleanly in voice mode diagnostics
- add regression coverage for disabled STT provider selection
pattern_key was derived by splitting the regex on \b and taking [1],
so patterns starting with the same word (e.g. find -exec rm and
find -delete) produced the same key "find". Approving one silently
approved the other. Using the unique description string as the key
eliminates all collisions.
The fork bomb regex used `()` (empty capture group) and unescaped `{}`
instead of literal `\(\)` and `\{\}`. This meant the classic fork bomb
`:(){ :|:& };:` was never detected. Also added `\s*` between `:` and
`&` and between `;` and trailing `:` to catch whitespace variants.
The execute_code sandbox spawns a child process with cwd set to a
temporary directory, but never adds the hermes-agent project root to
PYTHONPATH. This makes project-root modules like minisweagent_path
unreachable from sandboxed scripts, causing ImportError when the
agent runs self-diagnostic or analysis code via execute_code.
Fix by prepending the hermes-agent root directory to PYTHONPATH in
the child process environment.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Add base_url/api_key overrides for auxiliary tasks and delegation so users can
route those flows straight to a custom OpenAI-compatible endpoint without
having to rely on provider=main or named custom providers.
Also clear gateway session env vars in test isolation so the full suite stays
deterministic when run from a messaging-backed agent session.
Allow cron runs to keep using send_message for additional destinations, but
skip same-target sends when the scheduler will already auto-deliver the final
response there. Add prompt/tool guidance, docs, and regression coverage for
origin/home-channel resolution and thread-aware comparisons.
- Merge _init_persistent_shell + _start_persistent_shell into single method
- Move execute() dispatcher and cleanup() into PersistentShellMixin
so LocalEnvironment and SSHEnvironment inherit them
- Remove broad except Exception wrappers from _execute_oneshot in both backends
- Replace try/except with os.path.exists checks in local _read_temp_files
and _cleanup_temp_files
- Remove redundant bash -c from SSH oneshot (SSH already runs in a shell)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
When vision_analyze_tool fails, the except block was returning a
generic 'could not be analyzed' message that gave the agent no
actionable information about the failure cause.
Replace the generic message with the actual exception string so the
agent can distinguish between backend errors, missing dependencies,
network failures, and unsupported image paths.
Also add an 'error' field to the failure response for structured
error handling by callers.
Fixes#1034
Remove diary-style memory framing from the system prompt and memory tool
schema, explicitly steer task/session logs to session_search, and clarify
that session_search is for cross-session recall after checking the current
conversation first. Add regression tests for the updated guidance text.
Resolve the cherry-pick against current browser_tool structure without carrying unrelated formatting churn, while preserving the intended cleanup, PATH, and screenshot recovery changes from PR #1001.
Unify browser session teardown so manual close, inactivity cleanup, and emergency shutdown all follow the same cleanup path instead of partially duplicating logic.
This changes browser_close() to delegate to cleanup_browser(), which means recording shutdown, Browserbase release, activity bookkeeping cleanup, and local socket-directory removal now happen consistently. It also updates emergency cleanup to route through cleanup_all_browsers() and explicitly clear in-memory tracking state after teardown so stale active-session, last-activity, and recording entries are not left behind on exit.
The screenshot fallback path has also been fixed. _extract_screenshot_path_from_text() now matches real absolute PNG paths, including quoted output, so browser_vision() can recover screenshots when agent-browser emits human-readable text instead of JSON.
Regression coverage was added in tests/tools/test_browser_cleanup.py for screenshot path extraction, cleanup_browser() state removal, browser_close() delegation, and emergency cleanup state clearing.
Verified with:
- python -m pytest tests/tools/test_browser_cleanup.py -q
- python -m pytest tests/tools/test_browser_console.py tests/gateway/test_send_image_file.py -q
Salvaged from PR #1007 by stablegenius49.
- let INSTALL_POLICY decide dangerous verdict handling for builtin skills
- allow --force to override blocked dangerous decisions for trusted and community sources
- accept --yes / -y as aliases for --force in /skills install
- update regression tests to match the intended policy precedence
Round out the skills hub integration with:
- richer skills.sh metadata and security surfacing during inspect/install
- generic check/update flows for hub-installed skills
- support for well-known Agent Skills endpoints via /.well-known/skills/index.json
Also persist upstream bundle metadata in the lock file and add
regression coverage plus live-compatible path handling for both
skills.sh aliases and well-known endpoints.
* improve: add exc_info to MoA error logging
* refactor: tighten MoA traceback logging scope
Follow up on salvaged PR #998 by limiting exc_info logging to terminal
failure paths, avoiding duplicate aggregator errors, and refreshing the
MoA default OpenRouter model lineup to current frontier options.
---------
Co-authored-by: aydnOktay <xaydinoktay@gmail.com>
Harden the skills.sh hub adapter by parsing skill detail pages when
search slugs do not map cleanly onto GitHub skill folder names.
This adds detail-page resolution for alias-style skills, improves
inspect metadata from the page itself, and covers the behavior with
regression tests plus live smoke validation for json-render-react.
Add regression coverage for the standalone email send path and pass an explicit default SSL context to STARTTLS for certificate verification, matching the gateway email adapter hardening salvaged from PR #994.
Add a skills.sh-backed source adapter for the Hermes Skills Hub.
The new adapter uses skills.sh search results for discovery, falls back to
featured homepage links for browse-style queries, and resolves installs /
inspects through the underlying GitHub repo using common Agent Skills
layout conventions. Also expose skills-sh in CLI source filters and add
regression coverage for search, alias resolution, and source routing.
Extend the salvaged MCP filtering work so utility tools are also governed by policy and server capabilities. Store the registered tool subset per server so rediscovery and status reporting stay accurate after filtering.
Add optional config keys under each mcp_servers entry:
- tools.include: whitelist, only listed tools are registered
- tools.exclude: blacklist, all tools except listed are registered
- enabled: false: skip server entirely, no connection attempt
Backward-compatible: no config keys = all tools registered as before.
Tests: TestMCPSelectiveToolLoading (4 tests), 134 passed total.
Salvaged from PR #979 onto current main.
Preserve the current terminal backend checks while surfacing actionable
preflight errors for unknown TERMINAL_ENV values, missing SSH host/user
configuration, and missing Modal credentials/config. Tighten the modal
regression test so it deterministically exercises the config-missing
path.
Salvaged from PR #977 onto current main.
Preserves the MCP stdio command resolution and improved error diagnostics,
with deterministic regression tests for the npx/node PATH cases.
Co-authored-by: kshitij <82637225+kshitijk4poor@users.noreply.github.com>
1. Anthropic + ElevenLabs TTS silence: forward full response to TTS
callback for non-streaming providers (choices first, then native
content blocks fallback).
2. Subprocess timeout kill: play_audio_file now kills the process on
TimeoutExpired instead of leaving zombie processes.
3. Discord disconnect cleanup: leave all voice channels before closing
the client to prevent leaked state.
4. Audio stream leak: close InputStream if stream.start() fails.
5. Race condition: read/write _on_silence_stop under lock in audio
callback thread.
6. _vprint force=True: show API error, retry, and truncation messages
even during streaming TTS.
7. _refresh_level lock: read _voice_recording under _voice_lock.
Remove web UI gateway (web.py, tests, docs, toolset, env vars, Platform.WEB
enum) per maintainer request — Nous is building their own official chat UI.
