On vulnerable SQLite (e.g. 3.50.4), do not enable WAL for fresh/non-WAL
shared databases — prefer DELETE instead. Leave existing on-disk WAL
alone (no live downgrade under concurrent gateway/cron openers). Surface
Python/SQLite version details as a doctor warning (#69784).
Client leg of M2 org-shared skills (hsp-1-contract.md §11), pairing with
gateway-gateway #162 and NAS #768 (both merged).
- resolve_org_identity(): org_id + org_role from the token claims. NO
org_role claim (personal org — NAS only stamps it for multi-member orgs)
=> SyncInertError => every org surface is inert; personal M1 sync
untouched (contract §11.1 REFINED). org_sync_available() for callers.
- pull_org_skills(): materialize the org canonical set (refs/org/<org_id>/
HEAD) into ~/.hermes/skills/_org/<org_id>/ — fast-forward only, no client
merge on the org path (design.md §2.6); read-only mirror by convention
(§7.1: a local edit is a personal fork until proposed). maybe_pull_org_
skills() best-effort hook (never raises; inert without the claim).
- propose_skill(): snapshot the LOCAL skill dir, splice it into the org
HEAD's skill-tree map (per-skill delta, never wholesale replace), upload
?scope=org, CAS the org HEAD. ADMIN => direct merge; MEMBER => server
converts to a proposal — cas_ref now surfaces 202 as
{proposal_pending: True, proposal_id, ref} (success-shaped, NEVER
presented as live). Non-interactive by design for the future automated
submitter (Ben's trajectory note).
- put_objects(org_scope=True) adds ?scope=org (contract §11.5).
- is_sync_eligible(): skills under _org/ are excluded from PERSONAL sync —
enterprise content never rides a personal push (§11.11).
- CLI: hermes skills propose <name> [-m msg] — prints 'pending admin
review' for 202, 'merged' for admin, and a plain 'org sync unavailable'
for personal orgs instead of a raw 403.
Tests: 56 in the sync client suite (+10 org: identity gate both ways, _org/
personal-sync exclusion, admin direct merge, member 202 w/ HEAD untouched +
proposal ref parked + never-merged, splice-not-replace root, pull mirror
materialization, no-head noop, org-feature gate, maybe_pull inert). Mock
server extended (org feature flag, member-CAS→202). 103 across skills
suites. Live CLI smoke: propose --help + personal-org inert path verified.
Guard the post-start set_agent_cursor_enabled on _session._started so
call_tool cannot re-enter session.start() (matches the start_session
lifecycle guard).
Cut steady-state Computer Use latency without changing default behavior
or waiting on cua-driver:
- Cap screenshots via set_config(max_image_dimension) on session start
(config: computer_use.max_image_dimension, default 1456)
- Cache aux-vision routing per (provider, model) so captures skip
repeated load_config()
- Add computer_use.capture_after_mode (default som) so users can opt
follow-ups down to ax (elements only) for speed
Auto-detect now disables the cursor overlay on darwin as well as
headless/WSL2 Linux. After start_session, also call
set_agent_cursor_enabled(false) when the policy is on so older drivers
without --no-overlay still tear the overlay down.
Co-authored-by: David Metcalfe <80915+DavidMetcalfe@users.noreply.github.com>
The hermes-sweeper review #4701565902 (2026-07-15) flagged two
consistency issues in `_cua_driver_supports_no_overlay` and one
additive-config concern:
1. `cua-backend.py:260` — the `cua-driver --help` support probe
inherited the full parent environment. cua-driver is a third-party
binary; every other spawn site in this file (manifest probe at
`:214`, MCP spawn at `:697`, install probe at `:997`) uses
`_sanitize_subprocess_env(cua_driver_child_env())`. The `--help`
probe should match. This was a low-impact leak (only help output
exits), but inconsistency is the wrong default for a third-party
subprocess.
2. `cua_backend.py:238` — when the manifest returned a `command`
different from the input `driver_cmd` parameter (e.g. a relocated
executable at `/opt/relocated/cua-driver` while the system binary
is at `/usr/bin/cua-driver`), the support probe ran against
`_CUA_DRIVER_CMD` (the default) instead of the manifest-discovered
`command`. Two failure modes:
- The wrapper binary supports `--no-overlay` but the system binary
doesn't → probe returns False → overlay kept despite capability.
- The system binary supports `--no-overlay` but the wrapper doesn't
→ probe returns True → MCP spawn crashes on the unknown flag.
3. The original commit bumped `_config_version` 31→32 for an additive
default (`computer_use.no_overlay: None`). AGENTS.md specifies that
additive defaults in existing sections are handled by deep merge
and should NOT trigger a version bump. After cherry-picking onto
current `origin/main` (which is already at 33), the bump is
effectively dropped — resolved to main's 33.
Changes:
- Add `env=_sanitize_subprocess_env(cua_driver_child_env())` to the
`--help` subprocess (with the same import + rationale comment as
the manifest probe).
- Pass `driver_cmd=command` (or `driver_cmd=driver_cmd` for the
fallback path) into `_mcp_args_with_overlay_flag`, so the support
probe runs against the binary that will actually be launched.
Tests (3 new):
- `test_help_probe_passes_sanitized_env` — verifies `subprocess.run`
is called with an `env=` kwarg.
- `test_manifest_command_drives_support_probe` — verifies the probe
runs against the manifest command when it differs from the input
driver_cmd.
- `test_fallback_uses_input_driver_cmd_for_support_probe` — verifies
the fallback path (no command in manifest) uses the input
driver_cmd.
- `test_probe_distinguishes_support_between_binaries` — sanity check
that the lru_cache key on `driver_cmd` prevents cross-binary
cache leakage.
File-revert negative test confirmed all three of the new
"manifest/probe" tests are load-bearing: with the pre-fix code, they
fail (probe runs against the default binary instead of the resolved
one); with the fix, they pass. 20/20 tests in
`tests/computer_use/test_cua_no_overlay.py` green.
`TestMcpInvocationResolution` (8/8) still green.
Refs: sweeper review #4701565902
Address review feedback from cross-vendor review (Flash + GPT-OSS):
1. Auto-detect now checks for headless Linux (no DISPLAY), WSL2
(/proc/version contains 'microsoft'), instead of all Linux.
Desktop Linux with a compositor keeps the overlay.
2. Add _cua_driver_supports_no_overlay() that probes cua-driver --help
to check if the flag is supported. Older drivers (< 0.6.x) reject
unknown flags, so passing --no-overlay would crash the MCP spawn.
3. Update tests to cover headless vs desktop Linux, WSL2 detection,
version probe, and the unsupported-driver fallback path.
cua-driver's cursor overlay rendering loop can consume CPU indefinitely
when idle (#28152, #47032). On Linux/WSL2, the overlay serves no visual
purpose and the rendering path is the primary source of idle CPU usage.
Add computer_use.no_overlay config option (default: auto-detect) that
passes --no-overlay to cua-driver when enabled. Auto-detection disables
the overlay on Linux (covers WSL2, headless, containers) where it has no
benefit, and keeps it enabled on macOS/Windows where it is visually
useful.
Refs: #28152, #47032
Adds the missing write path for the per-task model_override column (which
was previously only settable via manual SQL) and pairs it with a
provider_override so cross-provider switches resolve correctly:
- kanban_db: provider_override column (+migration), set_model_override()
with model_override_set event, create_task(model_override=,
provider_override=), dispatcher spawns worker with -m <model>
[--provider <name>]
- dashboard: Model row in the task drawer — dropdown fed by a new
/model-options endpoint (build_models_payload substrate, provider-grouped,
free-text fallback), PATCH + bulk model override support
- CLI: kanban create --model/--provider, new kanban set-model subcommand,
show prints the provider
- agent tools: kanban_create accepts model/provider; show/list expose
provider_override
Rate-limit recovery flow: override is settable on running tasks and takes
effect on the next dispatch, without touching the worker profile's config.
Slack could already deliver files in-channel via the gateway, but
send_message omitted MEDIA for Slack and told the model it was
unsupported — causing agents to inconsistently refuse PDF sends.
Wire Slack through files_upload_v2 in the standalone sender.
* test(clarify-gateway): cover signature, timeout fallback, and notify paths for 100% coverage
Fixes#36531
(cherry picked from commit 5265dfe2f5)
* fix(clarify): one canonical timeout across CLI, TUI/desktop, and gateway
The clarify wait timeout was resolved three different (wrong) ways:
- CLI (`cli.py`, `hermes_cli/callbacks.py`) read a non-existent top-level
`clarify.timeout`, so it always fell through to a hardcoded 120s instead of
the canonical `agent.clarify_timeout` (default 3600) the gateway uses (#42969).
- The TUI/desktop bridge called `_block("clarify.request", …)` with no timeout,
so it used the hardcoded 300s `_block` default and ignored config (#51960).
- There was no way to disable the auto-skip: a user who wanted the agent to wait
indefinitely while they think couldn't get it.
Collapse all of this onto a single resolver:
- `tools.clarify_gateway.resolve_clarify_timeout(config)` is the one source of
truth. Order: explicit legacy `clarify.timeout` (back-compat) → canonical
`agent.clarify_timeout` → 3600. `<= 0` is preserved verbatim as "unlimited".
- CLI, callbacks, and the TUI bridge (`_clarify_timeout_seconds`) all route
through it, so the three surfaces can't drift.
- `<= 0` means unlimited everywhere: `wait_for_response` and `_block` drop the
deadline (heartbeat still fires), and the CLI hides its countdown.
Tests: resolver order / default / non-numeric / unlimited-sentinel; an
unlimited `wait_for_response` blocks until resolved rather than auto-skipping;
the TUI clarify bridge passes the configured timeout to `_block`.
Supersedes #42974 (CLI key), #51993 (TUI honors config), and #68986 (unlimited
wait); folds in #52031 (clarify_gateway coverage).
Co-authored-by: liuhao1024 <liuhao1024@users.noreply.github.com>
Co-authored-by: lkevincc0 <lkevincc0@users.noreply.github.com>
Co-authored-by: theone139344 <theone139344@users.noreply.github.com>
Co-authored-by: baauzi <baauzi@users.noreply.github.com>
---------
Co-authored-by: Christopher-Schulze <210261288+Christopher-Schulze@users.noreply.github.com>
Co-authored-by: liuhao1024 <liuhao1024@users.noreply.github.com>
Co-authored-by: lkevincc0 <lkevincc0@users.noreply.github.com>
Co-authored-by: theone139344 <theone139344@users.noreply.github.com>
Co-authored-by: baauzi <baauzi@users.noreply.github.com>
Limit the _NET_ACTIVE_WINDOW xprop fallback to unqualified default
captures so exact pid/window_id targeting does not pay up to a 2s
subprocess probe on Linux/X11.
When Linux/X11 reports the same z_index for every on-screen window,
prefer _NET_ACTIVE_WINDOW via xprop instead of list order. Keep the
higher-z-index-is-frontmost contract when ordering is informative.
Commit author.device was an opaque uuid4 hex, so the sync console showed a hash
per device. Make it human-friendly:
- Default: seed new devices from the short hostname + a short random suffix
(e.g. bens-macbook-a1b2c3) instead of a bare uuid. Existing .sync_device_id
files are honored verbatim (a machine keeps its id) — backward-compatible.
- hermes sync device [--name N]: show or set an explicit label
(set_device_name(), written to ~/.hermes/skills/.sync_device_id). New commits
use it; past commits keep their old label (author.device is immutable).
- Hermes Cloud: HERMES_SYNC_DEVICE_NAME env seeds the first-use label so a
hosted instance shows a recognizable name with no CLI call. Precedence:
explicit .sync_device_id file > HERMES_SYNC_DEVICE_NAME env > hostname default.
Env seeds first-use only, then persists, so a later --name still wins locally.
Tests: 46 pass (+5: hostname default, file-wins precedence, env first-use seed +
persistence, set/trim, empty-name reject). CLI verb smoke-verified end-to-end.
Addresses @teknium1's second review round on #63144:
1. _is_compacted_message now checks both active=0 AND compacted=1.
Previously checked active=0 alone, which also matched rewind/undo rows
(active=0, compacted=0) that must stay hidden.
2. Added _is_compression_ended() — checks only the session's own
end_reason, not the lineage-wide has_compression_hop flag. This prevents
delegation children living under a compression continuation from leaking
through the lineage filter.
3. _discover lineage skip now uses is_ended_session (session-level check)
instead of has_compression (lineage-level flag).
New tests:
- TestRewindExclusion: rewind rows stay hidden alongside compacted rows
- TestCompressionEndedHelper: session-level end_reason checks
- TestLegacyContinuationPlusDelegation: delegate child excluded while
compression ancestors surface
78 passed (71 + 7 new).
After context compaction, pre-compaction content was invisible to
session_search — a memory black hole. The _discover() skip logic
filtered both same-session and same-lineage hits unconditionally,
without distinguishing compression-summarised content (gone from live
context) from delegation children (still visible to the parent agent).
Reworked _resolve_to_parent to return (root_id, has_compression_hop),
checking end_reason='compression' on every hop during the same
db.get_session() traversal — zero extra queries.
_discover() now has three compression-aware paths:
- In-place compaction: FTS hits on active=0 (compacted=1) rows pass
through even when raw_sid == current_session_id
- Legacy rotation: lineage hits pass through when has_compression_hop
is true on either side of the chain
- Delegation children: still excluded (no compression edge)
18 new tests covering all three scenarios + unit tests for the helpers.
Addresses Teknium's review feedback on #6256.
Closes#13840, #13841.
mark_speech_interrupted() / take_speech_interrupted(): a one-shot,
TTL'd (120s) latch plus SPEECH_INTERRUPTED_NOTE. Barge-in paths mark
it when they cut live speech; the next turn's submit path pops it and
prepends the note to the model-bound message — API-call local, never
persisted, so history and prompt caching are untouched.
listen_for_speech(): sustained-RMS speech detection with the noise
floor calibrated against audible playback (speaker bleed doesn't
self-trigger). With capture=True it keeps a rolling pre-roll buffer and
records through to a silence endpoint, so the interruption is
transcribed from its first syllable — detection alone loses the opening
words to the detector's sustain window plus the mic re-open.
transcribe_recording() maps no_speech provider results to a successful
empty transcript (silence, not an error).
An STT provider that hears no words is reporting silence, not failing.
ElevenLabs/xAI empty-transcript errors now carry no_speech so live
voice loops can re-listen quietly instead of surfacing an error on
every pause.
stream_tts_to_speaker() drops its hardcoded ElevenLabs client for
resolve_streaming_provider() + SentenceChunker, so any provider speaks
sentence-by-sentence while the model is still generating. Markdown
stripping also drops emoji — providers stall on them or read them out
loud.
