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
Removed the hard block on base_url containing 'api.anthropic.com'.
Anthropic now offers an OpenAI-compatible /chat/completions endpoint,
so blocking their URL prevents legitimate use. If the endpoint isn't
compatible, the API call will fail with a proper error anyway.
Removed from: run_agent.py, mini_swe_runner.py
Updated test to verify Anthropic URLs are accepted.
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).
Authored by christomitov. Auto-detects sk-kimi- key prefix and routes
to api.kimi.com/coding/v1. Adds User-Agent header for Kimi Code API
compatibility. Legacy Moonshot keys continue to work unchanged.
Critical fixes:
- Add --worktree/-w to hermes_cli/main.py argparse (both chat
subcommand and top-level parser) so 'hermes -w' works via the
actual CLI entry point, not just 'python cli.py -w'
- Pass worktree flag through cmd_chat() kwargs to cli_main()
- Handle worktree attr in bare 'hermes' and --resume/--continue paths
Bug fixes in cli.py:
- Skip worktree creation for --list-tools/--list-toolsets (wasteful)
- Wrap git worktree subprocess.run in try/except (crash on timeout)
- Add stale worktree pruning on startup (_prune_stale_worktrees):
removes clean worktrees older than 24h left by crashed/killed sessions
Documentation updates:
- AGENTS.md: add --worktree to CLI commands table
- cli-config.yaml.example: add worktree config section
- website/docs/reference/cli-commands.md: add to core commands
- website/docs/user-guide/cli.md: add usage examples
- website/docs/user-guide/configuration.md: add config docs
Test improvements (17 → 31 tests):
- Stale worktree pruning (prune old clean, keep recent, keep dirty)
- Directory symlink via .worktreeinclude
- Edge cases (no commits, not a repo, pre-existing .worktrees/)
- CLI flag/config OR logic
- TERMINAL_CWD integration
- System prompt injection format
Add a --worktree (-w) flag to the hermes CLI that creates an isolated
git worktree for the session. This allows running multiple hermes-agent
instances concurrently on the same repo without file collisions.
How it works:
- On startup with -w: detects git repo, creates .worktrees/<session>/
with its own branch (hermes/<session-id>), sets TERMINAL_CWD to it
- Each agent works in complete isolation — independent HEAD, index,
and working tree, shared git object store
- On exit: auto-removes worktree and branch if clean, warns and
keeps if there are uncommitted changes
- .worktreeinclude file support: list gitignored files (.env, .venv/)
to auto-copy/symlink into new worktrees
- .worktrees/ is auto-added to .gitignore
- Agent gets a system prompt note about the worktree context
- Config support: set worktree: true in config.yaml to always enable
Usage:
hermes -w # Interactive mode in worktree
hermes -w -q "Fix issue #123" # Single query in worktree
# Or in config.yaml:
worktree: true
Includes 17 tests covering: repo detection, worktree creation,
independence verification, cleanup (clean/dirty), .worktreeinclude,
.gitignore management, and 10 concurrent worktrees.
Closes#652
Long-lived gateway sessions can accumulate enough history that every new
message rehydrates an oversized transcript, causing repeated truncation
failures (finish_reason=length).
Add a session hygiene check in _handle_message that runs right after
loading the transcript and before invoking the agent:
1. Estimate message count and rough token count of the transcript
2. If above configurable thresholds (default: 200 msgs or 100K tokens),
auto-compress the transcript proactively
3. Notify the user about the compression with before/after stats
4. If still above warn threshold (default: 200K tokens) after
compression, suggest /reset
5. If compression fails on a dangerously large session, warn the user
to use /compress or /reset manually
Thresholds are configurable via config.yaml:
session_hygiene:
auto_compress_tokens: 100000
auto_compress_messages: 200
warn_tokens: 200000
This complements the agent's existing preflight compression (which
runs inside run_conversation) by catching pathological sessions at
the gateway layer before the agent is even created.
Includes 12 tests for threshold detection and token estimation.
Kimi Code (platform.kimi.ai) issues API keys prefixed sk-kimi- that require:
1. A different base URL: api.kimi.com/coding/v1 (not api.moonshot.ai/v1)
2. A User-Agent header identifying a recognized coding agent
Without this fix, sk-kimi- keys fail with 401 (wrong endpoint) or 403
('only available for Coding Agents') errors.