Fix 1: Replace sd.wait() with polling pattern in play_audio_file() to prevent
indefinite hang when audio device stalls (consistent with play_beep()).
Fix 2: Use importlib.util.find_spec() for faster_whisper/openai availability
checks instead of module-level imports that trigger heavy native library
loading (CUDA/cuDNN) at import time.
Fix 3: Remove inspect.signature() hack in _send_voice_reply() — add **kwargs
to Telegram send_voice() so all adapters accept metadata uniformly.
Fix 4: Make session loading resilient to removed platform enum values — skip
entries with unknown platforms instead of crashing the entire gateway.
- web.py: pass stt_model from config like discord.py and run.py do
- run.py: match new error messages (No STT provider / not set)
- _transcribe_local: add missing "provider": "local" to return dict
Move stream close outside the lock in shutdown() to prevent deadlock
when audio callback tries to acquire the same lock. Replace single
t.join(timeout) with a polling loop (0.1s intervals) so KeyboardInterrupt
is not blocked during stream cleanup.
Voice status was hardcoded to check API keys only. Now uses the actual
provider resolution (local/groq/openai) so it correctly shows
"local faster-whisper" when installed instead of "Groq" or "MISSING".
Merge main's faster-whisper (local, free) with our Groq support into a
unified three-provider STT pipeline: local > groq > openai.
Provider priority ensures free options are tried first. Each provider
has its own transcriber function with model auto-correction, env-
overridable endpoints, and proper error handling.
74 tests cover the full provider matrix, fallback chains, model
correction, config loading, validation edge cases, and dispatch.
Duplicated YAML config parsing for stt.model existed in gateway/run.py
and gateway/platforms/discord.py. Moved to a single helper in
transcription_tools.py and added 5 tests covering all edge cases.
Code fixes:
- STT model, Groq base URL, and OpenAI STT base URL are now
configurable via env vars (STT_GROQ_MODEL, STT_OPENAI_MODEL,
GROQ_BASE_URL, STT_OPENAI_BASE_URL) instead of hardcoded
- Gateway and Discord VC now read stt.model from config.yaml
(previously only CLI did this — gateway always used defaults)
Doc fixes:
- voice-mode.md: move Web UI troubleshooting to web.md (was duplicated)
- voice-mode.md: simplify "How It Works" for end users (remove NaCl,
DAVE, RTP internals)
- voice-mode.md: clarify STT priority (OpenAI used first if both keys
set, Groq recommended for free tier)
- voice-mode.md: document new STT env overrides in config reference
- web.md: remove duplicate Quick Start / Step 1-3 sections
- web.md: add mobile HTTPS mic workarounds (moved from voice-mode.md)
- web.md: clarify STT fallback order
The counter was incremented in start/stop/cancel but never read
anywhere in the codebase. The race condition it was meant to guard
against is practically impossible with the persistent stream design.
- Keep InputStream alive across recordings to avoid CoreAudio hang on
repeated open/close cycles on macOS. New _ensure_stream() creates the
stream once; start()/stop()/cancel() only toggle frame collection.
- Add _close_stream_with_timeout() with daemon thread to prevent
stream.stop()/close() from blocking indefinitely.
- Add generation counter to detect stale stream-open completions after
cancel or restart.
- Run recorder.cancel() in background thread from Ctrl+C handler to
keep the event loop responsive.
- Add shutdown() method called on /voice off to release audio resources.
- Fix silence timer reset during active speech: use dip tolerance for
_resume_start tracker so natural speech pauses (< 0.3s) don't prevent
the silence timer from being reset.
- Update tests to match persistent stream behavior.
AudioRecorder now auto-stops after 15 seconds if no speech is detected
(_has_spoken remains False). In quiet environments where ambient RMS
never exceeds the silence threshold (200), the recording would wait
indefinitely. The new _max_wait parameter fires the silence callback
after the timeout, triggering the normal "No speech detected" flow.
- Set max_retries=0 on the STT OpenAI client. The SDK default (2) honors
Groq's retry-after header (often 53s), blocking the thread for up to
~106s on rate limits. Voice STT should fail fast, not retry silently.
- Stop continuous recording mode after 3 consecutive no-speech cycles to
prevent infinite restart loops when nobody is talking.
- Set OpenAI client timeout=30s in transcribe_audio() — default 600s
blocks _voice_processing for 10 min if Groq/OpenAI stalls
- Move _voice_start_recording in _voice_stop_and_transcribe finally
block to a daemon thread (same pattern as Ctrl+B handler and
process_loop)
- Add _should_exit guard at top of _voice_start_recording so all 4
call sites respect shutdown without individual checks
- Replace sd.wait() with a poll loop + sd.stop() in play_beep().
sd.wait() calls Event.wait() without timeout — hangs forever if the
audio device stalls. Poll with a 2s ceiling and force-stop instead.
- Wrap _on_silence callback in try-except so exceptions are logged
instead of silently lost in the daemon thread. Prevents recording
state from becoming inconsistent on unexpected errors.
browser_tool.py registered SIGINT/SIGTERM handlers that called sys.exit()
at module import time. When a signal arrived during a lock acquisition
(e.g. AudioRecorder._lock in voice mode), SystemExit was raised inside
prompt_toolkit's async event loop, corrupting coroutine state and making
the process unkillable (required SIGKILL).
atexit handler already ensures browser sessions are cleaned up on any
normal exit path, so the signal handlers were redundant and harmful.
- edge_tts NameError: _generate_edge_tts now calls _import_edge_tts()
instead of referencing bare module name (tts_tool.py)
- TTS thread leak: chat() finally block sends sentinel to text_queue,
sets stop_event, and joins tts_thread on exception paths (cli.py)
- output_stream leak: moved close() into finally block so audio device
is released even on exception (tts_tool.py)
- Ctrl+C continuous mode: cancel handler now resets _voice_continuous
to prevent auto-restart after user cancels recording (cli.py)
- _disable_voice_mode: now calls stop_playback() and sets
_voice_tts_done so TTS stops when voice mode is turned off (cli.py)
- _show_voice_status: reads record key from config instead of
hardcoding Ctrl+B (cli.py)
- AudioRecorder.start() now catches InputStream errors gracefully
with a clear error message about microphone availability
- Fix config key mismatch: cli.py was reading "push_to_talk_key"
but config.py defines "record_key" -- now consistent
- Add format conversion from config format ("ctrl+b") to
prompt_toolkit format ("c-b")
1. Fully lazy imports: sounddevice, numpy, elevenlabs, edge_tts, and
openai are never imported at module level. Each is imported only when
the feature is explicitly activated, preventing crashes in headless
environments (SSH, Docker, WSL, no PortAudio).
2. No core agent loop changes: streaming TTS path extracted from
_interruptible_api_call() into separate _streaming_api_call() method.
The original method is restored to its upstream form.
3. Configurable key binding: push-to-talk key changed from Ctrl+R
(conflicts with readline reverse-search) to Ctrl+B by default.
Configurable via voice.push_to_talk_key in config.yaml.
4. Environment detection: new detect_audio_environment() function checks
for SSH, Docker, WSL, and missing audio devices before enabling voice
mode. Auto-disables with clear warnings in incompatible environments.
5. Graceful degradation: every audio touchpoint (sd.play, sd.InputStream,
sd.OutputStream) wrapped in try/except with ImportError/OSError
handling. Failures produce warnings, not crashes.