StreamingTTSProvider ABC + registry + resolve_streaming_provider() —
ElevenLabs (pcm_24000) and OpenAI (response_format=pcm) stream chunked
PCM; every other provider keeps its configured voice and falls back to
per-sentence sync synthesis. SentenceChunker is the one incremental
cutter every surface shares: sentence-boundary cuts on the delta
stream, <think> blocks stripped even when split across deltas, short
fragments merged forward so they never stall as tiny clips.
Two changes:
1) Rename the client-local head-bookkeeping file .sync_manifest -> .sync_state
(read_sync_state/write_sync_state) to remove the name collision with the
§2.8 plane 'sync-manifest' OBJECT. read_sync_state migrates an existing
.sync_manifest on first read so no device loses its head record.
2) Make the knobs a Hermes Cloud instance needs env-configurable, so an
instance can be set up to use sync BY DEFAULT with no config.yaml edit or
per-skill CLI call. Precedence: HERMES_SYNC_* env -> config.yaml sync.* ->
built-in default (mirrors the existing HERMES_SYNC_BASE_URL bridge).
- HERMES_SYNC_ENABLED -> sync.enabled (master on/off; def off)
gated in maybe_push/maybe_pull alongside the dev-gate + base_url.
- HERMES_SYNC_DEFAULT_OPT_IN -> sync.default_opt_in (M1-D policy; def off)
opt-out mode: every eligible skill syncs unless usage rec says sync:false;
'your skills follow you with no setup' — the Cloud default. opt-in mode
(default) unchanged: only sync:true skills sync.
sync_status() + 'hermes sync status' surface feature_enabled/default_opt_in.
Cross-repo byte-compat re-verified: client manifest bytes still parse cleanly
through gateway-gateway parseSyncManifest.
Tests: 41 pass (+7: env precedence, opt-out/opt-in policy, rename migration).
Replace the device-local opt-in model (`.usage.json` `sync` flag as the sole
source of truth) with the §2.8 content model: a root-level `sync-manifest` blob
in the tree at `refs/user/<owner>/HEAD` recording per-skill {name, enabled},
matching gateway-gateway src/sync/manifest.ts byte-for-byte.
- build/parse_sync_manifest: canonical {type,version:1,skills:[{name,enabled}]},
strict parse (malformed != empty).
- snapshot_profile embeds the manifest as a root-level blob alongside skill
subtrees; the skill walk skips it (blob, not a SKILL.md-bearing tree).
- pull reconciles local opt-in intent FROM the plane manifest, so a skill opted
in on one device becomes opted in on the others (opt-in is now cross-device,
not per-device). Never silently disables a locally-enabled skill on pull.
- .usage.json `sync` flag kept as the editable local intent; plane manifest is
authoritative.
Cross-repo byte-compat verified: the Python client's manifest bytes parse
cleanly through gateway-gateway's real parseSyncManifest (tsx harness).
Tests: 34 pass incl. 5 new (roundtrip, wire shape, strict-reject, root-blob
embed, pull adopts opt-in from manifest).
Removes Homebrew and PyPI wheel/sdist as Hermes distribution paths while
preserving the supported source, Docker, and Nix workflows.
Changes:
- Removes the Homebrew formula, PyPI publish workflow, sdist manifest
(MANIFEST.in), and wheel/sdist release-attachment logic from scripts/release.py.
- Keeps setuptools metadata and entry points required by editable installs
and Docker/Nix builds, but adds a setup.py guard that rejects wheel/sdist
builds outside a sealed Nix derivation (HERMES_NIX_BUILD=1).
- Removes pip/Homebrew install detection, PyPI update checks, the pip
self-update path, the deprecation-banner state, the postinstall subcommand,
wheel data-directory fallbacks in agent/i18n.py and hermes_constants.py,
and the ACP Registry manifest/version-lockstep release logic.
- Adds /nix/store/ path detection so `nix run` / `nix profile install`
installs (which don't set HERMES_MANAGED) are correctly identified as
"nix" rather than falling through to "git"/"unknown".
- Retired install-method values ("pip", "homebrew") in existing
.install_method stamps (both code-scoped and home-scoped) are ignored by
the allowlist reader and fall through to "unknown" instead of resurrecting
a retired enum value.
- Updates Nix packaging to ship bare runtime data (locales, optional-mcps)
through store symlinks and wrapper env vars instead of wheel data-files.
- Removes the ACP Registry manifest/icon and their version-lockstep tests.
- Deletes or rewrites packaging, pip-update, Homebrew, and ACP Registry
tests; adds parametrized coverage for the packaging build guard covering
BOTH sdist and wheel paths (the guards live in separate cmdclass entries
— a passing sdist test proves nothing about the wheel path).
- Updates installation/platform documentation and related user-facing copy.
- Adjusts the supply-chain scan so deleted install-hook files do not trigger
a finding, while additions or modifications still require the existing
ci-reviewed label gate.
Supported installation paths (unchanged):
- git installer (install.sh)
- Docker
- Nix/NixOS
- editable development installs (uv sync, uv pip install -e ., pip install -e .)
Extract the open_preview emitter into a shared tools/desktop_ui bridge
(one gateway-injected sink, routed by HERMES_UI_SESSION_ID) and add a
second desktop-gated tool on top of it:
- focus_pane(chat|files|terminal|review|sessions) -> pane.reveal event.
The desktop runs each pane's own reveal path (revealDesktopPane table)
and only acts on the active window -- a background turn never moves the
user's focus (desktop AGENTS.md: offer, don't hijack).
open_preview now emits through the same bridge. Both tools are check_fn
on HERMES_DESKTOP (zero footprint elsewhere), sitting beside
read_terminal/close_terminal in _HERMES_CORE_TOOLS.
Deliberately not adding run_slash: letting the agent fire slash commands
mid-turn (/model, /new, /clear) fights prompt-cache + conversation
invariants.
Add a desktop-gated open_preview tool so 'open cnn.com in the preview
pane' works. The tool (check_fn on HERMES_DESKTOP, zero footprint
elsewhere) emits a preview.open event through a gateway-injected emitter,
mirroring the close_terminal -> terminal.close bridge. The desktop
handles it in usePreviewRouting, normalizing the target and opening the
pane for the active session only -- a background turn never hijacks it.
Bare domains and localhost are coaxed into fetchable URLs (www.cnn.com ->
https://, localhost:3000 -> http://); file paths and schemes pass through
to the renderer's normalizer.
In Slack threads, ordinary follow-up messages sent while a native
multi-choice clarify was pending were consumed as clarify answers —
_coerce_text_response accepted arbitrary text for any pending entry, so
the gateway text-intercept swallowed unrelated thread messages and the
user's messages appeared to be ignored.
Tighten resolve_text_response_for_session for native interactive
multi-choice prompts (buttons rendered, awaiting_text=False):
- numeric selections ("2") still resolve to the canonical choice
- exact choice-label matches (case-insensitive) still resolve
- arbitrary prose is now REJECTED (returns False) so the message
continues as a normal turn instead of vanishing into the clarify
Behavior is preserved everywhere free text is legitimately the answer:
open-ended clarifies, explicit 'Other' text-capture mode, and the base
adapter's numbered-text fallback (which flips awaiting_text at send
time).
Salvaged from PR #62042 by @liuhao1024.
Fixes#62034
Context-compaction handoff summaries (prefixed with [CONTEXT COMPACTION])
were being returned as normal bookend_start/bookend_end messages in
session_search discovery mode. A single compaction handoff could be 57K+
chars, immediately bloating a fresh session prompt to 73K+ chars from one
search hit.
Changes:
- Add _COMPACTION_PREFIXES and _is_compaction_summary() helper
- Filter compaction summaries from bookend_start and bookend_end in _discover()
- Cap bookend content to 1200 chars and window content to 4000 chars
- Add content_truncated/original_content_chars metadata when truncation occurs
- Add 6 regression tests covering prefix detection, bookend filtering,
content capping, and legacy [CONTEXT SUMMARY] prefix
Fixes#43175
Children inherit the parent's env, repo, and toolsets but were denied
execute_code ('children should reason step-by-step, not write scripts').
That forces subagents doing mechanical multi-step work (batch file
reads, fetch-N-pages loops, filter-before-context reductions) to burn
reasoning iterations one tool call at a time.
- Remove execute_code from DELEGATE_BLOCKED_TOOLS
- Stop stripping the code_execution toolset from child bundles
- No recursion risk: the sandbox bridges only the 7 SANDBOX_ALLOWED_TOOLS
(web/file/terminal) — delegate_task and execute_code itself are not
reachable from inside a sandbox script
- Update schema text, AGENTS.md, and delegation docs
- Tests: blocked-constant, strip, and child-assembly tests updated;
new test pins execute_code as intentionally unblocked
Adapter acceptance is not proof of delivery: the inner #55578 resolver can
still fail closed inside the message pipeline after the adapter accepted the
synthetic event, which falsely acknowledged the durable row as delivered and
silently discarded the delegation result.
Pre-flight the delivery target in _deliver_completion_notification before
adapter acceptance (adapted from #65838 by @henrynguyeninfo1):
- live parent or verified live compression tip -> deliver (inner resolver
still owns the actual route retarget)
- explicit-reset / unknown parent -> terminal 'dropped' disposition via new
drop_completion_delivery() (not falsely 'delivered', not eternally
'pending')
- transient uncertainty (DB error, mid-flight rotation without a visible
continuation) -> release the claim for retry
release_completion_delivery() now converges to a terminal 'dropped' state
once _MAX_DELIVERY_ATTEMPTS is exhausted, so an undeliverable completion
cannot replay on every gateway restart forever.
* fix(desktop): refresh repo status on session switch with unchanged cwd (#68208)
fix(desktop): refresh repo status on session switch with unchanged cwd
* fix(checkpoints): honor gateway config and task cwd (#68195)
* fix(gateway): wire checkpoint config into agents
* fix(checkpoints): resolve gateway file paths by task cwd
* ci: live-updating PR review comment with structured job statuses
Replace the static comment-pending + comment-results two-job pattern
with a live-updating comment system that polls the GitHub Actions API
every 15s, re-assembles the review comment from whatever results are
available, and upserts it via the <!-- hermes-ci-review-bot --> marker.
The comment updates in real time as each job finishes — no waiting for
the full pipeline.
Every CI job that wants to appear in the review comment emits a
review_status output — a JSON array of objects, each with a source
and a results array:
[
{
"source": "review-label-gate",
"results": [
{"kind": "action_required", "title": "...", "summary": "...",
"how_to_fix": "..."},
{"kind": "info", "title": "...", "summary": "..."}
]
},
{
"source": "ci timing",
"results": [
{"kind": "warning", "title": "CI timings", "summary": "...",
"detail": "...", "link": "..."}
]
}
]
One job can emit multiple results of different kinds. The source field
is used to exclude the corresponding job from the synthesized error
list (case-insensitive, hyphen-normalized matching against GitHub
Actions job display names).
| job | source | kind (on failure) | section |
|----------------------------|--------------------------|---------------------------|----------------------|
| review-labels | review label gate | action_required / info | Action required |
| lockfile-diff | lockfile-diff | action_required | Action required |
| ci-timings | ci timing | warning / info | Warnings |
| supply-chain scan | supply chain | error / (none) | Job failures |
| supply-chain dep-bounds | supply chain | action_required / (none) | Action required |
| osv-scanner | osv scan | warning / (none) | Warnings |
| uv-lockfile-check | uv.lock check | action_required / (none) | Action required |
| history-check | unrelated histories | action_required | Action required |
| contributor-check | contributor attribution | action_required | Action required |
Jobs that find nothing emit [] (empty array) — no noise info items.
A single comment-live job polls the GitHub Actions API every 15s,
classifies jobs into (completed, pending), assembles the comment, and
upserts it. Merges review_status outputs from all needs jobs via
toJSON(needs.*.outputs.review_status), and downloads the ci-timings
artifact when it becomes available. Shows commit SHA + message below
the header.
The assembler has ZERO job-specific knowledge. It just:
1. collect_from_statuses() — flattens all nested status objects into ReviewItems
2. collect_failed_jobs() — synthesizes errors for failed jobs with no declared status
3. _attach_job_urls() — fills in per-job log links for ALL items
4. render_comment() — groups by severity, renders with group headers
Each item shows links inline next to the title: View report (job-emitted
URL) and View job (auto-attached logs link). Each info item is its own
collapsible <details> block.
# ૮ >ﻌ< ა ci review
running on abc1234 — commit message first line
## ❌ Job failures
### {title} · [View job](url)
{summary}
## ⚠️ Action required
### {title} · [View job](url)
{summary}
**How to fix:**
{how_to_fix}
## ⚠️ Warnings
### {title} · [View report](url) · [View job](url)
{summary}
{detail}
<details><summary>{title}</summary>
{content}
</details>
Still running 3 jobs: ci-timings, docker
- test_assemble_review_comment.py (48 tests): collect_from_statuses,
collect_failed_jobs with exclude_sources, _attach_job_urls,
render_comment (group headers, inline links, commit info, per-item
details, pending footer), assemble integration
- test_live_comment.py (16 tests): classify_jobs pure function
- test_timings_report.py (10 tests): generate_review_status nested format
- test_lockfile_diff.py (6 tests)
- test_classify_changes.py (32 tests, pre-existing)
* ci: migrate AUTOFIX_BOT_PAT to GitHub App token
Replace the long-lived fine-grained PAT (AUTOFIX_BOT_PAT) with short-lived
(1-hour) installation access tokens minted via a new get-app-token composite
action wrapping actions/create-github-app-token@v3.2.0.
The PAT was used in 13 spots across 8 workflow files for gh CLI / GitHub API
calls. The per-repo GITHUB_TOKEN (1,000 req/hr) was getting rate-limited when
multiple workflows fire concurrently (deploy-site, skills-index, ci-timings,
supply-chain-audit, js-autofix). App installation tokens get 5,000 req/hr
per installation and are scoped to the App's permissions, not a user account.
New composite action: .github/actions/get-app-token/
- Wraps actions/create-github-app-token@bcd2ba49 (v3.2.0, SHA-pinned)
- Reads APP_ID + APP_PRIVATE_KEY repo secrets
- Outputs a 1hr installation token via steps.app-token.outputs.token
Requires two new repo secrets (set after creating the GitHub App):
- APP_ID: the App's numeric ID
- APP_PRIVATE_KEY: the PEM private key
App installation permissions needed:
contents: write (js-autofix push, pypi release upload)
pull-requests: write (js-autofix PR create/merge, supply-chain comment)
issues: write (skills-index-freshness issue creation)
actions: write (skills-index workflow trigger)
workflows: write (skills-index triggers deploy-site.yml)
The AUTOFIX_BOT_PAT secret can be deleted once CI passes on this PR.
The comment in js-autofix.yml noting that PAT pushes trigger downstream
workflows is updated — App tokens have the same property (they are not
GITHUB_TOKEN), so the concurrency-cancel loop logic is unchanged.