Changes:
- Auto-detect sk-kimi- key prefix and route to api.kimi.com/coding/v1
- Send User-Agent: KimiCLI/1.0 header for Kimi Code endpoints
- Legacy Moonshot keys (api.moonshot.ai) continue to work unchanged
- KIMI_BASE_URL env var override still takes priority over auto-detection
- Updated .env.example with correct docs and all endpoint options
- Fixed doctor.py health check for Kimi Code keys
Reference: https://github.com/MoonshotAI/kimi-cli (platforms.py)
Previously, when a session expired (idle/daily reset), the memory flush
ran synchronously inside get_or_create_session — blocking the user's
message for 10-60s while an LLM call saved memories.
Now a background watcher task (_session_expiry_watcher) runs every 5 min,
detects expired sessions, and flushes memories proactively in a thread
pool. By the time the user sends their next message, memories are
already saved and the response is immediate.
Changes:
- Add _is_session_expired(entry) to SessionStore — works from entry
alone without needing a SessionSource
- Add _pre_flushed_sessions set to track already-flushed sessions
- Remove sync _on_auto_reset callback from get_or_create_session
- Refactor flush into _flush_memories_for_session (sync worker) +
_async_flush_memories (thread pool wrapper)
- Add _session_expiry_watcher background task, started in start()
- Simplify /reset command to use shared fire-and-forget flush
- Add 10 tests for expiry detection, callback removal, tracking
Reduces token usage and latency for most tasks by defaulting to
medium reasoning effort instead of xhigh. Users can still override
via config or CLI flag. Updates code, tests, example config, and docs.
_convert_to_png() renamed the original file to .bmp before calling
ImageMagick convert, then unconditionally deleted the .bmp regardless
of whether convert succeeded. If convert failed, both files were gone.
- Only delete .bmp after confirmed successful conversion
- Restore original file on convert failure, timeout, or missing binary
- Add 3 tests covering failure, not-installed, and timeout scenarios
_make_cli() did not clear HERMES_MAX_ITERATIONS env var, so tests
failed in CI where the var was set externally. Also, default max_turns
changed from 60 to 90 in 0a82396 but tests were not updated.
- Clear HERMES_MAX_ITERATIONS in _make_cli() for proper isolation
- Add env_overrides parameter for tests that need specific env values
- Update hardcoded 60 assertions to 90 to match new default
- Simplify test_env_var_max_turns using env_overrides
When MarkdownV2 parsing fails, _strip_mdv2() removes escape backslashes
and bold markers (*text*) but missed italic markers (_text_). Users saw
raw underscores around italic text in the plaintext fallback.
- Add regex to strip _text_ italic markers in _strip_mdv2()
- Use word boundary lookaround to preserve snake_case identifiers
- Add tests for _strip_mdv2 covering italic, bold, snake_case, and edge cases
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.
Introduces a new OpenClaw-to-Hermes migration skill with a Python
helper script that handles importing SOUL.md, memories, user profiles,
messaging settings, command allowlists, skills, TTS assets, and
workspace instructions.
Supports two migration presets (user-data / full), three skill conflict
modes (skip / overwrite / rename), overflow file export for entries that
exceed character limits, and granular include/exclude option filtering.
Includes detailed SKILL.md agent instructions covering the clarify-tool
interaction protocol, decision-to-command mapping, post-run reporting
rules, and path resolution guidance.
Adds dynamic panel width calculation to CLI clarify/approval widgets so
panels adapt to content and terminal size.
Includes 7 new tests covering presets, include/exclude, conflict modes,
overflow exports, and skills_guard integration.
Adds 4 new direct API-key providers (zai, kimi-coding, minimax, minimax-cn)
to the inference provider system. All use standard OpenAI-compatible
chat/completions endpoints with Bearer token auth.
Core changes:
- auth.py: Extended ProviderConfig with api_key_env_vars and base_url_env_var
fields. Added providers to PROVIDER_REGISTRY. Added provider aliases
(glm, z-ai, zhipu, kimi, moonshot). Added auto-detection of API-key
providers in resolve_provider(). Added resolve_api_key_provider_credentials()
and get_api_key_provider_status() helpers.