- Fix Gemini streaming tool call merge bug: multiple tool calls with same
index but different IDs are now parsed as separate calls instead of
concatenating names (e.g. ha_call_serviceha_call_service)
- Handle partial results in voice mode: show error and stop continuous
mode when agent returns partial/failed results with empty response
- Fix error display during streaming TTS: error messages are shown in
full response box even when streaming box was already opened
- Add duplicate sentence filter in TTS: skip near-duplicate sentences
from LLM repetition
- Fix fake HA server state mutation: turn_on/turn_off/set_temperature
correctly update entity states; temperature sensor simulates change
when thermostat is adjusted
- Add _vprint() helper to suppress log output when stream_callback is active
- Expand Whisper hallucination filter with multi-language phrases and regex pattern for repetitive text
- Stop continuous voice mode when agent returns a failed result (e.g. 429 rate limit)
- Atomic check-and-set for _voice_recording flag with _voice_lock
- Guard _voice_stop_and_transcribe against concurrent invocation
- Remove premature flag clearing from Ctrl+R handler
- Clean up temp WAV files in finally block (_play_via_tempfile)
- Use buffer-level regex for <think> block filtering (handles chunked tags)
- Prevent /voice on prompt accumulation on repeated calls
- Include Groq in STT key error message
Move screen output from stream_callback to display_callback called by
TTS consumer thread. Text now appears sentence-by-sentence in sync with
audio instead of streaming ahead at LLM speed. Removes quiet_mode hack.
sounddevice raises OSError (not ImportError) when the PortAudio C
library is missing. This broke test collection on CI runners that
have the Python package installed but lack the native library.
Stream audio to speaker as the agent generates tokens instead of
waiting for the full response. First sentence plays within ~1-2s
of agent starting to respond.
- run_agent: add stream_callback to run_conversation/chat, streaming
path in _interruptible_api_call accumulates chunks into mock
ChatCompletion while forwarding content deltas to callback
- tts_tool: add stream_tts_to_speaker() with sentence buffering,
think block filtering, markdown stripping, ElevenLabs pcm_24000
streaming to sounddevice OutputStream
- cli: wire up streaming TTS pipeline in chat(), detect elevenlabs
provider + sounddevice availability, skip batch TTS when streaming
is active, signal stop on interrupt
Falls back to batch TTS for Edge/OpenAI providers or when
elevenlabs/sounddevice are not available. Zero impact on non-voice
mode (callback defaults to None).
- Audio cues: beep on record start (880Hz), double beep on stop (660Hz)
- Silence detection: auto-stop recording after 3s of silence (RMS-based)
- Continuous mode: auto-restart recording after agent responds
- Ctrl+R starts continuous mode, Ctrl+R during recording exits it
- Waits for TTS to finish before restarting to avoid recording speaker
- Tests: 7 new tests for beep generation and silence detection
- Add multi-provider STT support (OpenAI > Groq fallback) in transcription_tools
- Auto-correct model selection when provider doesn't support the configured model
- Change voice record key from Ctrl+Space to Ctrl+R (macOS compatibility)
- Fix duplicate transcript echo in voice pipeline
- Add GROQ_API_KEY to .env.example
- prevent raw MEDIA tag leakage outside the gateway pipeline
- make extract_media handle quoted/backticked paths and optional whitespace
- send Telegram media natively with explicit error/warning handling
- add regression tests for Telegram media dispatch and MEDIA parsing
The send_message tool's _send_telegram() sent MEDIA:<path> tags as
literal text instead of delivering actual files. This fixes it by
extracting MEDIA tags via BasePlatformAdapter.extract_media() and
routing files to the appropriate Telegram Bot API method by extension.
Changes:
- send_message_tool: extract MEDIA tags and send files natively as
photo/video/voice/audio/document based on file extension
- send_message_tool: add per-file error handling and missing-file logging
- send_message_tool: use cleaned text in fallback to avoid leaking tags
- base.py extract_media: handle optional space after MEDIA: colon
- base.py extract_media: strip surrounding backticks/quotes from paths
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* feat: improve context compaction handoff summaries
Adapt PR #916 onto current main by replacing the old context summary marker
with a clearer handoff wrapper, updating the summarization prompt for
resume-oriented summaries, and preserving the current call_llm-based
compression path.
* fix: clearer error when docker backend is unavailable
* fix: preserve docker discovery in backend preflight
Follow up on salvaged PR #940 by reusing find_docker() during the new
availability check so non-PATH Docker Desktop installs still work. Add
a regression test covering the resolved executable path.
---------
Co-authored-by: aydnOktay <xaydinoktay@gmail.com>
Restore gateway/run.py to current main behavior while keeping tirith startup
and pattern_keys replay, preserve yolo and non-interactive bypass semantics in
the combined guard, and add regression tests for yolo and view-full flows.
Integrate tirith as a pre-execution security scanner that detects
homograph URLs, pipe-to-interpreter patterns, terminal injection,
zero-width Unicode, and environment variable manipulation — threats
the existing 50-pattern dangerous command detector doesn't cover.
Architecture: gather-then-decide — both tirith and the dangerous
command detector run before any approval prompt, preventing gateway
force=True replay from bypassing one check when only the other was
shown to the user.
New files:
- tools/tirith_security.py: subprocess wrapper with auto-installer,
mandatory cosign provenance verification, non-blocking background
download, disk-persistent failure markers with retryable-cause
tracking (cosign_missing auto-clears when cosign appears on PATH)
- tests/tools/test_tirith_security.py: 62 tests covering exit code
mapping, fail_open, cosign verification, background install,
HERMES_HOME isolation, and failure recovery
- tests/tools/test_command_guards.py: 21 integration tests for the
combined guard orchestration
Modified files:
- tools/approval.py: add check_all_command_guards() orchestrator,
add allow_permanent parameter to prompt_dangerous_approval()
- tools/terminal_tool.py: replace _check_dangerous_command with
consolidated check_all_command_guards
- cli.py: update _approval_callback for allow_permanent kwarg,
call ensure_installed() at startup
- gateway/run.py: iterate pattern_keys list on replay approval,
call ensure_installed() at startup
- hermes_cli/config.py: add security config defaults, split
commented sections for independent fallback
- cli-config.yaml.example: document tirith security config
Salvage of PR #1246 by ChatGPT (teknium1 session), resolved against
current main which already includes #1239.
Changes:
- Add minisweagent_path.py: worktree-aware helper that finds
mini-swe-agent/src from either the current checkout or the main
checkout behind a git worktree
- Use the helper in tools/terminal_tool.py and mini_swe_runner.py
instead of naive path-relative lookup that fails in worktrees
- Clean up check_terminal_requirements():
- local: return True (no minisweagent dep, per #1239)
- singularity/ssh: remove unnecessary minisweagent imports
- docker/modal: use importlib.util.find_spec with clear error
- Add regression tests for worktree path discovery and tool resolution
- treat git diff --cached --quiet rc=1 as an expected checkpoint state
instead of logging it as an error
- downgrade expected write PermissionError/EROFS/EACCES failures out of
error logging while keeping unexpected exceptions at error level
- add regression tests for both logging behaviors
* fix: Home Assistant event filtering now closed by default
Previously, when no watch_domains or watch_entities were configured,
ALL state_changed events passed through to the agent, causing users
to be flooded with notifications for every HA entity change.
Now events are dropped by default unless the user explicitly configures:
- watch_domains: list of domains to monitor (e.g. climate, light)
- watch_entities: list of specific entity IDs to monitor
- watch_all: true (new option — opt-in to receive all events)
A warning is logged at connect time if no filters are configured,
guiding users to set up their HA platform config.