* style(desktop): satisfy merged eslint/prettier config
The SSH modules predate the stricter lint config that landed on main (curly, no-empty, perfectionist sorting, prettier). Mechanical lint:fix + fmt pass, empty catch blocks filled with the codebase's void-0 convention, and inline no-control-regex disables on the three deliberate control-char patterns (same pattern as lib/ansi.ts).
* fix(ci): pass App secrets as inputs to composite action
Composite actions cannot access the secrets context — the runner's
template engine rejects secrets.* references at load time with
'Unrecognized named-value: secrets'.
Move APP_ID and APP_PRIVATE_KEY from direct secrets.* references inside
the composite action to inputs passed by each calling workflow. The
fallback logic (GITHUB_TOKEN when APP_ID is empty, for fork PRs) stays
in the composite action's check step.
* fix(ci): add detect to all-checks-pass needs so its failure blocks merge
If detect fails, all downstream sub-workflows get SKIPPED (they have
needs: detect). all-checks-pass used if: always() and only checked the
sub-workflows — which all showed as 'skipped' (= success) — so it passed
even though the root cause (detect) failed. This made the PR mergeable
despite a broken CI pipeline.
Add detect to all-checks-pass needs so its failure propagates to the
gate job and blocks the merge.
* fix(desktop): bump skills test timeout to fix cold-start flake (#68235)
Test 1 in skills/index.test.tsx pays the full cold-start cost (jsdom env
init + module transform + the @/hermes/@/store/profile import graph),
which pushed past vitest's 5000ms default under load — caught at 8871ms
on one run, 6.6s pure test time on another. Tests 2-4 are ~30-130ms
each because all that setup is already cached, so only test 1 was at
risk of timing out.
Bump the describe-level timeout to 15s. Verified with 10 consecutive
runs, 4 of which took 5.5-6.6s of test time and would have hard-failed
under the old 5s default.
* feat(desktop): open multiple full app windows (electron)
Add createInstanceWindow() — a full-chrome peer of the primary that
renders the complete app (sidebar, routing, its own draft) against the
shared backend, so several GUI windows can run at once. Mirrors the
primary's window options + chatWindowWebPreferences (backgroundThrottling
stays off so a streamed answer never stalls when blurred) but never
overwrites the mainWindow global and doesn't respawn the backend — the
renderer's getConnection() joins the running one. New windows cascade off
their source via the pure, tested instanceWindowBounds().
Exposed via the hermes🪟openInstance IPC and a "New Window" File
menu item. Per-window fullscreen state now targets the window itself, and
titlebar/native-theme repaints reach every open chat window instead of
only the primary.
Retires the now-orphaned compact new-session pop-out (its only caller was
⌘⇧N, repointed in the follow-up commit): drops createNewSessionWindow,
the hermes🪟openNewSession handler, and the newSession/new=1 URL
flag.
* feat(desktop): wire New Window to ⌘⇧N + command palette
Repoint session.newWindow (⌘⇧N) from the compact new-session pop-out to
openNewWindow(), which opens a full peer instance via the new openWindow
bridge, and add a "New Window" entry to the ⌘K palette (shown with its
hotkey hint, gated on canOpenNewWindow()). Relabel the action "New window".
Drops the retired openNewSessionWindow bridge and the vestigial
isNewSessionWindow()/new=1 flag; renames the shared opener helper.
* fix(desktop): de-dupe cross-window cues so peers don't spam
With multiple full windows, each renderer independently reacts to the
same backend event, so one-shot cues fired N times: OS notifications
(the per-renderer throttle can't see other windows), the turn-end sound
(playCompletionSound runs on every message.complete, ungated by focus),
and auto-spoken replies (double voice when a chat is open in two windows).
Add a single race-free owner in the main process (electron/event-dedupe.ts):
main handles IPC serially, so the first window to claim a key within a
short window wins and peers stay quiet. Notifications collapse at the
hermes:notify choke point; the sound and spoken replies claim via a new
hermes:ambient:claim IPC (keyed by session / reply id). Off Electron the
claim falls back to "emit", preserving single-window behavior.
The sound's mute check runs before the claim so a muted window can't win
the cue and silence an audible peer.
* refactor(desktop): tidy the cross-window deduper
Drop the unused DEDUPE_WINDOW_MS export and rename its interval so
"window" isn't overloaded against BrowserWindow in a multi-window
feature (windowMs → intervalMs). DRY the completion-sound play path.
No behavior change.
* nix: add cage to devDeps
* fix(desktop): avoid false remote gateway reauthentication (#68250)
* fix(desktop): avoid false remote gateway reauthentication
Co-authored-by: Rod-fernandez <rodrigo@nxtlevelsaas.com>
Co-authored-by: David Andrews (LexGenius.ai) <david@lexgenius.ai>
* fix(desktop): harden remote revalidation state
---------
Co-authored-by: Rod-fernandez <rodrigo@nxtlevelsaas.com>
Co-authored-by: David Andrews (LexGenius.ai) <david@lexgenius.ai>
* fix(desktop): keep composer draft across compression tip rotation (#68079)
* fix(desktop): keep composer draft across compression tip rotation
Auto-compression swaps the live stored session id while the user may still
be typing. Scope the composer/queue key on the lineage root and migrate any
tip-keyed draft/queue entries onto that durable key when the tip rotates so
the in-progress prompt does not vanish when the response lands.
* test(desktop): cover draft survival across compression tip rotation
Add regression coverage for migrateSessionDraft, lineage-scoped composer
keys, and the rotation path that previously wiped an in-progress draft.
* fmt(js): `npm run fix` on merge (#68305)
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
* fix(desktop): Stop parks the queue instead of firing the next queued prompt
Interrupting a busy turn with the Stop button (or Esc) settles the
session to idle, and the edge-independent auto-drain immediately submits
the head of the composer queue. The user pressed Stop to halt the agent,
but it looks like Stop skipped the current turn and kept going — and the
queued text is hard to find, since its only surface is the collapsed
'N queued' pill above the composer.
The old userInterruptedRef latch (a23728dcc) fixed this but was removed
in #40221 because it also suppressed the drain that send-now-while-busy
depends on. This reintroduces the halt with source awareness instead of
a blanket latch:
- Explicit halts (Stop button, composer Esc, chat-focus Esc, the
streaming message's hover Stop, runtime cancel) park the session's
queue before interrupting. Parked queues are skipped by both
auto-drain paths (mounted ChatBar + background drainer).
- Interrupts that exist to advance the queue (send-now-while-busy)
unpark first, so the settle drain they rely on still flows.
- The park lifts on any renewed intent: resume, a manual drain (Enter
on empty composer or the per-row send arrow), queueing a new prompt,
or emptying the queue. It migrates with entries on a runtime re-key
and is deliberately not persisted (a fresh process starts unparked).
- The queue panel expands on park, switches to 'N Queued — paused' with
a pause icon, and grows a Resume action, so the held prompts are
visible instead of reading as vanished.
Store contract, hook wiring, and background-drain coverage included;
docs updated.
* fix(cli,tui): recall real paste content on up-arrow
Large pastes collapse to a placeholder in the composer, but input history
stored the placeholder — so up-arrow recall showed a truncated reference
(CLI) or lost the content entirely (TUI, where the `[[…]]` label has no
backing snip after submit).
Store the expanded content in history instead:
- CLI: `_inline_pastes()` expands `[Pasted text #N -> file]` into the buffer
before `reset(append_to_history=True)`; also reused by the external editor
(dedup). History nav suppresses re-collapse of recalled content.
- TUI: `dispatchSubmission` pushes `expandSnips(pasteSnips)(full)`; idempotent
on label-free text so re-submitting a recalled entry stays stable.
* fix(cli): suppress CPR on POSIX local TTYs under load
Delayed ESC[6n replies leak as ^[[row;colR into the classic CLI on
SSH/slow PTYs (#13870) and on local POSIX TTYs under heavy subagent
load. Suppress CPR on non-Windows platforms (layout hint only); keep
native Windows on prompt_toolkit's default pending native coverage.
Wire selection through _select_classic_cli_pt_output.
* test(cli): prove local CPR leak and Application CPR-disabled wiring
Add a delayed-CPR PTY harness (no SSH) plus selection/Application
assertions for POSIX local and Windows preserve-default. Update the
gating unit test to the new contract.
* refactor: drop platform kwarg, fix PTY test cleanup
- Remove redundant platform= test seam from _terminal_may_leak_cpr();
use monkeypatch.setattr(sys, 'platform', ...) consistently in both
test files.
- Wrap PTY tests in try/finally for fd cleanup on assertion failure.
- Guard select.select() in terminal thread against OSError after fd
close (fixes PytestUnhandledThreadExceptionWarning).
- Trim PR-number reference from test module docstring.
* docs(portal): remove retired Nous Chat references
* fix(web/ddgs): isolate DuckDuckGo search in a disposable process
ThreadPoolExecutor timeouts cannot fire when primp holds the GIL in
native code (#68096). Run each search in a child process the parent can
terminate/kill, and honor tools.interrupt between polls.
* test(web/ddgs): cover GIL-hold timeout, interrupt, and worker reap
Regression tests for #68096: native GIL-hold and sleep hooks must time
out or interrupt promptly with no orphaned search workers.
* fix: sanitize subprocess env for DDGS worker
os.environ.copy() passes all Hermes secrets (gateway tokens, API keys,
dashboard session tokens) into the DDGS child process. Use
_sanitize_subprocess_env() to strip Hermes-managed secrets before
spawning the worker.
* fix(agent): pass persisted-prefix boundary when rotation flushes on cold resume (#68196)
The legacy rotation branch in agent/conversation_compression.py flushes the
current turn to the OLD session before ending it (#47202) via
_flush_messages_to_session_db(messages) with no conversation_history boundary.
On the first turn after a cold Desktop resume, the restored transcript rows
live in the message list as plain dicts that have not yet been stamped with
_DB_PERSISTED_MARKER — the normal turn flush that stamps them runs after
preflight compression. With no boundary, _flush_messages_to_session_db builds
an empty history_ids set and treats every restored row as new, durably
re-appending the whole transcript to the parent session. Repeated
restart/resume + threshold compression keeps growing the parent transcript.
Pass messages[:_persist_user_message_idx] (the already-durable prefix that
turn_context anchors before preflight runs, guarded for int/bounds) as
conversation_history so the flush skips the persisted rows by identity and
writes only the current turn's new messages.
Adds a regression test that pre-populates SQLite, cold-loads the transcript,
appends one current user row, and forces rotating compression: it fails before
this change (parent grows to 5 rows) and passes after (parent holds the two
originals plus the single new turn).
* fix(desktop): prevent contentEditable composer input from visually collapsing to near-zero height
Fix#68095
The composer input box (contentEditable div) randomly shrank to a tiny/pixelated
size when typing character-by-character (paste worked fine). Root cause: during
per-keystroke input, the normalizeComposerEditorDom cleanup could briefly leave
the contentEditable with zero child nodes, and without intrinsic content the
browser collapsed it visually despite the CSS min-height.
Two-pronged fix:
1. Add min-h-[1.625rem] bracket syntax alongside the CSS variable min-height
to ensure the minimum height is enforced even if the CSS variable resolution
is delayed or overridden by browser defaults.
2. In normalizeComposerEditorDom, ensure the contentEditable always has at
least one <br> child when empty, giving it intrinsic height that the browser
cannot collapse. This is a belt-and-suspenders approach with the CSS min-height.
Closes#68095
* fix(agent): circuit-break AttributeError from commit-splice and detect code skew
Fix#68178
The git-install auto-updater rewrites source while the desktop backend
is live. Because agent/conversation_loop.py is imported lazily on the
first API call, a process can end up running two different commits
spliced together — one commit's AIAgent against another commit's
conversation_loop. When the interface differs, every turn fails
permanently with an AttributeError, and the loop retries indefinitely,
burning provider API calls (576 failures, 149 wasted API calls observed).
Three-prong fix:
1. Circuit-break AttributeError on agent objects: the outer-loop error
classifier now detects AttributeError targeting agent/run_agent
modules and breaks immediately instead of continuing the retry loop.
2. Code skew detection for desktop/serve backend: run_agent.py now
snapshots the checkout revision at import time and exposes a cheap
per-iteration check that the conversation loop uses to refuse new
work with a clear 'restart required' message before the lazy import
can crash.
3. Informative error message: when code skew is detected, the user
gets a clear explanation of the mismatch (boot revision vs current
revision) and actionable guidance to restart the application.
* fix(telegram): preserve fatal recovery handoff
Release the current polling-recovery task's ownership before invoking
the fatal-error handler. The runner bounds adapter cleanup in a child
task; disconnect() cancels the tracked polling-recovery task, so
retaining the current notifier in _polling_error_task would cancel the
fatal callback before the runner can finish its reconnect-queue or
shutdown decision.
The new _handoff_polling_fatal_error() helper clears
_polling_error_task only when it is the current notifier. Other
recovery tasks remain tracked and are still cancelled and awaited
during teardown.
Covers both network retry exhaustion and polling-conflict exhaustion.
Replaces the misleading "Restarting gateway" message with "Escalating
to gateway recovery".
Fixes#68406.
* fix(telegram): widen fatal handoff to heartbeat watchdog path
The wedged-recovery heartbeat watchdog (line 2526) calls
_notify_fatal_error() directly from the heartbeat task. disconnect()
cancels _polling_heartbeat_task unconditionally (no current_task guard,
unlike _polling_error_task). Same bug class as #68406: the child
disconnect cancels the heartbeat parent before the runner can queue
reconnect.
Widen _handoff_polling_fatal_error() to also clear
_polling_heartbeat_task when it is the current task, and route the
heartbeat watchdog call site through the handoff helper.
Co-authored-by: Imgaojp <6065749+Imgaojp@users.noreply.github.com>
* fix(tests): make the live-system-guard canary fail closed
tests/test_live_system_guard_self_test.py executes real kill primitives
(os.kill(-1, SIGTERM), os.killpg, pkill -f python) and depends entirely on
the autouse _live_system_guard fixture in tests/conftest.py to intercept
them. That makes the canary fail-OPEN: in any collection context where the
file is present but its home conftest is not — a published sdist that ships
tests/ but not tests/conftest.py, a tree assembled by copying test*.py (that
glob does not match conftest.py), pytest --noconftest, or a foreign rootdir —
the primitives fire for real, and os.kill(-1, SIGTERM) SIGTERMs every process
the invoking user owns (a full desktop-session kill was reported in the field).
Add an autouse fixture that refuses to run any canary test unless the guard is
provably active. The one thing the canary can detect about its own safety is
that the guard monkeypatches os.kill with a plain Python function, whereas the
unguarded primitive is a C builtin — so the probe keys off that. Tests marked
@pytest.mark.live_system_guard_bypass still opt out, matching the guard's own
bypass contract (e.g. test_bypass_marker_disables_guard). With the guard loaded
every canary test behaves exactly as before; without it each test refuses at
setup with zero side effects.