- runtime_provider.py: Added generic API-key provider branch in
resolve_runtime_provider() — any provider with auth_type='api_key'
is automatically handled.
- main.py: Added providers to hermes model menu with generic
_model_flow_api_key_provider() flow. Updated _has_any_provider_configured()
to check all provider env vars. Updated argparse --provider choices.
- setup.py: Added providers to setup wizard with API key prompts and
curated model lists.
- config.py: Added env vars (GLM_API_KEY, KIMI_API_KEY, MINIMAX_API_KEY,
etc.) to OPTIONAL_ENV_VARS.
- status.py: Added API key display and provider status section.
- doctor.py: Added connectivity checks for each provider endpoint.
- cli.py: Updated provider docstrings.
Docs: Updated README.md, .env.example, cli-config.yaml.example,
cli-commands.md, environment-variables.md, configuration.md.
Tests: 50 new tests covering registry, aliases, resolution, auto-detection,
credential resolution, and runtime provider dispatch.
Inspired by PR #33 (numman-ali) which proposed a provider registry approach.
Credit to tars90percent (PR #473) and manuelschipper (PR #420) for related
provider improvements merged earlier in this changeset.
Two bugs fixed:
1. search_messages() crashes with OperationalError when user queries
contain FTS5 special characters (+, ", (, {, dangling AND/OR, etc).
Added _sanitize_fts5_query() to strip dangerous operators and a
fallback try-except for edge cases.
2. _append_to_sqlite() in mirror.py creates a new SessionDB per call
but never closes it, leaking SQLite connections. Added finally block
to ensure db.close() is always called.
API key selection is now base_url-aware: when the resolved base_url
targets OpenRouter, OPENROUTER_API_KEY takes priority (preserving the
#289 fix). When hitting any other endpoint (Z.ai, vLLM, custom, etc.),
OPENAI_API_KEY takes priority so the OpenRouter key doesn't leak.
Applied in both the runtime provider resolver (the real code path) and
the CLI initial default (for consistency).
Fixes#560.
_make_cli() now patches CLI_CONFIG with clean defaults so
test_cli_init tests don't depend on the developer's local config.yaml.
test_empty_dir_returns_empty now mocks Path.home() so it doesn't pick
up a global SOUL.md.
Credit to teyrebaz33 for identifying and fixing these in PR #557.
Fixes#555.
tool_call_count was inaccurate in two ways:
1. Under-counting: an assistant message with N parallel tool calls
(e.g. "kill the light and shut off the fan" = 2 ha_call_service)
only incremented tool_call_count by 1 instead of N.
2. Over-counting: tool response messages (role=tool) also incremented
tool_call_count, double-counting every tool interaction.
Combined: 2 parallel tool calls produced tool_call_count=3 (1 from
assistant + 2 from tool responses) instead of the correct value of 2.
Fix: only count from assistant messages with tool_calls, incrementing
by len(tool_calls) to handle parallel calls correctly. Tool response
messages no longer affect tool_call_count.
This impacts /insights and /usage accuracy for sessions with tool use.
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.
Issues found and fixed during deep code path review:
1. CRITICAL: Prefix matching returned wrong prices for dated model names
- 'gpt-4o-mini-2024-07-18' matched gpt-4o ($2.50) instead of gpt-4o-mini ($0.15)
- Same for o3-mini→o3 (9x), gpt-4.1-mini→gpt-4.1 (5x), gpt-4.1-nano→gpt-4.1 (20x)
- Fix: use longest-match-wins strategy instead of first-match
- Removed dangerous key.startswith(bare) reverse matching
2. CRITICAL: Top Tools section was empty for CLI sessions
- run_agent.py doesn't set tool_name on tool response messages (pre-existing)
- Insights now also extracts tool names from tool_calls JSON on assistant
messages, which IS populated for all sessions
- Uses max() merge strategy to avoid double-counting between sources
3. SELECT * replaced with explicit column list
- Skips system_prompt and model_config blobs (can be thousands of chars)
- Reduces memory and I/O for large session counts
4. Sets in overview dict converted to sorted lists
- models_with_pricing / models_without_pricing were Python sets
- Sets aren't JSON-serializable — would crash json.dumps()
5. Negative duration guard
- end > start check prevents negative durations from clock drift
6. Model breakdown sort fallback
- When all tokens are 0, now sorts by session count instead of arbitrary order
7. Removed unused timedelta import
Added 6 new tests: dated model pricing (4), tool_calls JSON extraction,
JSON serialization safety. Total: 69 tests.