All 49 gateway HA tests + 52 HA tool tests pass.
* docs: update Home Assistant integration documentation
- homeassistant.md: Fix event filtering docs to reflect closed-by-default
behavior. Add watch_all option. Replace Python dict config example with
YAML. Fix defaults table (was incorrectly showing 'all'). Add required
configuration warning admonition.
- environment-variables.md: Add HASS_TOKEN and HASS_URL to Messaging section.
- messaging/index.md: Add Home Assistant to description, architecture
diagram, platform toolsets table, and Next Steps links.
* fix(terminal): strip provider env vars from background and PTY subprocesses
Extends the env var blocklist from #1157 to also cover the two remaining
leaky paths in process_registry.py:
- spawn_local() PTY path (line 156)
- spawn_local() background Popen path (line 197)
Both were still using raw os.environ, leaking provider vars to background
processes and interactive PTY sessions. Now uses the same dynamic
_HERMES_PROVIDER_ENV_BLOCKLIST from local.py.
Explicit env_vars passed to spawn_local() still override the blocklist,
matching the existing behavior for callers that intentionally need these.
Gap identified by PR #1004 (@PeterFile).
* feat(delegate): add observability metadata to subagent results
Enrich delegate_task results with metadata from the child AIAgent:
- model: which model the child used
- exit_reason: completed | interrupted | max_iterations
- tokens.input / tokens.output: token counts
- tool_trace: per-tool-call trace with byte sizes and ok/error status
Tool trace uses tool_call_id matching to correctly pair parallel tool
calls with their results, with a fallback for messages without IDs.
Cherry-picked from PR #872 by @omerkaz, with fixes:
- Fixed parallel tool call trace pairing (was always updating last entry)
- Removed redundant 'iterations' field (identical to existing 'api_calls')
- Added test for parallel tool call trace correctness
Co-authored-by: omerkaz <omerkaz@users.noreply.github.com>
* feat(stt): add free local whisper transcription via faster-whisper
Replace OpenAI-only STT with a dual-provider system mirroring the TTS
architecture (Edge TTS free / ElevenLabs paid):
STT: faster-whisper local (free, default) / OpenAI Whisper API (paid)
Changes:
- tools/transcription_tools.py: Full rewrite with provider dispatch,
config loading, local faster-whisper backend, and OpenAI API backend.
Auto-downloads model (~150MB for 'base') on first voice message.
Singleton model instance reused across calls.
- pyproject.toml: Add faster-whisper>=1.0.0 as core dependency
- hermes_cli/config.py: Expand stt config to match TTS pattern with
provider selection and per-provider model settings
- agent/context_compressor.py: Fix .strip() crash when LLM returns
non-string content (dict from llama.cpp, None). Fixes#1100 partially.
- tests/: 23 new tests for STT providers + 2 for compressor fix
- docs/: Updated Voice & TTS page with STT provider table, model sizes,
config examples, and fallback behavior
Fallback behavior:
- Local not installed → OpenAI API (if key set)
- OpenAI key not set → local whisper (if installed)
- Neither → graceful error message to user
Co-authored-by: Jah-yee <Jah-yee@users.noreply.github.com>
---------
Co-authored-by: omerkaz <omerkaz@users.noreply.github.com>
Co-authored-by: Jah-yee <Jah-yee@users.noreply.github.com>
* fix: Home Assistant event filtering now closed by default
Previously, when no watch_domains or watch_entities were configured,
ALL state_changed events passed through to the agent, causing users
to be flooded with notifications for every HA entity change.
Now events are dropped by default unless the user explicitly configures:
- watch_domains: list of domains to monitor (e.g. climate, light)
- watch_entities: list of specific entity IDs to monitor
- watch_all: true (new option — opt-in to receive all events)
A warning is logged at connect time if no filters are configured,
guiding users to set up their HA platform config.
All 49 gateway HA tests + 52 HA tool tests pass.
* docs: update Home Assistant integration documentation
- homeassistant.md: Fix event filtering docs to reflect closed-by-default
behavior. Add watch_all option. Replace Python dict config example with
YAML. Fix defaults table (was incorrectly showing 'all'). Add required
configuration warning admonition.
- environment-variables.md: Add HASS_TOKEN and HASS_URL to Messaging section.
- messaging/index.md: Add Home Assistant to description, architecture
diagram, platform toolsets table, and Next Steps links.
* fix(terminal): strip provider env vars from background and PTY subprocesses
Extends the env var blocklist from #1157 to also cover the two remaining
leaky paths in process_registry.py:
- spawn_local() PTY path (line 156)
- spawn_local() background Popen path (line 197)
Both were still using raw os.environ, leaking provider vars to background
processes and interactive PTY sessions. Now uses the same dynamic
_HERMES_PROVIDER_ENV_BLOCKLIST from local.py.
Explicit env_vars passed to spawn_local() still override the blocklist,
matching the existing behavior for callers that intentionally need these.
Gap identified by PR #1004 (@PeterFile).
* feat(delegate): add observability metadata to subagent results
Enrich delegate_task results with metadata from the child AIAgent:
- model: which model the child used
- exit_reason: completed | interrupted | max_iterations
- tokens.input / tokens.output: token counts
- tool_trace: per-tool-call trace with byte sizes and ok/error status
Tool trace uses tool_call_id matching to correctly pair parallel tool
calls with their results, with a fallback for messages without IDs.
Cherry-picked from PR #872 by @omerkaz, with fixes:
- Fixed parallel tool call trace pairing (was always updating last entry)
- Removed redundant 'iterations' field (identical to existing 'api_calls')
- Added test for parallel tool call trace correctness
Co-authored-by: omerkaz <omerkaz@users.noreply.github.com>
---------
Co-authored-by: omerkaz <omerkaz@users.noreply.github.com>
* fix: Home Assistant event filtering now closed by default
Previously, when no watch_domains or watch_entities were configured,
ALL state_changed events passed through to the agent, causing users
to be flooded with notifications for every HA entity change.
Now events are dropped by default unless the user explicitly configures:
- watch_domains: list of domains to monitor (e.g. climate, light)
- watch_entities: list of specific entity IDs to monitor
- watch_all: true (new option — opt-in to receive all events)
A warning is logged at connect time if no filters are configured,
guiding users to set up their HA platform config.
All 49 gateway HA tests + 52 HA tool tests pass.
* docs: update Home Assistant integration documentation
- homeassistant.md: Fix event filtering docs to reflect closed-by-default
behavior. Add watch_all option. Replace Python dict config example with
YAML. Fix defaults table (was incorrectly showing 'all'). Add required
configuration warning admonition.
- environment-variables.md: Add HASS_TOKEN and HASS_URL to Messaging section.
- messaging/index.md: Add Home Assistant to description, architecture
diagram, platform toolsets table, and Next Steps links.
* fix(terminal): strip provider env vars from background and PTY subprocesses
Extends the env var blocklist from #1157 to also cover the two remaining
leaky paths in process_registry.py:
- spawn_local() PTY path (line 156)
- spawn_local() background Popen path (line 197)
Both were still using raw os.environ, leaking provider vars to background
processes and interactive PTY sessions. Now uses the same dynamic
_HERMES_PROVIDER_ENV_BLOCKLIST from local.py.
Explicit env_vars passed to spawn_local() still override the blocklist,
matching the existing behavior for callers that intentionally need these.