Fixes#68311
* fix(billing): rename user-facing "terminal billing" copy to Remote Spending (#68355)
* fix(billing): rename user-facing "terminal billing" copy to Remote Spending
The capability was renamed Remote Spending on the portal (consent CTA:
"Allow Remote Spending"; per-terminal states Granted/Stopped), but the
terminal, desktop, and docs still said "terminal billing" everywhere.
- Feature name: Remote Spending in titles/labels, lowercase mid-sentence.
- Step-up action verb is now "allow", matching the portal consent CTA.
- Kill-switch-off recovery copy points at the actual control ("a billing
admin can turn it on from the portal's Hermes Agent page") instead of
the dead-end "manage it on the portal".
- Per-terminal revoke copy uses the portal vocabulary ("stopped").
- Wire identifiers (cli_billing_enabled, cli_billing_disabled, ...) are
unchanged; copy, comments, docs, and test expectations only.
* fix(billing): correct the post-step-up denial diagnosis + finish the desktop rename
Adversarial review findings: (1) a repeated insufficient_scope after a
successful step-up is a per-terminal authorization failure, but the copy
blamed the org kill-switch and pointed at the wrong recovery control —
now: "Remote Spending still isn't active for this terminal — the
authorization didn't take. Retry, or make this change on the portal."
(2) the desktop step-up flow started in Remote Spending vocabulary but
finished in "billing management access" — renamed both end states.
(3) prettier formatting on the touched files (matches the post-merge
fmt bot).
* feat(tui): show the plan catalog in /subscription on Free (#68357)
* feat(tui): show the plan catalog in /subscription on Free
The server returns the tier list even with no subscription, but the
overlay hid the picker behind can_change_plan && !isFree, so a Free
account got only "Start a subscription" with no idea what the plans
cost. Now:
- Overview on Free offers "Choose a plan" whenever the catalog has
enabled paid tiers.
- The picker on Free lists each plan as name · price · monthly credits
(no upgrade/downgrade hints — there is nothing to move from), and
picking one opens the portal, where starting a subscription actually
happens (card capture + checkout live there; the upgrade RPC requires
an existing subscription).
- Paid-plan behavior (preview → confirm → apply) is unchanged.
* refactor(tui): compute the picker row suffix once
Review feedback: the isFree fork duplicated the label template and run
handler; only the suffix differs.
* fix(tui): arm the busy guard before the Free portal handoff
Adversarial review: the Free branch returned before setting busyRef, so
a double-Enter could open the portal twice; and the picker narrated a
handoff that openManageLink already narrates (duplicate on success,
contradictory on failure). Guard first, let the helper do the talking.
* fix(tui): monthly credits are dollars — label them as such
The Free picker showed "1000 credits/mo" for what is $1,000 of monthly
credit — render "$1,000 credits/mo" (grouped, dollar-signed).
* feat(tui): render the Free-plan catalog inline in the /subscription overview
Sid ruling: the upsell belongs where the user already is — no
intermediate "Choose a plan" hop. On Free the overview lists each paid
plan (name · $/mo · $credits/mo) as a pickable row; picking opens the
portal (openManageLink narrates). The generic "Start a subscription"
row survives only when the catalog is empty. The picker reverts to its
original change-only form (Free never reaches it).
* feat(desktop): tier catalog chips on the Subscription row
Desktop parity with the TUI inline catalog (Sid ruling): accounts that
can act see the plans where they already are — Free gets the upsell
list (every chip opens the portal), a subscriber sees all tiers with
the current one marked inert. Members and team contexts see no chips.
Chips learn an optional url (portal handoff) in the shared row model.
* chore(tui): fixture harness mirrors the live tier catalog
The dev screenshot fixtures showed invented plans ($50 Super / $99
Ultra, "1,000 credits"); align with the real catalog ($20/$100/$200
with $22/$110/$220 monthly credits) so fixture renders cannot be
mistaken for product truth. The overlay itself always reads tiers from
the subscription API.
* chore: trim narration comments
* fmt(js): `npm run fix` on merge (#68462)
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
* fix(relay): attach metadata.user_id on guild replies for egress fallback (#68320)
The relay adapter re-attaches an egress discriminator on outbound replies
so the connector can resolve the owning tenant. It captured scope_id for
scoped (guild) messages and user_id for DMs, but as MUTUALLY EXCLUSIVE:
a scoped inbound hit an early return, so the author's user_id was never
recorded, and _with_scope only attached user_id when there was no
scope_id. Guild replies therefore went out with scope_id only.
That's fine while the guild has a provision-time route row. But a MANAGED
Discord agent joins guilds dynamically (the shared bot is added to /
removed from servers at runtime), and GATEWAY_RELAY_ROUTE_KEYS — the only
thing that writes guild route rows — is a self-hosted, static field never
stamped for managed agents. So their guild has no route row, the
connector's guild-route lookup misses, and with no user_id on the frame
there's nothing to fall back to → every guild reply is declined
"discord egress declined: target not routed to an onboarded tenant"
even though INBOUND resolved the same guild fine (via the author-first
SharedSocketRouter.targets() fallback).
Fix: capture the authentic author user_id for EVERY inbound (DM and
scoped alike) and re-attach it on the outbound reply alongside scope_id.
The connector consults it only on a route/scope miss, so carrying both
never overrides routing-table resolution. This is the gateway half of the
paired gateway-gateway change (makeDiscordTenantOf guild-route-miss
author-binding fallback); together they make guild replies resolve the
same observed-author way inbound already does.
Tests (tests/gateway/relay/test_relay_adapter.py): a guild reply now
carries both scope_id AND user_id; a scoped inbound with no author still
yields scope_id only (never invents one). Verified fail-without /
pass-with.
* build: declare pywin32 as a direct win32 dependency
hermes_cli/windows_ssh_runtime.py imports win32security/win32file/etc.
directly but pywin32 only arrived transitively via concurrent-log-handler
-> portalocker. Declare it with a sys_platform gate so the Windows SSH
runtime doesn't depend on the logging dep chain. Review follow-up on
PR #68130.
* fix(desktop): preserve dragging with empty titlebar slots
* Revert "fix(agent): circuit-break AttributeError from commit-splice and detect code skew"
This reverts commit 3a9b9d65d5.
* fix(context): revalidate Codex OAuth context windows
* test(context): document Codex cache persistence coverage
* fix(context): scope Codex catalogue cache by credential
* test(context): cover Codex context rollback
* fix(compression): report live-resolved Codex window in the autoraise notice
The autoraise banner hardcoded '272K' for the gpt-5.4/5.5/5.6 family, but
the Codex /models catalog is authoritative and shifts server-side (gpt-5.6
served 372K during July 9-18, 2026 before OpenAI rolled it back). Pass the
compressor's live-resolved context_length through so the notice reports the
window the session actually got; the static 272K/128K text remains as the
fallback when no resolved value is available.
* fix(codex): send ChatGPT-Account-Id on /models probes
The Codex backend returns the per-account model catalog only when the
ChatGPT-Account-Id header is present. Without it, GET /backend-api/codex/models
responds 200 OK with {"models":[]} and the picker silently degrades to the
hardcoded fallback list — which is stale or wrong for the active plan
(no GPT-5.6 family, wrong context windows).
This was the upstream bug behind slow first responses and HTTP 520/120s SSE
hangs: Hermes was sending invalid slugs because the probe never saw them in
the catalog, and Codex's request builder also depends on the same JWT claim
that's now being threaded through both probe paths.
Fixes the probe-side paths in hermes_cli/codex_models.py and
agent/model_metadata.py by extracting chatgpt_account_id from the OAuth JWT
(mirroring the request-side logic already in auxiliary_client.py) and sending
it as a header.
Verified live:
- _fetch_models_from_api now returns the 10-model catalog (gpt-5.6-sol,
gpt-5.6-terra, gpt-5.6-luna, gpt-5.5, gpt-5.4, gpt-5.4-mini,
gpt-5.3-codex-spark, 3x -pro variants) instead of [].
- _fetch_codex_oauth_context_lengths resolves all 8 account models to 272K
context (matches direct API probes of the same account).
- end-to-end: hermes chat -m gpt-5.6-sol -q 'Reply with one word: pong'
returns 'pong' cleanly via the openai-codex route.
Same class of bug as PR #64760.
* test(codex): cover ChatGPT-Account-Id header on /models probe
Add regression tests locking in the new behavior: a JWT carrying a
chatgpt_account_id claim causes the probe to send ChatGPT-Account-Id,
while a malformed token omits the header instead of crashing.
* fix(tools): make the tool-search context gate provider-aware (#68589)
_resolve_active_context_length() called get_model_context_length() with the
model id alone, so provider-enforced windows (e.g. Codex OAuth's 272K for
gpt-5.x vs the direct API's 1.05M) never reached the tool-search activation
gate — it sized against generic metadata for the same slug.
Resolve the runtime provider for the configured model and pass provider,
base_url, and api_key through. If credential resolution fails (offline, no
keys), degrade to a provider+base_url-only lookup so the static
provider-aware fallbacks still apply; explicit model.context_length keeps
short-circuiting as before (#46620). Gap flagged during review of #16735.
* feat(skills): bundle docx, xlsx, and pdf office skills; refresh powerpoint (#68595)
Non-technical users asking for Word docs, spreadsheets, or PDF work had
no bundled skill coverage — docx/xlsx creation required discovering and
installing hub skills, and PDF manipulation had no skill at all beyond
OCR extraction and nano-pdf edits.
- skills/productivity/docx: create (docx-js), edit (unzip -> XML -> zip),
tracked changes, comments, validation. Adapted from anthropics/skills.
- skills/productivity/xlsx: openpyxl creation/editing, mandatory
LibreOffice recalc gate, formula-compatibility rules, financial-model
conventions. Points at optional excel-author for finance-grade work.
- skills/productivity/pdf: merge/split/rotate/watermark/encrypt, form
filling (AcroForm + flat overlay scripts), text/table extraction,
reportlab creation, forms.md + reference.md companions.
- skills/productivity/powerpoint: synced to current upstream pptx skill —
richer pptxgenjs corruption footguns, template workflow, validate.py +
validators + thumbnail.py, font-substitution QA guidance; drops the
stale pack.py/editing.md/pptxgenjs.md workflow files.
- Cross-linked ocr-and-documents, nano-pdf, excel-author via
related_skills so each office skill routes to its siblings.
- deliverable-mode docs mention the new skills; regenerated per-skill
docs pages, catalogs, and sidebar.
- tests/skills/test_office_document_skills.py: frontmatter contracts,
referenced-script existence, schema-map integrity, cross-link
resolution, script compilation.
E2E validated: docx create->render->edit->validate, xlsx recalc
(SUM + _xlfn.TEXTJOIN evaluate correctly), pdf create->merge->extract,
pptx generate->validate->thumbnail.
* fix(approval): raise gateway approval timeout to 300s, honest stale-tap UX, offer Always on mixed prompts (#68597)
Three related messaging-approval fixes:
1. approvals.timeout default 60 -> 300. PR #63501 collapsed the gateway
wait onto the canonical approvals.timeout (previously
gateway_timeout=300), silently shrinking messaging approval windows
to 60s. Push-notification approvals routinely arrive later than a
minute; taps landed after the wait had already failed closed.
2. Stale-tap honesty: adapters resolved the approval AFTER rendering
'<checkmark> Approved by <user>' (Telegram/Discord/Slack), or ignored a zero
resolve count (WhatsApp Cloud/Feishu). A tap on an expired prompt
claimed approval while the command had already been denied. All
button paths now resolve first and render 'Approval expired -
command was not run' when nothing was waiting.
3. Mixed-warning prompts (dangerous pattern + tirith finding) now offer
Always: the persistence layer already permanently allowlists the
pattern key and downgrades the tirith key to session scope, but the
UI hid Always whenever ANY tirith warning was present. Pure-tirith
prompts still withhold Always (content findings are session-max by
design), and Smart-DENY overrides remain once-only.
* feat(secrets): one-command token rotation + actionable startup errors for all secret sources (#68605)
* feat(secrets): one-command token rotation + actionable startup errors for all secret sources
When a Bitwarden machine-account token expired, users saw a raw Rust
error dump (invalid_client + Location: + backtrace hints) and the only
fix was manually editing .env or re-running the whole setup wizard.
- New `hermes secrets bitwarden token` / `hermes secrets onepassword
token`: paste a new token (masked prompt or flag), the command probes
the backend BEFORE persisting — a rejected token changes nothing; a
good one is written to .env and the fetch caches are cleared.
- New optional SecretSource.remediation(kind, cfg) hook: startup
warnings now print a '→ Run `hermes secrets <name> token`…' fix-it
line after any fetch error, for bundled AND plugin sources (generic
per-ErrorKind defaults in the ABC).
- bws stderr is summarized to its cause line (Location:/backtrace noise
dropped) and invalid_client/invalid_grant/400 identity rejects are
now classified AUTH_FAILED (was INTERNAL) with a plain-English
explanation naming the token env var.
- op whoami probe accepts a candidate token so rotation validates the
NEW credential, not the ambient one.
Additive hook with defaults — no SECRET_SOURCE_API_VERSION bump.
* docs: fix MDX parse error in secret-source-plugin hook table
Escaped backticks around a <name> placeholder made MDX parse it as an
unclosed JSX tag, breaking the docs-site build. Use a plain code span
instead.
* feat(desktop): configure repository discovery (supersedes #67630) (#68642)
* feat(desktop): configure repository discovery
* fix(config): preserve additive default migration
* fix(desktop): stabilize session-actions-menu gateway mock for repo-scan subscribe
projects.ts now runs $gateway.subscribe(syncReposScanning) at module load, and
nanostores fires the subscriber synchronously. session-actions-menu.test.ts
reaches projects.ts transitively via the session store but mocked
@/store/gateway without $gateway, crashing the whole desktop vitest suite
("No \ export is defined"). Simply adding $gateway: atom(null) exposed a
second issue: the synchronous subscriber calls the mock's activeGateway()
during the transitive import, before the module-level const initializes (TDZ).
Hoist the mock fns via vi.hoisted() so activeGateway is defined before the
hoisted vi.mock factory runs, and add $gateway: atom(null) to the mock. Mirrors
the self-contained mock pattern already used in projects.test.ts. Also maps the
PR author's commit email for attribution.
Supersedes #67630; incorporates review feedback from that PR.
Co-authored-by: Rudimar Ronsoni <rudimar@outlook.com>
---------
Co-authored-by: Rudimar Ronsoni <rudimar@outlook.com>
Co-authored-by: Austin Pickett <austinpickett@users.noreply.github.com>
* fix(desktop): ⌘W closes visible file tab when preview selection is stale (#68639)
* fix(desktop): make ⌘W close visible file tab on stale preview selection
When the live preview target is gone but $rightRailActiveTabId still points
at preview, file tabs remain on screen while ⌘W fell through to a workspace
no-op. Close the visible file tab instead.