Custom OAI endpoints, self-hosted models, and local inference should NOT
show fabricated cost estimates. Changed default pricing from $3/$12 per
million tokens to $0/$0 for unrecognized models.
- Added _has_known_pricing() to distinguish commercial vs custom models
- Models with known pricing show $ amounts; unknown models show 'N/A'
- Overview shows asterisk + note when some models lack pricing data
- Gateway format adds '(excludes custom/self-hosted models)' note
- Added 7 new tests for custom model cost handling
Inspired by Claude Code's /insights, adapted for Hermes Agent's multi-platform
architecture. Analyzes session history from state.db to produce comprehensive
usage insights.
Features:
- Overview stats: sessions, messages, tokens, estimated cost, active time
- Model breakdown: per-model sessions, tokens, and cost estimation
- Platform breakdown: CLI vs Telegram vs Discord etc. (unique to Hermes)
- Tool usage ranking: most-used tools with percentages
- Activity patterns: day-of-week chart, peak hours, streaks
- Notable sessions: longest, most messages, most tokens, most tool calls
- Cost estimation: real pricing data for 25+ models (OpenAI, Anthropic,
DeepSeek, Google, Meta) with fuzzy model name matching
- Configurable time window: --days flag (default 30)
- Source filtering: --source flag to filter by platform
Three entry points:
- /insights slash command in CLI (supports --days and --source flags)
- /insights slash command in gateway (compact markdown format)
- hermes insights CLI subcommand (standalone)
Includes 56 tests covering pricing helpers, format helpers, empty DB,
populated DB with multi-platform data, filtering, formatting, and edge cases.
Authored by Farukest. Fixes#432. Extracts _kill_port_process() helper
that uses netstat+taskkill on Windows and fuser on Linux. Previously,
fuser calls were inline with bare except-pass, so on Windows orphaned
bridge processes were never cleaned up — causing 'address already in use'
errors on reconnect. Includes 5 tests covering both platforms, port
matching edge cases, and exception suppression.
Authored by Farukest. Fixes#435. The retry summary in
_handle_max_iterations() hardcoded max_tokens instead of using
_max_tokens_param(), which returns max_completion_tokens for direct
OpenAI API (required by gpt-4o, o-series). The first attempt already
used _max_tokens_param correctly — only the retry path was wrong.
Includes 4 tests for _max_tokens_param provider detection.
Verifies explicit allowlist keys, catch-all _API_KEY/_TOKEN patterns,
case insensitivity, TERMINAL_SSH prefix, and config.yaml routing for
non-secret keys. Covers the fix from PR #469.
The mock handler checked for function_name == 'search' but the RPC
sends 'search_files'. Any test exercising search_files through the
mock would get 'Unknown tool' instead of the canned response.
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.
The original implementation only supported xclip (X11), which silently
fails on WSL2 (can't access Windows clipboard for images), Wayland
desktops (xclip is X11-only), and VSCode terminal on WSL2.
Clipboard backend changes (hermes_cli/clipboard.py):
- WSL2: detect via /proc/version, use powershell.exe with .NET
System.Windows.Forms.Clipboard to extract images as base64 PNG
- Wayland: use wl-paste with MIME type detection, auto-convert BMP
to PNG for WSLg environments (via Pillow or ImageMagick)
- Dispatch order: WSL → Wayland → X11 (xclip), with fallthrough
- New has_clipboard_image() for lightweight clipboard checks
- Cache WSL detection result per-process
CLI changes (cli.py):
- /paste command: explicit clipboard image check for terminals where
BracketedPaste doesn't fire (image-only clipboard in VSCode/WinTerm)
- Ctrl+V keybinding: fallback for Linux terminals where Ctrl+V sends
raw byte instead of triggering bracketed paste
Tests: 80 tests (up from 37) covering WSL, Wayland, X11 dispatch,
BMP conversion, has_clipboard_image, and /paste command.