Gap identified by PR #1004 (@PeterFile).
Terminal subprocesses inherit OPENAI_BASE_URL and other provider env
vars loaded from ~/.hermes/.env, silently misrouting external CLIs
like codex. Build a blocklist dynamically from the provider registry
so new providers are automatically covered. Callers that truly need
a blocked var can opt in via the _HERMES_FORCE_ prefix.
Closes#1002
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
The old message referenced 'hermes setup' which doesn't handle
skill-specific env vars. Updated to direct users to load the skill
in the local CLI (which triggers the secure prompt) or add the key
to ~/.hermes/.env manually.
When a skill declares required_environment_variables in its YAML
frontmatter, missing env vars trigger a secure TUI prompt (identical
to the sudo password widget) when the skill is loaded. Secrets flow
directly to ~/.hermes/.env, never entering LLM context.
Key changes:
- New required_environment_variables frontmatter field for skills
- Secure TUI widget (masked input, 120s timeout)
- Gateway safety: messaging platforms show local setup guidance
- Legacy prerequisites.env_vars normalized into new format
- Remote backend handling: conservative setup_needed=True
- Env var name validation, file permissions hardened to 0o600
- Redact patterns extended for secret-related JSON fields
- 12 existing skills updated with prerequisites declarations
- ~48 new tests covering skip, timeout, gateway, remote backends
- Dynamic panel widget sizing (fixes hardcoded width from original PR)
Cherry-picked from PR #723 by kshitijk4poor, rebased onto current main
with conflict resolution.
Fixes#688
Co-authored-by: kshitijk4poor <kshitijk4poor@users.noreply.github.com>
The tinker-atropos submodule and its heavy dependencies (atroposlib, tinker,
wandb, fastapi, uvicorn) were being installed for all users by default,
adding significant install time and disk usage for most users who don't
need RL training capabilities.
Changes:
- install.sh: Only init mini-swe-agent submodule by default; skip
tinker-atropos clone and install entirely
- install.sh: Remove --recurse-submodules from git clone (only fetches
what's needed)
- pyproject.toml: Add [rl] optional dependency group for explicit opt-in
- rl_training_tool.py: Move LOGS_DIR.mkdir() from module-level to lazy
init (_ensure_logs_dir) to avoid side effects on import
- README.md: Update contributor quick start to not auto-fetch
tinker-atropos; add RL opt-in instructions
Users who want RL training can opt in with:
git submodule update --init tinker-atropos
uv pip install -e ./tinker-atropos
The ClawHub API v1 version endpoint only returns file metadata
(path, size, sha256, contentType) without inline content or download
URLs. Our code was looking for inline content in the metadata, which
never existed, causing all ClawHub installs to fail with:
'no inline/raw file content was available'
Fix: Use the /api/v1/download endpoint (same as the official clawhub
CLI) to download skills as ZIP bundles and extract files in-memory.
Changes:
- Add _download_zip() method that downloads and extracts ZIP bundles
- Retry on 429 rate limiting with Retry-After header support
- Path sanitization and binary file filtering for security
- Keep _extract_files() as a fallback for inline/raw content
- Also fix nested file lookup (version_data.version.files)
When a dangerous command is detected and the user is prompted for
approval, long commands are truncated (80 chars in fallback, 70 chars
in the TUI). Users had no way to see the full command before deciding.
This adds a 'View full command' option across all approval interfaces:
- CLI fallback (tools/approval.py): [v]iew option in the prompt menu.
Shows the full command and re-prompts for approval decision.
- CLI TUI (cli.py): 'Show full command' choice in the arrow-key
selection panel. Expands the command display in-place and removes
the view option after use.
- CLI callbacks (callbacks.py): 'view' choice added to the list when
the command exceeds 70 characters.
- Gateway (gateway/run.py): 'full', 'show', 'view' responses reveal
the complete command while keeping the approval pending.
Includes 7 new tests covering view-then-approve, view-then-deny,
short command fallthrough, and double-view behavior.
Closes community feedback about the 80-char cap on dangerous commands.
- gateway/run.py: Take main's _resolve_gateway_model() helper
- hermes_cli/setup.py: Re-apply nous-api removal after merge brought
it back. Fix provider_idx offset (Custom is now index 3, not 4).
- tests/hermes_cli/test_setup.py: Fix custom setup test index (3→4)
Add centralized call_llm() and async_call_llm() functions that own the
full LLM request lifecycle:
1. Resolve provider + model from task config or explicit args
2. Get or create a cached client for that provider
3. Format request args (max_tokens handling, provider extra_body)
4. Make the API call with max_tokens/max_completion_tokens retry
5. Return the response
Config: expanded auxiliary section with provider:model slots for all
tasks (compression, vision, web_extract, session_search, skills_hub,
mcp, flush_memories). Config version bumped to 7.
Migrated all auxiliary consumers:
- context_compressor.py: uses call_llm(task='compression')
- vision_tools.py: uses async_call_llm(task='vision')
- web_tools.py: uses async_call_llm(task='web_extract')
- session_search_tool.py: uses async_call_llm(task='session_search')
- browser_tool.py: uses call_llm(task='vision'/'web_extract')
- mcp_tool.py: uses call_llm(task='mcp')
- skills_guard.py: uses call_llm(provider='openrouter')
- run_agent.py flush_memories: uses call_llm(task='flush_memories')
Tests updated for context_compressor and MCP tool. Some test mocks
still need updating (15 remaining failures from mock pattern changes,
2 pre-existing).
Route all remaining ad-hoc auxiliary LLM call sites through
resolve_provider_client() so auth, headers, and API format (Chat
Completions vs Responses API) are handled consistently in one place.
Files changed:
- tools/openrouter_client.py: Replace manual AsyncOpenAI construction
with resolve_provider_client('openrouter', async_mode=True). The
shared client module now delegates entirely to the router.
- tools/skills_guard.py: Replace inline OpenAI client construction
(hardcoded OpenRouter base_url, manual api_key lookup, manual
headers) with resolve_provider_client('openrouter'). Remove unused
OPENROUTER_BASE_URL import.
- trajectory_compressor.py: Add _detect_provider() to map config
base_url to a provider name, then route through
resolve_provider_client. Falls back to raw construction for
unrecognized custom endpoints.
- mini_swe_runner.py: Route default case (no explicit api_key/base_url)
through resolve_provider_client('openrouter') with auto-detection
fallback. Preserves direct construction when explicit creds are
passed via CLI args.
- agent/auxiliary_client.py: Fix stale module docstring — vision auto
mode now correctly documents that Codex and custom endpoints are
tried (not skipped).
Three interconnected fixes for auxiliary client infrastructure:
1. CENTRALIZED PROVIDER ROUTER (auxiliary_client.py)
Add resolve_provider_client(provider, model, async_mode) — a single
entry point for creating properly configured clients. Given a provider
name and optional model, it handles auth lookup (env vars, OAuth
tokens, auth.json), base URL resolution, provider-specific headers,
and API format differences (Chat Completions vs Responses API for
Codex). All auxiliary consumers should route through this instead of
ad-hoc env var lookups.
Refactored get_text_auxiliary_client, get_async_text_auxiliary_client,
and get_vision_auxiliary_client to use the router internally.
2. FIX CODEX VISION BYPASS (vision_tools.py)
vision_tools.py was constructing a raw AsyncOpenAI client from the
sync vision client's api_key/base_url, completely bypassing the Codex
Responses API adapter. When the vision provider resolved to Codex,
the raw client would hit chatgpt.com/backend-api/codex with
chat.completions.create() which only supports the Responses API.