* test(desktop): cover ⌘W close for file tabs and ghost preview selection
Lock the happy path and the stale-preview regression so ⌘W keeps closing
the file tab the rail is actually showing.
* fmt(js): `npm run fix` on merge (#68681)
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
* feat(billing): plan chips and rows deep-link their tier (#68666)
* fix(desktop): drop the decorative top-up credits bar (#68649)
The bar rendered full-or-empty (value 1|0) because top-ups have no
denominator — the wire carries only the current balance and the pool is
open-ended, so a fill fraction is fiction. Show the amount alone;
subscription credits and the monthly cap keep their bars (real
denominators).
* fix(ci): route critical supply-chain findings through review gate (#68833)
Let the scanner report critical findings without failing. The review-label
gate owns the action-required status and blocking result, allowing the
ci-reviewed label rerun to clear both CI and the PR comment.
* fix: `tool_calls` double-encoding on import (#68856)
* nix: add `cage` to devShell
* test(desktop): add pre-filled sessions support
Exports createSandbox, writeMockProviderConfig, writeEnvFile,
buildAppEnv, findElectron, and launchDesktop from fixtures.ts so
specs can compose their own seeded-backend fixtures without duplicating
the sandbox/config/launch logic.
* test(desktop): auto-fail e2e tests on error banner
Adds a shared test fixture (e2e/test.ts) that wraps @playwright/test's
page with an error-banner guard. When any [role="alert"] element
(error notification toast) appears in the DOM during a test, the test
fails with the error message text.
The guard uses:
- A MutationObserver (injected via addInitScript) that watches for
[role="alert"] elements appearing at any point during the test
- A final DOM scan in afterEach for alerts still visible at teardown
- Deduplication so the same error text only fires once
All existing e2e specs updated to import { test, expect } from './test'
instead of '@playwright/test'. No per-spec setup needed — the guard is
auto-installed on every page via the extended fixture.
This catches issues like the "resume failed" error banner that can
appear during session loading — previously the test would pass while
an error toast was silently visible on screen.
* fix(state): parse tool_calls JSON string before re-serializing
_insert_message_rows and append_message both do json.dumps(tool_calls)
to serialize the field for SQLite storage. But when tool_calls arrives
as a JSON string (from import_sessions / export_session, which store it
as TEXT), json.dumps double-encodes it — wrapping the already-serialized
string in quotes and escaping the inner quotes.
When _rows_to_conversation later does json.loads(row['tool_calls']),
the double-encoded string parses back to a plain string (not a list).
_history_to_messages then iterates this string character-by-character,
calling tc.get('function', {}) on each char — 'str' object has no
attribute 'get'.
This was a pre-existing bug (on main), but only triggered by the
import_sessions path (the live agent always passes tool_calls as a
Python list). The e2e error-banner guard caught it via the 'Resume
failed' notification toast.
Fix: in both append_message and _insert_message_rows, parse tool_calls
with json.loads first if it's a string, then re-serialize.
* fix(desktop): exempt boot-failure from error guard
- boot-failure: add allowErrorBanners() beforeEach — these tests
deliberately trigger boot errors, so error toasts are expected
- test.ts: export allowErrorBanners() opt-out + reset flag in afterEach
* feat(status-bar): add /battery toggle for a color-coded battery read-out
Add an opt-in battery indicator to the CLI and TUI status bars, shown as
the first element and colour-coded by charge (green/yellow/orange/red, or
green while charging). Off by default and a no-op on machines without a
battery.
- agent/battery.py: shared psutil-backed reader with a short TTL cache,
category bucketing, and a compact 🔋/⚡ label. Fails open to
"unavailable" everywhere.
- CLI: /battery [on|off|status] toggle persisted to display.battery,
rendered first in every status-bar width tier.
- TUI: /battery slash command, config sync, a system.battery RPC polled
while enabled, and a pinned first segment in StatusRule.
* fix(approval): restore session approval for Tirith-flagged commands
Adds an allow_session flag to the gateway approval payload so adapters
can render the session tier independently of the permanent tier. Matrix
gains a session reaction (🌀) and a reaction legend; pure-tirith prompts
now offer once/session/deny instead of collapsing to once/deny.
Salvaged from PR #67312, adapted to the allow_permanent semantics that
landed in #68597 (Always offered when any dangerous-pattern warning is
persistable; pure-tirith prompts stay session-max).
* fix(approval): honor allow_session across all button adapters
Widen the allow_session tier from Matrix to every adapter the gateway
notifies: Telegram, Discord, Slack, Feishu, and Teams gate their Session
button on it; WhatsApp Cloud and qqbot accept the kwarg (no session tier
in their button sets). Also thread allow_session through the plugin-
escalation gate, the execute_code guard payload, and the plain-text
fallback so every notify path carries the same capability flags.
* test(approval): cover allow_session tiers in Matrix reaction seeding and gateway payload
Update the Matrix reaction-seeding contract to the four-reaction default
(once/session/always/deny), add tirith-tier (session without always) and
no-session-tier cases, and assert allow_session=True in the tirith
gateway payload.
* fix(desktop): wrap missing sidebar icon-button tooltips (#67500)
* fix(desktop): wrap sidebar icon buttons in Tip tooltips
Several icon-only buttons in the sidebar (header actions, workspace
menu, project menu, session actions, load-more) had aria-label but
no visual tooltip on hover. Wrap them in the existing <Tip> component,
matching the pattern already used elsewhere (e.g. ProfilePill).
No behavioral changes -- purely wraps existing buttons.
Adds vitest coverage asserting the Tip wrapper (data-slot=tooltip-trigger)
for 6 of 7 files; index.tsx is a 1500+ line top-level page component and
was verified manually via screenshots instead.
* fix(desktop): satisfy consistent-type-imports lint rule in project-dialog test
* test(desktop): update session-row mocks for restored sessionColorById
* fix(desktop): compose Tip around the real trigger instead of inside it
Tip was being placed as SessionActionsMenu's/PlatformAvatar's DIRECT child,
which asChild then cloned instead of the actual button/span. Neither Tip nor
PlatformAvatar forwarded the injected onClick/ref, so both silently dropped
the wiring:
- session-actions-menu.tsx: Tip now wraps DropdownMenuTrigger internally
(new ooltip prop) instead of the caller wrapping its children in Tip.
- platform-icon.tsx: PlatformAvatar now forwards ref and spreads rest props
onto its span so a wrapping Tip's trigger actually attaches.
- session-row.tsx: updated call site to use the new tooltip prop.
- Added session-actions-menu.test.tsx exercising the real DropdownMenu open
behavior end-to-end (no Tip/Dropdown mocks).
- session-row.test.tsx no longer mocks PlatformAvatar's behavior; it now
exercises the real (fixed) component for the handoff-avatar tooltip.
* fix(desktop): compose Tip outside PopoverAnchor in ProjectMenu (#67500)
* test(desktop): update session-row test for the tooltip-prop composition (cbbbeb2fd)
* fix(desktop): satisfy consistent-type-imports in session-row.test.tsx mocks
* chore: retrigger CI
* test(desktop): stop mocking PlatformAvatar's behavior (#67500, third pass)
The mock was re-introduced by a prior edit that fixed an unrelated lint
error, silently undoing the earlier fix where this test started exercising
the real (forwardRef) PlatformAvatar. Removed the mock; updated the two
handoff-avatar tests to query the real component's rendered span instead of
text content, since it renders a brand SVG icon for known platforms rather
than the platform name as text.
* fmt(js): `npm run fix` on merge (#68867)
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
* fix(gateway): detect stale lock when macOS psutil returns valid start_time for recycled PID
On macOS, the lock record's start_time is None (no /proc at creation),
but psutil.Process(recycled_pid).create_time() returns a valid float
for the unrelated process that now owns the PID. The old condition
required both sides to be None before falling back to cmdline checking,
so the recycled PID was never detected as stale.
Change the fallback condition from AND to OR: when either side's
start_time is missing, fall back to cmdline-based gateway detection.
Fixes#53763
* fix(gateway): handle PermissionError on stale root-owned lock file
When the macOS launchd service runs in a Background session, the gateway
process spawns as root and creates a root-owned gateway.lock. On restart
as the normal user, open() on that file raises PermissionError, crashing
the gateway immediately and entering a launchd crash loop.
Catch PermissionError in is_gateway_runtime_lock_active(), remove the
stale lock file, and return False so the new process can start cleanly.
Fixes#42685
* fix(gateway): guard acquire_gateway_runtime_lock against root-owned lock PermissionError
Widen the PermissionError handling from is_gateway_runtime_lock_active
(#42689) to the sibling open() in acquire_gateway_runtime_lock: a stale
root-owned gateway.lock left by a launchd Background session previously
crashed the acquiring process. Unlink the stale file and retry once; if
the unlink or retry fails, return False cleanly instead of raising.
* fix(gateway): make stale scoped-lock removal atomic via tombstone rename
Replace the unlink()+O_EXCL sequence in acquire_scoped_lock with an
atomic os.replace() of the stale lock to a <lock>.stale tombstone
followed by the existing O_EXCL create. With plain unlink(), two racing
starters could both judge the lock stale and the second unlink() would
silently delete the first racer's freshly-created lock — both would then
'win'. os.replace() guarantees exactly one racer claims the stale file;
the loser gets FileNotFoundError and falls through to O_EXCL, which
admits at most one winner. Tombstones are cleaned up immediately;
behavior is otherwise identical.
* fix(gateway): detect stale gateway_state.json in `gateway status` (TTL + PID liveness)
Verified: applies cleanly and the patched module compiles. Tests are
described in the PR body (not bundled in this commit).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
* test(gateway): cover stale gateway_state.json detection (TTL + PID liveness)
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
* fix(gateway): take over live platform-lock token holders once
When --replace misses a cross-HERMES_HOME Telegram token holder, platform
connect used to retry forever. Terminate a verified gateway holder once
(with the takeover marker) and re-acquire the scoped lock (#65176).
Co-authored-by: Cursor <cursoragent@cursor.com>
* chore(contributors): map jaretbottoms@gmail.com -> jbbottoms (PR #65178 salvage)
* fix(gateway): reap the replaced gateway's orphaned children on POSIX
Builds on jbbottoms's #65178 takeover fix (cherry-picked as the previous
commit). Windows --replace already tree-kills via taskkill /T, but the
POSIX paths signalled only the recorded gateway PID — adapter
subprocesses that outlived their parent kept holding scoped token locks
and blocked the replacement gateway.
- gateway/status.py: _snapshot_gateway_children() captures the old
gateway's descendants (psutil, recursive) while it is still alive;
reap_gateway_children() SIGTERMs verified orphans after the main PID
is confirmed dead, waits bounded, SIGKILLs survivors. Identity-aware
(psutil is_running is PID+create-time), skips zombies and children
whose ppid still equals the old gateway (parent actually alive), and
never raises — best-effort with debug/info logging only.
- take_over_scoped_lock_holder() snapshots before terminating and reaps
only on a confirmed successful handoff.
- gateway/run.py: start_gateway --replace snapshots before SIGTERM and
reaps after the old PID is confirmed gone, mirroring taskkill /T.
- tests/gateway/test_replace_child_reap.py: reap/skip/never-raise unit
coverage plus end-to-end --replace ordering (snapshot → terminate →
reap) and the no---replace path never touching the old process.
* chore(contributors): map emails for PRs #66906, #66420, #63398 salvage
* fix(state): probe FTS5 read path in _db_opens_cleanly so partial index corruption is detected (#66724)
`hermes sessions repair --check-only` opens cleanly on state.db files
with partial FTS5 index corruption — base tables read fine, the rolled-back
write probe from #50502 succeeds, and `PRAGMA integrity_check` returns
"ok". But every session_search / /resume title resolution / feature
backed by MATCH / snippet / rank queries errors out with
`database disk image is malformed` because internal shadow-table segments
are bad. The official repair tool then gives false confidence.
Add a representative FTS5 read probe against both `messages_fts` and
`messages_fts_trigram` (the latter backs title resolution). Empty MATCH
strings are accepted by every FTS5 index without requiring populated
content, so the probe is safe on a freshly-init'd DB; missing-table /
missing-column errors fall through to the existing "not yet a populated
DB" branch, matching the write-probe's behaviour. Any other OperationalError
is surfaced as the check reason, which sends `hermes sessions repair` to
its existing FTS 'rebuild' path (repair_state_db_schema, line 616).
Single-file change in hermes_state.py::_db_opens_cleanly. No public API
change. No new imports. Fixes#66724.
* fix(state): also catch sqlite3.DatabaseError in FTS5 read probe (#66724)
The FTS5 read probe in _db_opens_cleanly() only caught
sqlite3.OperationalError. But the corruption class #66724 actually
wants caught — partial shadow-table damage where MATCH / snippet / rank
queries raise DatabaseError("database disk image is malformed") — is a
DatabaseError, not OperationalError. Without this catch the probe
crashes the caller instead of returning a reason, which is exactly the
silent-fail mode the issue describes.
Move the try/except inside the for-loop so each FTS table is probed
independently (one table corrupted should still surface as a reason),
add a separate except clause for DatabaseError that surfaces the same
reason format, and use continue instead of pass so the loop still walks
both tables when only one is missing on a brand-new DB.
Tested by hand: with a corrupted messages_fts_trigram shadow table the
function now returns 'fts5 read probe failed on messages_fts_trigram:
database disk image is malformed' instead of crashing out. Without this
fix it would still crash.
* fix(state): preserve degraded-runtime read probe + use canonical FTS5 classifier
Two follow-ups on top of f842733 (the FTS5 read probe added in #66906):
1. The original probe query used MATCH '', which FTS5 rejects with
'fts5: syntax error near '. Empty MATCH syntax is not valid FTS5.
Switch to MATCH '""' — a quoted empty phrase that parses, scans
zero rows, and exercises the same shadow-table read path the
search tools use. The probe previously never reached the shadow
segments at all on a healthy DB; the read-corruption class was
only being detected because the existing write probe happens to
fail first on a DatabaseError.
2. The probe's degraded-runtime branch only checked the substrings
'no such table' / 'no such column'. On a SQLite build without the
fts5 module, MATCH against a legacy messages_fts table raises
'no such module: fts5' (a different OperationalError class). The
substring check would misclassify that as corruption and trigger
repair, whose final fallback deletes the messages_fts% schema
(#66906 review). Use SessionDB._is_fts5_unavailable_error() — the
canonical classifier already used by the degraded-runtime init
path — to recognize both 'no such module: fts5' and
'no such tokenizer: trigram' as capability errors.
Add tests covering:
- Partial shadow-table damage (read-corruption class)
- Repair brings reads back online
- Healthy degraded DB without fts5 module stays healthy (regression
for the misclassification risk)
- Healthy degraded DB without trigram tokenizer stays healthy
Closes#66906 review feedback
Refs #66724
* fix(state): self-heal FTS corruption on the SessionDB search path too
Complements #66296 (self-heal on the write path): search_messages()'s main
FTS5 MATCH query caught only sqlite3.OperationalError (a query-syntax error →
return empty). A corrupt FTS index raises the malformed / "fts5: corrupt
structure record" class, which is a sqlite3.DatabaseError — the parent of
OperationalError, so it was NOT caught and propagated straight out of
search_messages, crashing session/history search.