Copy an image to clipboard (screenshot, browser, etc.) and paste into
the Hermes CLI. The image is saved to ~/.hermes/images/, shown as a
badge above the input ([📎 Image #1]), and sent to the model as a
base64-encoded OpenAI vision multimodal content block.
Implementation:
- hermes_cli/clipboard.py: clean module with platform-specific extraction
- macOS: pngpaste (if installed) → osascript fallback (always available)
- Linux: xclip (apt install xclip)
- cli.py: BracketedPaste key handler checks clipboard on every paste,
image bar widget shows attached images, chat() converts to multimodal
content format, Ctrl+C clears attachments
Inspired by @m0at's fork (https://github.com/m0at/hermes-agent) which
implemented image paste support for local vision models. Reimplemented
cleanly as a separate module with tests.
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).
Replaces the unsafe 128K fallback for unknown models with a descending
probe strategy (2M → 1M → 512K → 200K → 128K → 64K → 32K). When a
context-length error occurs, the agent steps down tiers and retries.
The discovered limit is cached per model+provider combo in
~/.hermes/context_length_cache.yaml so subsequent sessions skip probing.
Also parses API error messages to extract the actual context limit
(e.g. 'maximum context length is 32768 tokens') for instant resolution.
The CLI banner now displays the context window size next to the model
name (e.g. 'claude-opus-4 · 200K context · Nous Research').
Changes:
- agent/model_metadata.py: CONTEXT_PROBE_TIERS, persistent cache
(save/load/get), parse_context_limit_from_error(), get_next_probe_tier()
- agent/context_compressor.py: accepts base_url, passes to metadata
- run_agent.py: step-down logic in context error handler, caches on success
- cli.py + hermes_cli/banner.py: context length in welcome banner
- tests: 22 new tests for probing, parsing, and caching
Addresses #132. PR #319's approach (8K default) rejected — too conservative.
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
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>
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.
The retry summary in _handle_max_iterations hardcodes max_tokens instead
of calling _max_tokens_param(). For direct OpenAI API users (gpt-4o,
o-series), the correct parameter name is max_completion_tokens. The first
attempt at line 2697 already uses _max_tokens_param correctly but the
retry path at line 2743 was missed.
fuser command does not exist on Windows, causing orphaned bridge processes
to never be cleaned up. On crash recovery, the port stays occupied and the
next connect() fails with address-already-in-use.
Add _kill_port_process() helper that uses netstat+taskkill on Windows and
fuser on Linux/macOS. Replace both call sites in connect() and disconnect().
Adds a /update command to Telegram, Discord, and other gateway platforms
that runs `hermes update` to pull the latest code, update dependencies,
sync skills, and restart the gateway.
Implementation:
- Spawns `hermes update` in a separate systemd scope (systemd-run --user
--scope) so the process survives the gateway restart that hermes update
triggers at the end. Falls back to nohup if systemd-run is unavailable.
- Writes a marker file (.update_pending.json) with the originating
platform and chat_id before spawning the update.
- On gateway startup, _send_update_notification() checks for the marker,
reads the captured update output, sends the results back to the user,
and cleans up.
Also:
- Registers /update as a Discord slash command
- Updates README.md, docs/messaging.md, docs/slash-commands.md
- Adds 18 tests covering handler, notification, and edge cases
Authored by FarukEst. Fixes#392.
1. Initialize data={} before health-check loop to prevent NameError when
resp.json() raises after http_ready is set to True.
2. Extract _close_bridge_log() helper and call on all return False paths
to prevent file descriptor leaks on failed connection attempts.
Refactors disconnect() to reuse the same helper.
Authored by 0xbyt4.
The italic regex [^*]+ matched across newlines, corrupting bullet lists
using * markers (e.g. '* Item one\n* Item two' became italic garbage).
Fixed by adding \n to the negated character class: [^*\n]+.
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 0xbyt4.