Fix: Added get_async_vision_auxiliary_client() which properly wraps
Codex into AsyncCodexAuxiliaryClient. vision_tools.py now uses this
instead of manual client construction.
3. FIX COMPRESSION FALLBACK + VISION ERROR HANDLING
- context_compressor.py: Removed _get_fallback_client() which blindly
looked for OPENAI_API_KEY + OPENAI_BASE_URL (fails for Codex OAuth,
API-key providers, users without OPENAI_BASE_URL set). Replaced
with fallback loop through resolve_provider_client() for each
known provider, with same-provider dedup.
- vision_tools.py: Added error detection for vision capability
failures. Returns clear message to the model when the configured
model doesn't support vision, instead of a generic error.
Addresses #886
Adds full stack traces to error logs in _upscale_image() and
image_generate_tool() for better debugging. Matches the pattern
used across the rest of the codebase.
Cherry-picked from PR #868 by aydnOktay.
Co-authored-by: aydnOktay <aydnOktay@users.noreply.github.com>
- Log command, return code, and stderr on non-zero exit
- Add exc_info=True to timeout, FileNotFoundError, and catch-all handlers
- Add debug field to restore() error responses with raw git output
- Keeps user-facing error messages clean while preserving detail for debugging
Inspired by PR #843 (aydnOktay).
Previously the early return for unconfigured vision model was silent.
Now logs an error so the failure is visible in logs for debugging.
Inspired by PR #839 by aydnOktay.
Co-authored-by: aydnOktay <aydnOktay@users.noreply.github.com>
Address merge-blocking review feedback by removing unsafe signal handler overrides, wiring next-turn Honcho prefetch, restoring per-directory session defaults, and exposing all Honcho tools to the model surface. Also harden prefetch cache access with public thread-safe accessors and remove duplicate browser cleanup code.
Made-with: Cursor
_discover_one() caught all exceptions and returned [], making
asyncio.gather(return_exceptions=True) redundant. The
isinstance(result, Exception) branch in _discover_all() was dead
code, so failed_count was always 0. This caused:
- No summary printed when all servers fail (silent failure)
- ok_servers always equaling total_servers (misleading count)
- Unused variables transport_desc and transport_type
Fix: let exceptions propagate to gather() so failed_count increments
correctly. Move per-server failure logging to _discover_all(). Remove
dead variables.
Fixes discovered while running TBLite baseline evaluation:
1. ephemeral_disk param not supported in modal 1.3.5 - check before passing
2. Modal legacy image builder requires working pip - add ensurepip fix via
setup_dockerfile_commands to handle task images with broken pip
3. Host cwd leaked into Modal sandbox - add /home/ to host prefix check
4. Tilde ~ not expanded by subprocess.run(cwd=) in sandboxes - use /root
5. install_pipx must stay True for swerex-remote to be available
Dependencies also needed (not in this commit):
- git submodule update --init mini-swe-agent
- uv pip install swe-rex boto3
Allow users to interact with Hermes by sending and receiving emails.
Uses IMAP polling for incoming messages and SMTP for replies with
proper threading (In-Reply-To, References headers).
Integrates with all 14 gateway extension points: config, adapter
factory, authorization, send_message tool, cron delivery, toolsets,
prompt hints, channel directory, setup wizard, status display, and
env example.
65 tests covering config, parsing, dispatch, threading, IMAP fetch,
SMTP send, attachments, and all integration points.
Adds delegation.model and delegation.provider config fields so subagents
can run on a completely different provider:model pair than the parent agent.
When delegation.provider is set, the system resolves the full credential
bundle (base_url, api_key, api_mode) via resolve_runtime_provider() —
the same path used by CLI/gateway startup. This means all configured
providers work out of the box: openrouter, nous, zai, kimi-coding,
minimax, minimax-cn.
Key design decisions:
- Provider resolution uses hermes_cli.runtime_provider (single source of
truth for credential resolution across CLI, gateway, cron, and now
delegation)
- When only delegation.model is set (no provider), the model name changes
but parent credentials are inherited (for switching models within the
same provider like OpenRouter)
- When delegation.provider is set, full credentials are resolved
independently — enabling cross-provider delegation (e.g. parent on
Nous Portal, subagents on OpenRouter)
- Clear error messages if provider resolution fails (missing API key,
unknown provider name)
- _load_config() now falls back to hermes_cli.config.load_config() for
gateway/cron contexts where CLI_CONFIG is unavailable
Based on PR #791 by 0xbyt4 (closes#609), reworked to use proper
provider credential resolution instead of passing provider as metadata.
Co-authored-by: 0xbyt4 <0xbyt4@users.noreply.github.com>
Adds defensive guard against empty/None/missing choices in SamplingHandler.__call__
before accessing response.choices[0]. Returns proper ErrorData instead of crashing
with IndexError/TypeError on content filtering, provider errors, or rate limits.
Authored by 0xbyt4.
Co-authored-by: 0xbyt4 <0xbyt4@users.noreply.github.com>
Stray print() in write_file_tool exception handler leaked debug output
to stdout. Replaced with logger.error() which is already set up in
the file.
Authored by memosr.
Co-authored-by: memosr <memosr@users.noreply.github.com>
- Add `agent`, `tools.*`, `gateway.*` to packages.find include
- Add `hermes_state`, `hermes_time`, `mini_swe_runner`, `rl_cli`, `utils` to py-modules
- Move rl_training_tool LOGS_DIR to ~/.hermes/logs/rl_training/ (was writing
into the package source tree, which fails on read-only installs)
These were masked in development (editable installs see the whole source tree)
but broke any non-editable install like `pip install .` or wheel builds.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Isolate Telegram forum topic sessions — each topic gets its own independent session key, history, and interrupt tracking. Progress, hygiene, and cron messages all route to the correct topic.
Four cleanups to code merged today:
1. New hermes_cli/curses_ui.py — shared curses_checklist() used by both
hermes tools and hermes skills. Eliminates ~140 lines of near-identical
curses code (scrolling, key handling, color setup, numbered fallback).
2. Fix _find_all_skills() perf — was calling load_config() per skill
(~100+ YAML parses). Now loads disabled set once via
_get_disabled_skill_names() and does a set lookup.
3. Eliminate _list_all_skills_unfiltered() duplication — _find_all_skills()
now accepts skip_disabled=True for the config UI, removing 30 lines
of copy-pasted discovery logic from skills_config.py.
4. Fix fragile label round-trip in skills_command — was building label
strings, passing to checklist, then mapping labels back to skill names
(collision-prone). Now works with indices directly, like tools_config.
Replaces head-only stdout capture with a two-buffer approach (40% head,
60% tail rolling window) so scripts that print() their final results
at the end never lose them. Adds truncation notice between sections.
Cherry-picked from PR #755, conflict resolved (test file additions).
3 new tests for short output, head+tail preservation, and notice format.
Authored by dmahan93. Adds HERMES_YOLO_MODE env var and --yolo CLI flag
to auto-approve all dangerous command prompts.
Post-merge: renamed --fuck-it-ship-it to --yolo for brevity,
resolved conflict with --checkpoints flag.
On macOS, Docker Desktop installs the CLI to /usr/local/bin/docker, but
when Hermes runs as a gateway service (launchd) or in other non-login
contexts, /usr/local/bin is often not in PATH. This causes the Docker
requirements check to fail with 'No such file or directory: docker' even
though docker works fine from the user's terminal.