The write path now rebuilds and retries on that class, but a read-only
session (cron/CLI history search, or a search issued before any write) never
triggers a write, so its search stayed broken until the next process restart
ran the offline repair.
Catch the DatabaseError corruption class on the search MATCH read too and
route it through the existing one-shot _try_runtime_fts_rebuild(), then retry
the query. The catch is moved outside `with self._lock` so rebuild_fts() can
re-acquire the lock (mirrors _execute_write). The one-shot guard is shared
with the write path, so a single instance never loops on a genuinely
unrecoverable index. OperationalError syntax handling is unchanged (caught
first).
Adds a regression test: with a corrupted messages_fts and no post-corruption
write, search_messages() rebuilds in place and returns the match; without the
fix it raises DatabaseError.
* fix(state): extend search-path FTS self-heal to the CJK/trigram branch
The trigram MATCH branch in search_messages() had the same
OperationalError-only catch that #66420 fixed on the main FTS5 branch: a
corrupt messages_fts_trigram shadow table raises the malformed /
'fts5: corrupt structure record' class (sqlite3.DatabaseError, parent of
OperationalError), which propagated straight out of search_messages and
crashed CJK session/history search for read-only sessions.
Route that class through the shared one-shot _try_runtime_fts_rebuild()
and retry the trigram query (catch moved outside self._lock so
rebuild_fts() can re-acquire it, mirroring the main branch). If the
rebuild is refused (guard consumed / FTS disabled / different error) or
the retry fails, fall through to the existing LIKE substring fallback —
which reads only the canonical messages table — instead of raising, so
CJK search degrades gracefully rather than crashing.
Adds two regression tests: trigram search self-heals in place after
shadow-table corruption (answers from the rebuilt trigram index, not the
LIKE fallback), and degrades to LIKE without raising when the one-shot
rebuild was already consumed.
Follow-up to #66420; refs #66296#66724
* fix(state): add REINDEX strategy to repair stale B-tree indexes (#63386)
When PRAGMA integrity_check reports 'wrong # of entries in index' for
B-tree indexes (e.g. idx_sessions_handoff_state), the existing repair
strategies (FTS rebuild, sqlite_master dedup, drop-FTS+VACUUM) don't
address the mismatch. Add Strategy 0.5: run REINDEX to rewrite the
index b-tree from canonical table rows before escalating to more
destructive strategies.
* test(state): exercise REINDEX repair against a REAL stale B-tree index
Replace the mocked test for #63398's REINDEX strategy: the original
monkeypatched _db_opens_cleanly to return the corruption string, so the
REINDEX pass itself was never exercised against actual index corruption —
the test would pass even if REINDEX didn't fix anything.
New fixture _corrupt_btree_index() builds genuine on-disk staleness with a
writable_schema hack: rewrite the index definition to a partial index
(WHERE 0), REINDEX so the b-tree is rebuilt empty, then restore the full
definition. integrity_check then reports the real
'wrong # of entries in index idx_messages_session' / 'row N missing from
index' class from #63386 — no mocks anywhere.
The rewritten test asserts end-to-end with real function calls:
- the real _db_opens_cleanly detects the stale index,
- repair_state_db_schema repairs it with strategy 'reindex_btree',
- post-repair the detector and raw PRAGMA integrity_check both report
healthy, and a query forced through the rebuilt index (INDEXED BY) sees
every row.
Adds a second test asserting the REINDEX strategy is non-destructive
(all sessions/messages survive, readable via SessionDB).
Follow-up to #63398; refs #63386
* fix(kanban): auto-repair index-only kanban.db corruption via REINDEX
_guard_existing_db_is_healthy previously failed closed on ANY
integrity_check failure, including the index-scoped class ('wrong # of
entries in index <name>' / 'row N missing from index <name>') where the
table b-trees are intact and REINDEX rebuilds the damaged indexes
losslessly. Boards hit by that class were bricked until manual surgery
even though SQLite can fix them in-place.
Now, when integrity_check output consists ONLY of index-scoped errors
(index name parsed generically from the message — no hardcoded list):
1. quarantine the corrupt bytes FIRST via the existing content-
addressed _backup_corrupt_db,
2. under the caller-held cross-process init flock, REINDEX each named
index (falling back to bare REINDEX if a parsed name doesn't
resolve),
3. re-run integrity_check and proceed only if it comes back clean.
Any non-index error class (page corruption, malformed image, freelist
damage) — or a REINDEX whose re-check is still dirty — fails closed
exactly as before: backup + KanbanDbCorruptError, no silent recreation.
Transient OperationalError (locked/busy) still propagates raw with no
quarantine.
Tests build a real board DB and corrupt a live index via the
writable_schema/partial-index REINDEX trick to produce the genuine
'wrong # of entries in index' shape, then assert auto-repair recovers
with data intact, page corruption still raises, and a dirty re-check
fails closed.
* fix(kanban): cap corrupt-backup retention at 10 files per board DB
Content-addressed quarantine backups dedupe identical corrupt bytes,
but corruption that keeps mutating between failures (partial repairs,
further damage across dispatcher retries, multi-profile fleets) mints a
new sha-named backup every round — a user accumulated 124
.corrupt.*.bak files with no bound.
After each NEW backup is created, prune oldest-by-mtime backups beyond
_CORRUPT_BACKUP_RETENTION (module constant, default 10), including the
copied -wal/-shm sidecars. The just-created backup is always exempt
(copy2 preserves the source mtime, which can be older than existing
backups). Pruning is best-effort and never masks the corruption error
about to be raised; dedupe of identical corrupt bytes is unchanged.
* feat(kanban): periodic WAL checkpoint (TRUNCATE) on the dispatcher tick
Kanban connections set wal_autocheckpoint=100, but SQLite's passive
autocheckpoint backs off whenever any reader holds an open snapshot —
on a busy multi-process board the -wal file can grow without bound
between gateway restarts.
After each successful dispatch tick, while still holding the board's
single-writer dispatch flock, run PRAGMA wal_checkpoint(TRUNCATE)
best-effort at a coarse interval (>=5 min since this process last
checkpointed that board; module-level per-path monotonic timestamp, so
multi-board dispatchers checkpoint each board on its own clock).
Success and busy/locked skips are both logged at DEBUG; a failing
checkpoint can never fail the tick.
* feat(kanban): add `hermes kanban repair` CLI verb
Adds kanban_db.repair_db() — a structured, non-raising wrapper around
the same narrow repair policy as the connect-time guard: probe with
PRAGMA integrity_check under the board's cross-process init flock;
quarantine the corrupt bytes FIRST via the content-addressed backup;
REINDEX only when every integrity message is index-scoped; re-check;
report ok / repaired / corrupt / missing. Locked/busy OperationalError
still propagates raw (a locked healthy DB is not corruption and gets
no quarantine), and a repair invalidates the per-process healthy-path
cache so the next connect() re-probes.
The CLI verb reports status human-readably (or --json), exits 0 for
ok/repaired/missing and 1 when the DB is still corrupt (non-index
corruption stays fail-closed with manual-recovery guidance). It
dispatches BEFORE kanban_command's auto-init: init_db() raises
KanbanDbCorruptError on a corrupt board, which previously would have
made a repair verb unreachable on exactly the boards that need it.
CLI tests drive the real argparse surface (build_parser +
kanban_command) against real corrupted SQLite fixtures.
* fix(packaging): graft web_dist in MANIFEST.in and add sdist regression test
Wheels ship hermes_cli/web_dist via pyproject package-data, but the sdist
did not: MANIFEST.in had no graft and .gitignore excludes web_dist, so
source tarballs installed a dashboard-less package. Graft the directory
and add an sdist regression test that builds the tarball and asserts
index.html is inside.
Salvaged from #29661; the PR's [web]-extra 404-message change was dropped
per maintainer review (misleading guidance for source installs).
* fix(dashboard): attempt one recovery build when --skip-build finds no dist
--skip-build with a missing web_dist/index.html previously hard-failed
with sys.exit(1) (issue #59288). The desktop launcher passes
--build-mode skip on every boot, so a wiped or never-populated dist
bricked the dashboard until the user manually rebuilt.
Now the default-dist path logs a clear warning and attempts exactly ONE
recovery build through the existing _build_web_ui path. If the recovery
build also fails to produce index.html, the original fatal behavior is
preserved with a clear message. A custom HERMES_WEB_DIST stays fail-fast:
the build writes to the default dist location and cannot populate a
caller-managed directory.
Closes#59288
* fix(web): guard _serve_index against a missing or unreadable index.html
mount_spa degrades to a JSON 404 catch-all when the dist directory is
fully missing, but _serve_index read index.html unguarded — so a dist
dir that exists while index.html is missing (partial build, wiped dist,
permissions) raised FileNotFoundError on EVERY request instead of
returning a useful error.
Catch OSError around the read and return the same JSON 404 payload the
fully-missing-dist path uses, so clients get a consistent signal. The
route recovers automatically once a rebuild restores the file.
* fix(dashboard): content-hash web UI freshness check
Replace mtime-based web UI dist staleness with a content-hash stamp under HERMES_HOME, matching the desktop build freshness model.
This avoids false stale/fresh decisions when git operations rewrite source mtimes without changing bytes, while preserving the existing stale-dist fallback. Also treats malformed or missing stamp data as needing a rebuild.
* fix(cli): serialize concurrent web-UI builds with an exclusive flock
Concurrent dashboard boots (desktop retry loop) each spawned their own
npm install + vite build over the same tree; parallel builds starved
each other, the dist sentinel never advanced, and every boot re-triggered
the build — cascading into orphan backends, port collisions, and CPU
storms that also knocked out the Telegram gateway's heartbeat (2026-07-12).
One process now builds under flock; the rest serve the existing dist
(stale is acceptable) or wait for the first-ever build to finish.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* refactor(cli): run the web-UI staleness walk once, under the build lock
The flock wrapper checked _web_ui_build_needed twice (pre-lock fast path
and post-lock re-check) and _do_build_web_ui checks it again internally,
so a boot that actually built walked the whole web/ source tree three
times. The callee's own check already runs under the lock on every path
through the wrapper, so it alone is sufficient: drop both wrapper checks.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* chore(git): ignore the web UI build lock file
The cross-process build flock (.web_ui_build.lock at the repo root) is
an empty coordination file created on first dashboard boot; it must
never be tracked or show up as untracked noise. Also keeps it out of
the content-hash staleness digest, which skips gitignored paths.
* test(cli): cover the web UI build flock contention paths
PR #63455 shipped the flock without tests. Cover the three contention
paths: contended+dist serves stale without a second build, uncontended
builds under the lock (creating the lock file), and contended+no-dist
blocks then skips the rebuild because the staleness re-check runs under
the lock and sees the winner's stamp. Also pin the lock filename into
.gitignore via a regression assertion.
* style(cli): open the build lock file with explicit encoding (ruff PLW1514)
* chore(contributors): map eazye19@users.noreply.github.com -> eazye19 (PR #42387 salvage)
* fix(mcp): reconnect immediately on transport TaskGroup drop instead of backoff/park
Streamable-HTTP / SSE MCP transports run their stream pump inside an anyio
TaskGroup. A transient stream drop (idle timeout, brief backend blip,
server-side TCP close) surfaces as a BaseExceptionGroup escaping the
transport context manager. It reached run()'s error path, which applied
exponential backoff (1s..16s) and eventually parked the server for 300s and
deregistered its tools — turning a sub-second glitch into a multi-minute
tool outage even though the POST path was still healthy.
_run_http now wraps all three transport branches (SSE, new + deprecated
Streamable-HTTP) and routes a BaseExceptionGroup through a new
_reconnect_or_reraise_group() helper that returns 'reconnect' (immediate
rebuild, no backoff/park/deregister). It re-raises instead of masking when
shutdown is in progress, when the group carries a real CancelledError
(cancellation must propagate, cf. #9930), or when no live session was
established this attempt (a connect/handshake failure that should fall
through to run()'s backoff rather than hot-loop).
Fixes#66092
Co-authored-by: 雨哥 <hanyu1212@users.noreply.github.com>
* fix(mcp): charge immediate reconnects against a rapid-drop budget (#62212)
Harden the immediate-reconnect path from #66271:
- _reconnect_or_reraise_group re-raises when the transport TaskGroup
carries a KeyboardInterrupt / SystemExit leaf — fatal signals must
propagate to the interpreter, never be converted into a reconnect.
- Rapid-drop budget: a completed handshake alone no longer clears
_reconnect_retries/backoff on clean transport return. A session is
UNPROVEN until it demonstrates real health — survived >=1 full
keepalive interval (keepalive success path) or served >=1 successful
tool call (_mark_session_proven). Unproven clean returns are charged
against _MAX_RECONNECT_RETRIES, so a flapping transport that
handshakes fine and drops moments later still reaches the park
instead of respawning forever (#62212: 6212 spawns in 63h).
Proven sessions keep the #57604 behaviour: budget clears, transient
blips over a long-lived session never accumulate toward parking.
* fix(mcp): unwrap exception groups and park permanent failures immediately (#65673)
- Add module-level _unwrap_exception_group (ported from
hermes_cli/mcp_config.py, adapted): handles nested groups, prefers
non-cancellation leaves over the CancelledErrors anyio sprays across
sibling tasks, and re-raises KeyboardInterrupt/SystemExit leaves.
- Add _classify_mcp_failure(exc) -> 'permanent'|'transient'.
Permanent: auth 401/403, NonMcpEndpointError, InvalidMcpUrlError,
FileNotFoundError/ENOENT on the stdio command. Transient: everything
else (network/EOF/ClosedResource/TaskGroup drops). Permanent failures
park immediately without burning the retry ladder — every retry
against a missing binary or a revoked credential hits the same wall.
- Every log site in run() and the keepalive failure log now logs the
unwrapped root cause as 'TypeName: message', so a dead stdio pipe
says 'BrokenPipeError' instead of the opaque
'unhandled errors in a TaskGroup (1 sub-exception)' (and empty
str(exc) dead-pipe errors are no longer blank).
* fix(mcp): one WARNING per state transition, DEBUG retry chatter, jittered backoff
Log-storm hygiene for the reconnect machinery (#65673, #66092):
- State transitions carry exactly one WARNING each:
connected → degraded (keepalive failure), degraded → parked
(budget exhausted / rapid-drop park / permanent error),
parked → connected (revival proven healthy, via
_mark_session_proven).
- Per-attempt retry logs (initial-connect attempts, connection-lost
reconnect attempts, per-cycle rebuild notices, park-wake probes)
demoted to DEBUG.