Two fixes:
- extract_images(): only remove extracted image tags, not all markdown image
tags. Previously  was silently dropped when real images
were also present.
- truncate_message(): walk chunk_body not full_chunk when tracking code block
state, so the reopened fence prefix doesn't toggle in_code off and leave
continuation chunks with unclosed code blocks.
Authored by 0xbyt4.
144 new tests covering gateway/pairing.py, tools/skill_manager_tool.py,
tools/skills_tool.py, honcho_integration/session.py, and
agent/auxiliary_client.py.
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.
Fixes a bug where the refresh token was not persisted when the API key
mint failed (e.g., 402 insufficient credits, timeout). The rotated
refresh token was lost, causing subsequent auth attempts to fail with
a stale token.
Changes:
- Persist auth state immediately after each successful token refresh,
before attempting the mint
- Use latest in-memory refresh token on mint-retry paths (was using
the stale original)
- Atomic durable writes for auth.json (temp file + fsync + replace)
- Opt-in OAuth trace logging (HERMES_OAUTH_TRACE=1, fingerprint-only)
- 3 regression tests covering refresh+402, refresh+timeout, and
invalid-token retry behavior
Author: Robin Fernandes <rewbs>
Authored by PercyDikec. Fixes#394.
The transcript extraction used len(history) to find new messages, but
history includes session_meta entries stripped before reaching the agent.
This caused 1 message lost per turn from turn 2 onwards. Fix returns
history_offset (filtered length) from _run_agent and uses it for the slice.
Two fixes for the case where a user switches to a model with a smaller
context window while having a large existing session:
1. Preflight compression in run_conversation(): Before the main loop,
estimate tokens of loaded history + system prompt. If it exceeds the
model's compression threshold (85% of context), compress proactively
with up to 3 passes. This naturally handles model switches because
the gateway creates a fresh AIAgent per message with the current
model's context length.
2. Error handler reordering: Context-length errors (400 with 'maximum
context length' etc.) are now checked BEFORE the generic 4xx handler.
Previously, OpenRouter's 400-status context-length errors were caught
as non-retryable client errors and aborted immediately, never reaching
the compression+retry logic.
Reported by Sonicrida on Discord: 840-message session (2MB+) crashed
after switching from a large-context model to minimax via OpenRouter.
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 transcript extraction used len(history) to find new messages, but
history includes session_meta entries that are stripped before passing
to the agent. This mismatch caused 1 message to be lost from the
transcript on every turn after the first, because the slice offset
was too high. Use the filtered history length (history_offset) returned
by _run_agent instead.
Also changed the else branch from returning all agent_messages to
returning an empty list, so compressed/shorter agent output does not
duplicate the entire history into the transcript.
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.
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.
The session key construction logic was duplicated in 4 places
(session.py + 3 inline copies in run.py), which is exactly the
kind of drift that caused issue #349 in the first place.
Extracted build_session_key() as a public function in session.py.
SessionStore._generate_session_key() now delegates to it, and all
inline key construction in run.py has been replaced with calls to
the shared function. Tests updated to test the function directly.
The previous implementation used `len(self._entries) > 1` to check if any
sessions had ever been created. This failed for single-platform users because
when sessions reset (via /reset, auto-reset, or gateway restart), the entry
for the same session_key is replaced in _entries, not added. So len(_entries)
stays at 1 for users who only use one platform.
Fix: Query the SQLite database's session count instead. The database preserves
historical session records (marked as ended), so session_count() correctly
returns > 1 for returning users even after resets.
This prevents the agent from reintroducing itself to returning users after
every session reset.
Fixes#351
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.
Three section header comments in tests/test_run_agent.py used
'Grup' instead of 'Group':
- Line 124: # Grup 1: Pure Functions
- Line 276: # Grup 2: State / Structure Methods
- Line 572: # Grup 3: Conversation Loop Pieces (OpenAI mock)
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.
When both OPENROUTER_API_KEY and OPENAI_API_KEY are set (e.g. OPENAI_API_KEY
in .bashrc), the wrong key was sent to OpenRouter causing auth failures.
Fixed key resolution order in cli.py and runtime_provider.py.