Add find_docker() helper that uses shutil.which() first, then probes
common Docker Desktop install paths on macOS (/usr/local/bin,
/opt/homebrew/bin, Docker.app bundle). The resolved path is cached and
passed to mini-swe-agent via its 'executable' parameter.
- tools/environments/docker.py: add find_docker(), use it in
_storage_opt_supported() and pass to _Docker(executable=...)
- tools/terminal_tool.py: use find_docker() in requirements check
- tests/tools/test_docker_find.py: 4 tests (PATH, fallback, not found, cache)
2877 tests pass.
The 4 early-return paths in _spawn_training_run (API exit, trainer
exit, env not found, env exit) were doing manual process.terminate()
or returning without cleanup, leaking open log file handles. Now all
paths call _stop_training_run() which handles both process termination
and file handle closure.
Also adds 12 tests for _stop_training_run covering file handle
cleanup, process termination, status transitions, and edge cases.
Inspired by PR #715 (0xbyt4) which identified the early-return issue.
Core file handle fix was already on main via e28dc13 (memosr.eth).
Follow-up to PR #705 (merged from 0xbyt4). Addresses several issues:
1. CONSECUTIVE-ONLY TRACKING: Redesigned the read/search tracker to only
warn/block on truly consecutive identical calls. Any other tool call
in between (write, patch, terminal, etc.) resets the counter via
notify_other_tool_call(), called from handle_function_call() in
model_tools.py. This prevents false blocks in read→edit→verify flows.
2. THRESHOLD ADJUSTMENT: Warn on 3rd consecutive (was 2nd), block on
4th+ consecutive (was 3rd+). Gives the model more room before
intervening.
3. TUPLE UNPACKING BUG: Fixed get_read_files_summary() which crashed on
search keys (5-tuple) when trying to unpack as 3-tuple. Now uses a
separate read_history set that only tracks file reads.
4. WEB_EXTRACT DOCSTRING: Reverted incorrect removal of 'title' from
web_extract return docs in code_execution_tool.py — the field IS
returned by web_tools.py.
5. TESTS: Rewrote test_read_loop_detection.py (35 tests) to cover
consecutive-only behavior, notify_other_tool_call, interleaved
read/search, and summary-unaffected-by-searches.
New tool lets Hermes persist conclusions about the user (preferences,
corrections, project context) directly to Honcho via the conclusions
API. Feeds into the user's peer card and representation.
Authored by johnh4098. Fixes CWE-214: SUDO_PASSWORD was visible in
/proc/PID/cmdline via echo pipe. Now passed through subprocess stdin.
All 6 backends updated: local, ssh, docker, singularity pipe via stdin;
modal and daytona use printf fallback (remote sandbox, documented).
Completes the fix started in 8318a51 — handle_function_call() accepted
enabled_tools but run_agent.py never passed it. Now both call sites in
_execute_tool_calls() pass self.valid_tool_names, so each agent session
uses its own tool list instead of the process-global
_last_resolved_tool_names (which subagents can overwrite).
Also simplifies the redundant ternary in code_execution_tool.py:
sandbox_tools is already computed correctly (intersection with session
tools, or full SANDBOX_ALLOWED_TOOLS as fallback), so the conditional
was dead logic.
Inspired by PR #663 (JasonOA888). Closes#662.
Tests: 2857 passed.
Authored by manuelschipper. Adds missing docker_volumes key to
container_config in file_tools.py, matching terminal_tool.py.
Without this, Docker sandbox containers created by file operations
lack user volume mounts when file tools run before terminal.
build_execute_code_schema(set()) produced "from hermes_tools import , ..."
in the code property description — invalid Python syntax shown to the model.
This triggers when a user enables only the code_execution toolset without
any of the sandbox-allowed tools (e.g. `hermes tools code_execution`),
because SANDBOX_ALLOWED_TOOLS & {"execute_code"} = empty set.
Also adds 29 unit tests covering build_execute_code_schema, environment
variable filtering, execute_code edge cases, and interrupt handling.
Authored by voteblake.
- Semaphore limits concurrent Modal sandbox creations to 8 (configurable)
to prevent thread pool deadlocks when 86+ tasks fire simultaneously
- Modal cleanup guard for failed init (prevents AttributeError)
- CWD override to /app for TB2 containers
- Add /home/ to host path validation for container backends
Authored by aydnOktay. Adds logging to skills_tool.py with specific
exception handling for file read errors (UnicodeDecodeError, PermissionError)
vs unexpected exceptions, replacing bare except-and-continue blocks.
file_tools.py creates its own Docker sandbox when read_file/search_files
runs before any terminal command. The container_config was missing
docker_volumes, so the sandbox had no user volume mounts — breaking
access to heartbeat state, cron output, and all other mounted data.
Matches the existing pattern in terminal_tool.py:872.
Missed in original PR #158 (feat: add docker_volumes config).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Automatic filesystem snapshots before destructive file operations,
with user-facing rollback. Inspired by PR #559 (by @alireza78a).
Architecture:
- Shadow git repos at ~/.hermes/checkpoints/{hash}/ via GIT_DIR
- CheckpointManager: take/list/restore, turn-scoped dedup, pruning
- Transparent — the LLM never sees it, no tool schema, no tokens
- Once per turn — only first write_file/patch triggers a snapshot
Integration:
- Config: checkpoints.enabled + checkpoints.max_snapshots
- CLI flag: hermes --checkpoints
- Trigger: run_agent.py _execute_tool_calls() before write_file/patch
- /rollback slash command in CLI + gateway (list, restore by number)
- Pre-rollback snapshot auto-created on restore (undo the undo)
Safety:
- Never blocks file operations — all errors silently logged
- Skips root dir, home dir, dirs >50K files
- Disables gracefully when git not installed
- Shadow repo completely isolated from project git
Tests: 35 new tests, all passing (2798 total suite)
Docs: feature page, config reference, CLI commands reference
PR #568 moved the close entirely to the finally block, but the success-path
close is needed to break the RPC thread out of accept() immediately. Without
it, rpc_thread.join(3) may block for up to 3 seconds if the child process
never connected. The finally-block close remains as a safety net for the
exception/error path (the actual fd leak fix).
Authored by alireza78a. Moves server_sock.close() into the finally block so
the socket fd is always cleaned up, even if an exception occurs between socket
creation and the success-path close.
Authored by aydnOktay. Adds _atomic_write_text() helper using tempfile.mkstemp()
+ os.replace() to prevent skill file corruption on crash/interrupt. All 7
write_text() calls in skill_manager_tool.py converted, including rollback writes
during security scans.
Authored by Himess. Three independent fixes:
- cron/jobs.py: respect HERMES_HOME env var (consistent with scheduler.py)
- gateway/run.py: add Platform.HOMEASSISTANT to toolset mappings
- tools/environments/daytona.py: use time.monotonic() for timeout deadline
Authored by Himess. Replaces split(':', 2) with regex that optionally
captures Windows drive-letter prefix in rg/grep output parsing. Fixes
search_files returning zero results on Windows where paths like
C:\path\file.py:42:content were misparsed by naive colon splitting.
No behavior change on Unix/Mac.
Authored by shitcoinsherpa. Adds encoding='utf-8' to all text-mode
open() calls in gateway/run.py, gateway/config.py, hermes_cli/config.py,
hermes_cli/main.py, and hermes_cli/status.py. Prevents encoding errors
on Windows where the default locale is not UTF-8.