- Backoff sleeps get +/-20% uniform jitter (_jittered) so a herd of
servers that lost the same backend doesn't retry — and log — in
lockstep.
* fix(mcp): re-register tools during parked revival
* fix(mcp): isolate a single failing stdio server from the bridge (#50394)
* fix(mcp): clear connect-cooldown state on every shutdown path
Builds on trevorgordon981's #50589 (cherry-picked as the previous commit).
The #50394 cooldown reset only ran inside the async _shutdown coroutine,
which is skipped on the empty-_servers fast path — the most common state
when a server failed to connect (failed servers are never recorded in
_servers). It was also skipped when the MCP loop wasn't running.
Clear _server_connect_retry_after/_server_connect_failures on the fast
path and in a final unconditional sweep so a full shutdown/restart always
re-attempts every configured server immediately. Adds regression tests
for both paths.
* chore(contributors): map diffen77 + fazerluga-creator emails
For the #66547 and #66981 salvage cherry-picks in this branch.
* fix(mcp): reconnect message-less closed transports
* test(mcp): isolate resource error breaker state
* fix(mcp): harden nested interruption detection
* fix(mcp): make transport classification cycle safe
* fix(mcp): cycle-guard cause/context traversal in transport classifier
Builds on diffen77's #66547 (cherry-picked as the previous commits).
Extend _is_session_expired_error's iterative traversal to follow
__cause__/__context__ in addition to ExceptionGroup .exceptions — SDK
wrappers often raise a generic RuntimeError *from* the message-less
ClosedResourceError, leaving the transport signal reachable only via
the chain. The identity-visited set guards chain cycles (handlers
re-raising previously seen exceptions), and a bounded node budget
(_EXC_TRAVERSAL_MAX_NODES) caps pathological acyclic graphs.
Adds regression tests: cause/context chain detection, interruption
precedence through chains, cyclic cause/context termination, and
budget-bounded termination.
* fix(mcp): allow background discovery retry after a run that connected nothing
start_background_mcp_discovery() sets _mcp_discovery_started once and never
resets it. If the first background run exits without connecting any MCP
server (startup cancellation, OOM restart, transient network failure), every
later call returns immediately and the process is permanently stuck with
zero MCP tools until a full restart.
Fix: when discovery is marked started but the thread is dead and no server
is connected, reset the flag and spawn a fresh discovery thread. Also log a
WARNING when a discovery run completes with zero connected servers, so the
condition is visible instead of silent.
Caught in production on a long-running gateway fleet where a gateway
restarted under memory pressure and came back with all MCP tools missing.
* fix(tui): centralize stdio TUI MCP discovery on the shared owner
Review follow-up: the stdio hermes --tui path spawned its own one-shot
discovery thread, so the retry-after-zero-connected semantics added to
start_background_mcp_discovery() did not cover it.
Spawn TUI discovery through the shared owner and make the entry-side
wait_for_mcp_discovery() fall through to the shared owner when no local
thread exists (mcp_discovery_in_flight/join_mcp_discovery already consult
both owners). Keeps the cheap no-mcp-servers config guard on the TUI path.
Adds regression tests for the entry-side wait delegation.
* fix(tui): give the stdio TUI discovery a retry-after-zero-connected path
Builds on fazerluga-creator's #66981 (cherry-picked as the previous two
commits). start_background_mcp_discovery()'s retry allowance only fires
when the function is CALLED again, but tui_gateway/entry.py main() calls
it exactly once at startup — so a first discovery run that connected
nothing still latched the stdio TUI MCP-less for the whole session.
Re-invoke the idempotent spawn from wait_for_mcp_discovery() (the
per-agent-build wait) when the process is MCP-enabled, gated on a flag
set in main() so non-MCP sessions never pay the MCP import on the wait
path. Adds regression tests for both the retry re-invocation and the
non-MCP skip.
* fix(tui): gate the shared-owner MCP discovery wait on the stdio TUI flag
The TUI retry-allowance follow-up made tui_gateway.entry.wait_for_mcp_discovery
delegate to hermes_cli.mcp_startup unconditionally when no entry-local thread
exists. But server._make_agent already calls the startup wait directly for
dashboard /api/ws sessions, so every non-stdio agent build paid the bounded
wait twice (caught by test_make_agent_waits_for_shared_mcp_discovery). Gate
the retry-spawn AND the delegated wait on _mcp_discovery_enabled, which only
the stdio TUI arms in main().
* fix(status-bar): address Copilot review on /battery
- TUI /battery matches the CLI surface: adds `status` (live reading via
system.battery), and the help/usage strings now consistently read
[on|off|status].
- batteryLabel() renders `--` for an unknown percent so a null can never
surface as "null%" even without the showBattery guard.
- Move the system.battery RPC out of the config section into
"Methods: tools & system" where system.* RPCs belong.
* fix(dashboard): cap gateway health probe timeout
* fix(gateway): report runtime source version in /health, not stale dist-info metadata
/health and /health/detailed resolve the version via importlib.metadata
first, falling back to hermes_cli.__version__. On editable/source
checkouts — including the standard git-based install that hermes-setup
performs — hermes_agent-*.dist-info can survive a source update
unchanged, so the health endpoints keep reporting the previous release
even though the running code (CLI, dashboard, release tags) is newer.
Stale metadata does not raise, so the source fallback never fires.
Flip the preference: use hermes_cli.__version__ (the runtime source of
truth shared by the CLI and dashboard) first, and fall back to
distribution metadata only when the source import fails. The
never-raise contract of the version probe is unchanged.
Observed live: CLI, dashboard, and pyproject all reported 0.18.2 while
/health returned 0.18.0 from a stale hermes_agent-0.18.0.dist-info left
behind by a source update.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(api-server): count "stopping" runs as active in readiness work counts
_readiness_work_counts()'s active_api_runs set is {"queued", "running",
"waiting_for_approval"} — it excludes "stopping", the status
_handle_stop_run() sets while a run is being interrupted. Since the stop
is fully cooperative (the run stays "stopping" — doing real
executor-thread work — until the agent actually notices the interrupt and
the task settles to "cancelled", an unbounded window, not a fixed
timeout), /health/detailed's background_queues.active_api_runs
undercounts real active work for that whole duration.
Fix: add "stopping" to the active-status set. background_queues.status
itself is hardcoded "ok" (gateway/readiness.py), so this doesn't change
overall readiness — it only corrects the count value external monitoring
tooling reads from this endpoint.
* chore(contributors): map yingwaizhiying@gmail.com -> tianma-if (supersedes stale msh01 entry)
* feat(dashboard): component-level health rollup on /api/status (#68662)
The dashboard's own liveness surface could report healthy while every
authenticated request 500'd (e.g. wedged state DB) — /api/status carried
gateway-only fields, no storage/dashboard signal, and no middleware
counted unhandled exceptions.
- DashboardHealth state holder: rolling 5-min deque of unhandled-error
timestamps + last self-test result. last_error_type/last_error_path
are internal-only diagnostics — snapshot() exports counts/enums/
timestamps exclusively (PUBLIC_API_PATHS no-secrets contract).
- Outermost @app.middleware('http') (registered last) wraps call_next
in try/except: records + re-raises unhandled exceptions, and records
responses with status >= 500.
- /api/status gains 'components' {gateway, storage, dashboard,
platforms} + top-level 'overall' ok|degraded. storage reuses the
gateway readiness state_db probe (read-only, 1s-bounded) in an
executor; platforms derive ok/degraded from existing
gateway_platforms states.
- Authenticated self-test task started in the lifespan: every 60s an
in-process httpx ASGITransport GET of /api/sessions?limit=1 with the
real _SESSION_TOKEN, feeding the dashboard component. Skips cleanly
when httpx is unavailable and while the OAuth gate is engaged (the
legacy token is not honoured there).
Tests: middleware increments on raising route and on 5xx, window
expiry, components shape + overall, storage degraded when the state_db
probe fails, dashboard degraded after an error, no secret-bearing
fields in the public payload, self-test pass/fail recording (mocked
client) + a real ASGI round trip.
* fix(status): make gateway_updated_at a stable RFC3339-or-null contract (#68657)
The /api/status gateway_updated_at field and the gateway /health/detailed
updated_at field passed through whatever gateway_state.json contained,
untyped. All current writers emit RFC3339 via _utc_now_iso(), but legacy
gateways wrote unix epoch floats, and a corrupt or hand-edited state file
can inject numbers or arbitrary garbage — while the frontend types
(web/src/lib/api.ts) declare string | null.
Add normalize_updated_at() in gateway/status.py as the single funnel:
- str: accepted iff datetime.fromisoformat parses (trailing Z tolerated);
naive timestamps coerced to UTC; canonical isoformat returned
- int/float: treated as unix epoch seconds -> UTC ISO string, with a
plausibility guard (reject < 2000-01-01, > now+1day, non-finite)
- bool: rejected explicitly (int subclass, but never a timestamp)
- anything else: None
Apply it at both emit sites: the dashboard /api/status handler (covers
both the local read_runtime_status() branch and the remote
/health/detailed cross-container fallback branch) and the gateway API
server's /health/detailed response.
Contract tests: parametrized /api/status normalization (epoch float/int,
garbage string, None, bool, dict, absent key), remote-health numeric and
garbage bodies, dashboard shape test round-trip assertion, direct
normalize_updated_at units (range guards, Z suffix, naive coercion,
non-finite floats), and a write_runtime_status -> read_runtime_status
round-trip proving the writer side stays tz-aware parseable.
* feat(desktop): type-to-focus the composer from empty chat chrome
Printable keys and soft `/`/Enter pull focus back to the composer when
nothing else owns the key (dialogs, terminal, buttons on Enter, …).
* feat(ui-tui): widget-grid 2-axis layout engine + overlay primitives (#20379 rebase)
Rebase of the widget-grid PR onto current main, then grow it into a real
2-axis engine: resolveGridTracks (grid-template tracks — fixed cell counts
and weighted fr shares with mins), layoutWidgetGrid (1D auto-packing flow),
and layoutGridAreas (2D absolute placement with row/col spans and implicit
row growth). Overlay/Dialog primitives give zoned viewport-level modals
with an optional scrim. /grid-test (interactive: areas, nesting, zoom,
gap/padding) and /grid-test streams (4x3 mission-control GridAreas board
with promote-to-main) exercise everything end to end.
* refactor(ui-tui): route production surfaces through the widget-grid engine
Banner (responsive tiers: full logo -> compact rule -> text -> hidden),
SessionPanel (fixed hero track + flexible info track, the desktop pane
shell's fixed-vs-flex contract), floating overlays, prompt zone, and the
pickers all render through WidgetGrid instead of hand-rolled flex math.
Pickers gain maxWidth so grid cells can cap them.
* feat(ui-tui): background-aware theme adaptation + paired palettes + /theme pin
OSC-11 asks the terminal for its actual background at startup (env
heuristics are blind on xterm.js hosts) and the theme re-derives against
the answer: desktop-contract adaptation (contrast floors + fill polarity),
a shared list-row selection primitive instead of per-picker panel fills,
paired light_colors/dark_colors skin blocks with a machine audit, and a
/theme auto|light|dark pin (display.tui_theme) for hosts whose probe lies.
E2E coverage for the OSC reply chain + /theme-info diagnostics.
* feat(ui-tui): theme engine — seeds to derived-tone ladder + flash-free boot
The desktop color-mix system, ported: a theme is a handful of identity
SEEDS (text, primary, accent, border, status hues); every secondary tone
(muted, label, surfaces, chips, selection) is a color-mix derivative
against the real terminal background. lib/color.ts consolidates the
primitives (parse/mix/luminance/contrast/retone + xterm.js's multiplicative
liftForContrast). Knobs are grid-search fitted so the math reproduces the
classic hand-tuned literals (contract-tested). The display shim renders
authored palettes RAW and only rescues near-invisible colors. Boot reads
the last resolved theme from $HERMES_HOME/tui-theme-boot.json so the first
frame paints in the right palette (no default-dark flash), and the
placeholder cursor follows the bubbles textinput pattern.
* perf(tui): skin switches stay hot — MCP gating, pooled reload, swap repaint
Four responsiveness/hardening fixes surfaced by rapid /skin switching:
config.get mtime now carries an mcp_rev hash so the TUI reloads MCP only
when MCP-relevant config changed (cosmetic /skin writes cost seconds of
reconnects before); reload.mcp runs on the RPC pool serialized by a lock
(inline it froze the stdio reader for the duration of a flapping server's
retry loop — config.set/complete.slash sat unread and the TUI appeared
dead); theme swaps schedule one full repaint (incremental diffs after a
recolor tear — stale cells keep the old palette, read as "shadows"); and
OSC-11 pure-black answers are distrusted universally (unset-default
fingerprint on xterm.js hosts and tmux). Parent EIO zombie fixed: a dead
PTY made every render write throw once a second forever; exit after 5
consecutive dead-stream errors.
* fix(ui-tui): light mode renders the vivid palette RAW, not WCAG-darkened
Pixel-sampled against the reference screenshots: the beloved classic
light-mode look is the vivid authored golds rendered essentially raw
(#FFD700 shows as bright #F5C242) — transparent terminal profiles apply no
contrast lift of their own, and pre-darkening every foreground to WCAG 4.5
produced the reported mustard mud. Light-mode display floors become
near-invisible rescues only (1.18 display / 1.6 semantic; dark keeps
1.45/2.2); the default skin ships a fills-only light_colors OVERLAY
(polarity-flip the navy menu/status fills, foregrounds inherit the vivid
colors), and themeForSkin merges polarity overlays instead of replacing.
Palette audit reworked: base colors audited fully, overlays for valid keys
and fill polarity.
* fix(ui-tui): eradicate every transparent-terminal black-slab trigger
On terminal.background #00000000 xterm paints "drawn blank" and
attribute-styled cells against an opaque black RGB the user never sees
elsewhere. Verified by PTY byte capture + stateful SGR replay that we emit
no black backgrounds — then removed every trigger: the banner's opaque
space-fills, the scrollbar's non-scrollable space column and SGR-dim
track, the placeholder's SGR inverse cursor and dim fallback, and the bold
full-width banner rule. Chrome styling is now explicit truecolor only:
scrollbar thumb rides primary (accent on hover/drag) over a blended track,
the placeholder cursor is a theme-colored chip (48;2 bg + luminance-picked
ink), hints always carry an explicit 38;2 foreground.
* fix(ui-tui): session-panel hierarchy + polarity-proof placeholder tone
Tool/skill rows rendered labels in muted and member lists in text — which
inverts per skin (muted is the strong family tone on gold skins but the
weak gray on blue ones; slate/poseidon read backwards). Labels now lead in
the theme's label tone, values recede via an explicit fade toward the
surface; audited across all 9 built-ins x both poles. The composer
placeholder lands on muted — the "(and N more toolsets…)" tone — a
mid-luminance family color that reads receded on both poles even when
polarity detection is wrong.