Fixes#289
- 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
Patch _servers to empty dict in tests that call discover_mcp_tools()
with mocked config, preventing interference from real MCP connections
that may exist when running within the full test suite.
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"]
Added a fixture to redirect HERMES_HOME to a temporary directory during tests, preventing writes to the user's home directory. Updated the test for DebugSession to create a dedicated log directory for saving logs, ensuring test isolation and accuracy in assertions.
The TestFlushSentinelNotLeaked test from PR #227 had two issues:
1. flush_memories() uses get_text_auxiliary_client() which could bypass
agent.client entirely — mock it to return (None, None)
2. No assertion that the API was actually called — added guard assert
Without these fixes the test passed vacuously (API never called).
The TestRetryExhaustion tests from PR #223 didn't mock time.sleep/time.time,
causing the retry backoff loops (275s+ total) to run in real time. Tests would
time out instead of running quickly.
Added _make_fast_time_mock() helper that creates a mock time module where
time.time() advances 500s per call (so sleep_end is always in the past) and
time.sleep() is a no-op. Both tests now complete in <1s.
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 OpenAI API returns content: null on assistant messages that only
contain tool calls. msg.get('content', '') returns None (not '') when
the key exists with value None, causing TypeError on len() and string
concatenation in _generate_summary and compress.
Fix: msg.get('content') or '' — handles both missing keys and None.
Tests from PR #216 (@Farukest). Fix also in PR #215 (@cutepawss).
Both PRs had stale branches and couldn't be merged directly.
Closes#211
Priority is: CLI arg > config file > env var > default
(not env var > config file as the old comment stated)
The test failed because config.yaml had max_turns at both root level
and inside agent section. The test cleared agent.max_turns but the
root-level value still took precedence over the env var. Fixed the
test to clear both, and corrected the comment to match the intended
priority order.
Tests added:
- Roundtrip serialization of chat_topic via to_dict/from_dict
- chat_topic defaults to None when missing from dict
- Channel Topic line appears in session context prompt when set
- Channel Topic line is omitted when chat_topic is None
Follow-up to PR #248 (feat: Discord channel topic in session context).
Enhanced the AIAgent class to capture and normalize summary information for reasoning items. Implemented logic to handle summaries as lists, ensuring proper formatting for API interactions. Updated tests to validate the inclusion of summaries in reasoning items, both for existing and default cases.
Updated the authentication mechanism to store Codex OAuth tokens in the Hermes auth store located at ~/.hermes/auth.json instead of the previous ~/.codex/auth.json. This change includes refactoring related functions for reading and saving tokens, ensuring better management of authentication states and preventing conflicts between different applications. Adjusted tests to reflect the new storage structure and improved error handling for missing or malformed tokens.
Introduced a new `provider_routing` section in the CLI configuration to control how requests are routed across providers when using OpenRouter. This includes options for sorting providers by throughput, latency, or price, as well as allowing or ignoring specific providers, setting the order of provider attempts, and managing data collection policies. Updated relevant classes and documentation to support these features, enhancing flexibility in provider selection.
Remove loop_scope="function" parameter from async test decorators in
test_hooks.py. This matches the existing convention in the repo
(test_telegram_documents.py) and avoids requiring pytest-asyncio 0.23+.
All 144 new tests from PR #191 now pass.
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.
Fixes#241
When users set HONCHO_API_KEY via `hermes config set` or environment
variable, they expect the integration to activate. Previously, the
`enabled` flag defaulted to `false` when reading from global config,
requiring users to also explicitly enable Honcho.
This change auto-enables Honcho when:
- An API key is present (from config file or env var)
- AND `enabled` is not explicitly set to `false` in the config
Users who want to disable Honcho while keeping the API key can still
set `enabled: false` in their config.
Also adds unit tests for the auto-enable behavior.
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.
print_above() used \033[K (erase-to-end-of-line) to clear the spinner
line before printing text above it. This causes garbled escape codes when
prompt_toolkit's patch_stdout is active in CLI mode.
Switched to the same spaces-based clearing approach used by stop() —
overwrite with blanks, then carriage return back to start of line.
Updated test assertion to match the new clearing method.
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.