Also fixed 4 additional open() calls in gateway/run.py that were added
after the PR branch was created.
The existing password prompt uses /dev/tty and termios to read input
with echo disabled. Neither exists on Windows.
On Windows, msvcrt.getwch() reads a single character from the console
without echoing it. This adds a Windows code path that uses getwch()
in a loop, collecting characters until Enter is pressed.
The Unix path using termios and /dev/tty is unchanged.
Authored by shitcoinsherpa. Imports winpty.PtyProcess on Windows instead
of ptyprocess.PtyProcess, and adds platform markers to the [pty] extra
so the correct package is installed automatically.
cap-drop ALL removes DAC_OVERRIDE, which root needs to write to
bind-mounted directories owned by the host user (uid 1000). This
broke persistent Docker sandboxes — the container couldn't write
to /workspace or /root.
Add back the minimum capabilities needed:
- DAC_OVERRIDE: root can write to bind-mounted dirs owned by host user
- CHOWN: package managers (pip, npm, apt) need to set file ownership
- FOWNER: needed for operations on files owned by other users
Still drops all other capabilities (NET_RAW, SYS_ADMIN, etc.) and
keeps no-new-privileges. Security boundary is the container itself.
Verified end-to-end: create files → destroy container → new container
with same task_id → files persist on host and are accessible in the
new container.
Validate that the username portion of ~username paths contains only
valid characters (alphanumeric, dot, hyphen, underscore) before passing
to shell echo for expansion. Previously, paths like '~; rm -rf /'
would be passed unquoted to self._exec(f'echo {path}'), allowing
arbitrary command execution.
The approach validates the username rather than using shlex.quote(),
which would prevent tilde expansion from working at all since
echo '~user' outputs the literal string instead of expanding it.
Added tests for injection blocking and valid ~username/path expansion.
Credit to @alireza78a for reporting (PR #442, issue #442).
SamplingHandler.__call__ accessed response.choices[0] without checking
if the list was non-empty. LLM APIs can return empty choices on content
filtering, provider errors, or rate limits, causing an unhandled
IndexError that propagates to the MCP SDK and may crash the connection.
Add a defensive guard that returns a proper ErrorData when choices is
empty, None, or missing. Includes three test cases covering all
variants.
Authored by aydnOktay. Improves URL validation with urlparse, adds exc_info
to error logs for full stack traces, and tightens type hints.
Resolved merge conflict in _handle_vision_analyze: kept PR's string formatting
with our AUXILIARY_VISION_MODEL env var logic.
Add MCP sampling/createMessage capability via SamplingHandler class.
Text-only sampling + tool use in sampling with governance (rate limits,
model whitelist, token caps, tool loop limits). Per-server audit metrics.
Based on concept from PR #366 by eren-karakus0. Restructured as class-based
design with bug fixes and tests using real MCP SDK types.
50 new tests, 2600 total passing.
Combine read/search loop detection with main's redact_sensitive_text
and truncation hint features. Add tracker reset to TestSearchHints
to prevent cross-test state leakage.
Terminal output was already redacted via redact_sensitive_text() but
read_file and search_files returned raw content. Now both tools
redact secrets before returning results to the LLM.
Based on PR #372 by @teyrebaz33 (closes#363) — applied manually
due to branch conflicts with the current codebase.
Uses temp file + fsync + os.replace() to avoid corruption if the
process crashes mid-write. Cleans up temp file on failure, logs
errors at debug level.
Based on PR #335 by @aydnOktay — adapted for the current v2
manifest format (name:hash).
Teknium
ddupont
Teknium
Dilee
Zhicheng Han
helix4u
Abd0r
Hirokazu Ogawa
luyao618
liuhao1024
Alexander Monas
Mason James
AJV20
Kowen Hao
LeonSGP43
Alan Chen
altmazza0-star
Andrew Ho
leon7609
Kailigithub
liuhao1024
Brian Su
stephen0110
ambition0802
Gutslabs
kshitij
misery-hl
jani
beardthelion
LeonSGP43
sprmn24
vominh1919
Chris Danis
briandevans
Yoimex
ClawdIA
lhysdl
ygd58
0668001438
阿泥豆
ms-alan
Ioodu
xyiy001
YAMAGUCHI Seiji
ideathinklab01-source
Emilien Domenge
pdonizete
nftpoetrist
yuehei
WanderWang
Yanzhong Su
dh
Kiala
Siddharth Balyan
Bartok9
JasonOA888
Mind-Dragon
hharry11
simbam99
johnncenae
Yukipukii1
y0shualee
0xDevNinja
Stephen Schoettler
Maxence Groine
Heltman
Ari Lotter
Shannon Sands
Scott Trinh
loongzhao
Ben Barclay
brooklyn!
Lyle Lengyel
Tranquil-Flow
Teknium
Teknium
Pony.Ma
墨綠BG
Christian Scheid
Teknium
voidborne-d
Ivan Tonov
bde3249023
hhuang91
alt-glitch
MorAlekss
AntAISecurityLab
Alexazhu
Amanuel Tilahun Bogale
Matt Maximo
say8hi
Devorun
Jefferson
j0sephz
MikeFac
Wysie
TaroballzChen
yuanhe
Julien Talbot
VantHoff
Loic Moncany
shushuzn
Ubuntu
WuTianyi123
ycbai
yudaiyan
Konstantinos Karachalios
niyoh
Jaffar Keikei
Kongxi
王强
pefontana
zhangguangtao
JackTheGit
teyrebaz33
ifrederico
AxDSan
Junass1
Aniruddha Adak
Allard
Mibayy
sjz-ks
Saurabh
handsdiff
etherman-os
ChimingLiu
liujinkun
Ubuntu
yeyitech
emozilla
kshitijk4poor
konsisumer
Markus Corazzione
Kovyrin Family Claw
jneeee
shin4
Ruzzgar
etcircle
bkadish
Greer Guthrie
Greer Guthrie
adybag14-cyber
N0nb0at
Dusk1e
walli
Junjun Zhang
haileymarshall
Al Sayed Hoota
dippwho
0xbyt4
alt-glitch
WAXLYY
0xbyt4
chqchshj
faishal
aaronagent
jjovalle99
hermes-agent-dhabibi
Hermes Agent
0xFrank-eth
KUSH42
coffee
angelos
win4r
tars
maxyangcn
Austin Pickett
dashed
Lumen Radley
mrshu
Vasanthdev2004
r266-tech
Marc Bickel
Mateus Scheuer Macedo
WAXLYY
charliekerfoot
BongSuCHOI
Mariano A. Nicolini
Awsh1
Dusk1e
donrhmexe
Xowiek
Git-on-my-level
LucidPaths
Octopus
acsezen
Jack
maymuneth
Nils
Robin Fernandes
memosr.eth
ctlst
Mibay
Ivelin Tenev
Test
hermes
cmcleay
rovle
Han
octo-patch
Bartok9
max
darya
sai-samarth
crazywriter1
Alex Ferrari
ShawnPana
Muhammet Eren Karakuş
aydnOktay
Oktay Aydin
Nikita
Synergy
anastazya
Dave Tist
Stable Genius
quabug
alireza78a
sheeki003
0xIbra
Erosika
aydnOktay
dmahan93
SPANISH FLU
johnh4098
Dev User
shitcoinsherpa