* feat(ui-tui): OSC-10 foreground polarity tiebreaker for transparent terminals
Transparent profiles make OSC-11 useless (xterm reports the unset default,
pure black, regardless of the composited surface) so polarity detection
lagged editor theme flips. OSC-10 reports the theme's REAL foreground on
those hosts — its luminance reveals the pole. hermes-ink grows a foreground
slot (shared reportedColorSlot factory), App.tsx queries both in the
startup batch (background first so a trusted answer wins without churn),
and the app commits an inferred pole only when the background was
distrusted AND the foreground is decisive (bright=dark theme, dark=light;
mid-grays and #000/#fff defaults commit nothing). User pins still outrank.
* fix(ui-tui): session-panel fade anchors on muted, not the surface — readable on every pole
mix(text, surface) was invisible whenever text was already pale (the
light-rendered default: cream blended toward white = nothing) and inherited
wrong-polarity detection through the surface. mix(muted, text, .5) is
pole-proof: muted is mid-luminance by construction, so the midpoint stays
readable everywhere. Audited 9 skins x both poles: 2.0-2.9:1 on white,
5.6-9.3:1 on dark (worst case 1.92 for a light-authored skin on dark).
* style(skins): default light overlay = goldenrod ladder, not neon or mustard
On white, the vivid #FFD700/#FFBF00 read as glare and the WCAG-darkened
mustard reads as mud. The sweet spot is the statusbar's goldenrod family
(hue kept, saturation tamed, mid luminance): title #C8961E, headers
#D89B04, labels #A97E10, muted #B8860B unchanged, warm bronze ink body
#5C4718, deepened semantics + shell blue. Hierarchy on white: ink 8.9:1 >
fade 5.2 > label 3.7 > muted 3.3 > title 2.7 > headers 2.4. Dark mode
renders the vivid block untouched (explicitly approved as-is).
* refactor(ui-tui): DRY the chrome primitives — shared scrollbarColors + hintRgb
scrollbarColors(t, hover, grabbed) in overlayPrimitives is now THE scheme
for both scrollbars (was duplicated formulas); textInput's hex parsing
collapses into one hintRgb helper feeding colorizeHint and hintCursorCell.
Comment bloat trimmed. No behavior change — 1243 tests byte-green.
* fix(ui-tui): completions popover — aligned name track + neutral descriptions
Two-column grid: the name track auto-sizes to the widest visible command so
descriptions align in their own column instead of running under the names.
Descriptions render in the neutral statusFg gray — label and muted are
near-twins on the gold skins, which made command and description read as
one unparseable run.
* refactor(ui-tui): every selectable list rides the shared selection chip — zero inverse left
/agents, /journey, model picker, skills hub, plugins hub, pet picker, and
the approval/confirm prompts all used SGR inverse for the active row —
terminal-interpreted against unknowable defaults (black slab on transparent
profiles) and visually divergent from completions/session-switcher. New
spreadable chipRowProps(t, active) in overlayPrimitives (chip bg + lifted
ink + bold; spread after `color` so chip ink wins) converts each site to
the one selection treatment. rg confirms zero inverse={} remaining in
components/.
* fix(ui-tui): overlay width caps are absolute — clampOverlayWidth (Copilot review)
Every picker/hub forced width >= 24 AFTER applying the caller's maxWidth,
so a grid cell narrower than 24 (or a FloatBox cell under 28 with its 4
cols of chrome) overflowed and clipped at the terminal edge. One shared
clampOverlayWidth(preferred, maxWidth, min=24): the caller's cap is
ABSOLUTE (a cell knows its budget), the usability floor applies only when
the cap allows it, uncapped keeps the old floor semantics. Five call
sites (model/pet pickers, skills/plugins hubs, session switcher) route
through it; the grid-test FloatBox drops its own 24 floor. Contract-tested
including the sub-floor cap case from the review.
* fix(tui): reload.mcp lock scope + coalesced refresh + macOS polarity flip (Bugbot)
Three real findings from the review of the reload.mcp pooling + OSC-10 work:
- Lock released too early: the leader now holds _mcp_reload_lock across
shutdown+discover AND its own agent refresh — releasing after discover let
a second reload tear the registry down while the first was still reading it
to rebuild the session's tool snapshot.
- Coalesced reload skipped the agent refresh: a follower returned
"reloaded" without rebuilding ITS OWN session's snapshot, so a coalesced
session kept stale tools. Followers now wait, then refresh their agent
against the freshly-built registry (under the lock, skipping the redundant
shutdown/discover). Shared _finish_reload tail for the `always` opt-out.
- macOS AppleInterfaceStyle fallback never set `resolved`, so a late OSC-10
foreground reply could re-flip the committed inference (visible churn). It
now marks resolved; a real OSC-11 background measurement still corrects it
(that listener intentionally doesn't gate on resolved — measurement beats
inference).
Bugbot's other four findings were diff-truncation false positives (color.ts,
themeBoot.ts, and the mcp_rev/light_colors/dark_colors types all exist;
typecheck + build green). 384 tui_gateway + TS suites pass.
* fix(tui): mcp_rev includes mcp_servers; reload survives leader failure; /theme persists first (Bugbot r2)
- mcp_rev hash now covers `mcp_servers` (the server DEFINITIONS the classic
CLI watches) alongside `mcp`/`tools` — editing a server previously bumped
mtime but not mcp_rev, so the TUI skipped reload.mcp and new servers never
connected until a manual /reload-mcp.
- reload.mcp coalescing survives a failed leader: a completed-generation
counter (bumped only after a successful shutdown+discover) gates the
follower. If the leader threw (flapping server) the follower re-runs the
full reload itself instead of returning a bogus success over an empty
registry.
- /theme applies AFTER config.set confirms (mirrors /indicator) with a
guardedErr catch — a failed persist no longer leaves the session showing a
theme that reverts on restart.
384 tui_gateway + 1246 TS tests, lint, build green.
* fix(ui-tui): first-swap repaint + config-auto respects shell theme pin (Bugbot r3)
- commitTheme compares the first commit against the SEED theme uiStore
mounted with (boot cache/default), not null — a cold start that resolves
to a skin differing from the default previously skipped the anti-tearing
forceRedraw on that first swap.
- applyConfiguredTuiTheme('auto') now clears only a pin CONFIG set (tracked
via configPinnedTheme), never a HERMES_TUI_THEME the user exported in their
shell — that env override outranks auto-detection per detectLightMode's
documented priority, and a config hydrate was wiping it.
(Bugbot's recurring "GatewaySkin missing paired palette fields" is a
diff-truncation false positive — gatewayTypes.ts declares light_colors/
dark_colors, typecheck green.) 81 handler + build/lint green.
* fix: restore package-lock.json @esbuild platform entries (Bugbot r4)
The branch had stripped every node_modules/@esbuild/* optional-platform
entry from the lockfile (442 deletions, 0 additions) — collateral from an
earlier local @esbuild/darwin-arm64 reinstall that rewrote the lockfile to
this machine's platform only. esbuild still declares them as
optionalDependencies, so `npm ci` on Linux CI would skip the platform
binary and break Vitest / ui-tui builds. No dependency was added on this
branch (the only package.json delta is a dev `visual` script), so the
lockfile is restored to match main exactly.
* fix(ui-tui): seed mcp_rev baseline at startup so first cosmetic write doesn't reload MCP (Bugbot r5)
The startup config.get seeded mtimeRef but not mcpRevRef; after a normal
boot mtime is already non-zero so the poller's baseline branch never runs,
leaving mcpRevRef empty. The first /skin (or other cosmetic) write bumped
mtime with an unchanged mcp_rev, and empty !== hash tripped a full
reload.mcp — exactly the reconnect this optimization removes. Seed the
revision alongside mtime at boot.
* fix(ui-tui): record config theme pin even when env already matches (Bugbot r6)
Regression from the r3 shell-pin fix: applyConfiguredTuiTheme('light'|'dark')
returned early when HERMES_TUI_THEME already matched, leaving
configPinnedTheme false — so a later '/theme auto' refused to clear the pin
and the session stayed forced to the old mode while config said auto. Set
configPinnedTheme before the match short-circuit.
Bugbot's other three r6 findings (GatewaySkin paired-palette fields,
ConfigMtimeResponse.mcp_rev, picker maxWidth props) are diff-truncation
false positives — all declared in gatewayTypes.ts / the picker prop
interfaces; typecheck is green.
* feat(ui-tui): shimmer skeleton for the lazy tools section
The lazy-loaded Available Tools section rendered a BLANK gap that popped
when data landed. It now shows animated shimmer rows shaped like the real
content (label block + value run, diagonal band sweep), and the summary
line prints "… tools · … skills" instead of "0 tools · 0 skills" while
counts load.
components/loaders.tsx ships the primitives (shimmerSegments band math,
Shimmer, useShimmerPhase, ShimmerRows) — colors are caller-owned theme
tones, one interval per composition. Cherry-picked from the SDK-shape
branch so the skeleton lands with this PR; the widget-SDK consumers stay
in the follow-up.
* fmt(js): `npm run fix` on merge (#68938)
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
* fmt(js): `npm run fix` on merge (#68976)
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
* fmt(js): `npm run fix` on merge (#68977)
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
* test(mcp): stamp breaker-open time on the monotonic clock, not a literal (#69003)
test_session_expired_retry_waits_for_new_session hardcoded
_server_breaker_opened_at["hindsight"] = 123.0 to simulate a circuit breaker
whose cooldown has already elapsed. But the breaker in tools/mcp_tool.py
compares that stamp against time.monotonic() (age = monotonic() - opened_at,
elapsed when age >= _CIRCUIT_BREAKER_COOLDOWN_SEC). time.monotonic()'s origin
is arbitrary and small on a freshly-booted CI container, so age worked out to
only a few seconds there (< the 60s cooldown) — the breaker stayed open, the
half-open probe never fired, and the retry returned the "unreachable" error
instead of "bank ok". It passed on long-uptime dev boxes (large monotonic)
and failed under CI, with the reported "Auto-retry available in ~Ns" drifting
run to run as the container's monotonic clock varied.
Stamp opened_at relative to the same clock the code reads
(time.monotonic() - _CIRCUIT_BREAKER_COOLDOWN_SEC - 1.0) so the cooldown is
provably elapsed regardless of the monotonic origin, exercising the intended
half-open transition deterministically.
* fix(windows): share one bounded, tree-killing git probe across both call sites (#68997)
subprocess.run(["git", ...], timeout=...) deadlocks on Windows: run()'s
post-timeout cleanup calls an unbounded communicate() after killing git.
Killing the PATH-resolved launcher can leave a suspended descendant git.exe
holding duplicates of the captured stdout/stderr handles, so the pipes never
reach EOF and the reader-thread join blocks forever — leaking a process +
two reader threads per fired timeout (the accumulating git.exe load behind
Windows Defender CPU spikes).
Two fail-open probe call sites had this identical flaw:
- tui_gateway/git_probe.py::run_git — on the Desktop agent-build path
(_start_agent_build -> _session_info -> branch() -> run_git), where the
hang turned an optional branch label into "agent initialization timed
out" (#68609).
- agent/coding_context.py::_git — hangs the agent turn inside
build_coding_workspace_block under an ACP host (#66037).
Consolidate both onto one shared bounded_git_probe() in
hermes_cli/_subprocess_compat.py (both files already import from there, so
no new import surface):
- explicit communicate(timeout), then on ANY failure a tree-kill —
proc.kill() AND, on Windows, best-effort taskkill /T /F so the suspended
descendant that holds the pipe writers dies too — plus a bounded 1s
post-kill drain; if the pipes are still held they're abandoned (the
orphaned reader threads are daemonic and cost nothing).
- fail open to "" on every path: spawn error, timeout, kill() raising
(access denied / already reaped — a raise inside the except handler
previously escaped the contract), and non-timeout communicate() failures
now also terminate the child instead of leaving it running.
- the taskkill spawn can't re-enter the deadlock class: it captures no
pipes (DEVNULL), so its own timeout cleanup has no reader threads to join.
Normal-path spawn contract is preserved byte-for-byte: PIPE/PIPE/DEVNULL,
text + utf-8 errors="replace", hidden-window creationflags on Windows only,
nonzero returncode -> "". Each call site keeps its own timeout (1.5s / 2.5s).
Supersedes #68622 (Sora-bluesky — git_probe fix + tree-kill) and #66038
(iamwongeeeee — coding_context fix), folding both into one shared helper so
the two sites can't drift and every timeout tree-kills the descendant. Tests
consolidated onto the helper, incl. the previously-missing assertion that a
Windows timeout escalates to taskkill /T /F.
Co-authored-by: Sora-bluesky <sora.bluesky.dev@gmail.com>
Co-authored-by: iamwongeeeee <wykim777@naver.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* fix(tui): revision-aware reload.mcp — an ack now means the revision was LOADED
Review on #20379, finding 1 (High). Two ways an MCP config revision could
be silently acknowledged without ever being applied:
Client: the poll advanced its accepted mcp_rev BEFORE calling reload.mcp,
and quietRpc collapses failures to null — a reload that failed against a
temporarily broken server left the revision recorded as applied, and no
subsequent poll retried until an unrelated MCP edit. The handshake is now
syncMcpReload(): send the observed rev with the request, advance `accepted`
only when the server answers status=reloaded (to the server's loaded_rev,
falling back to the requested rev on older gateways), and re-compare on
EVERY poll tick — decoupled from mtime — so a transient failure heals on
the next tick. An in-flight guard stops the 5s poll from stacking requests
behind a slow reload.
Server: generation-only coalescing let a follower triggered by revision B
ack against revision A's registry when the config changed under a slow
leader. The leader now re-hashes the MCP-relevant config after discovery
and repeats until stable (bounded), records _mcp_reload_loaded_rev, and a
follower coalesces only when the revision it was asked to load matches —
otherwise it re-runs the full reload itself. Responses carry loaded_rev.
Deterministic tests for the exact failure sequences: failed reload → no
ack, no generation advance; A-then-B overlap → follower re-runs; matching
rev → coalesces; failed leader → follower re-runs; legacy no-rev callers
keep generation-only coalescing (thread ordering via an instrumented lock,
no sleeps). Client: 6 vitest cases on the ack/retry/in-flight contract.
* fix(ui-tui): boot theme cache gets provenance — a stale hint can't pin the terminal
Review on #20379, finding 2 (High). The boot cache seeded the previous
session's background into HERMES_TUI_BACKGROUND, the same slot a CURRENT
OSC-11 answer occupies — so a cache written on a light terminal pinned a
now-pure-black terminal to light forever: the new OSC-11 #000000 answer is
distrusted by design, the pure-white OSC-10 foreground is distrusted too,
and the macOS appearance fallback refuses to run while the slot is set.
Seeding now records provenance (themeBoot.seedBootEnvironment, extracted
and testable against a passed env). When the current terminal answers the
background probe with the untrusted fingerprint, the gateway handler calls
invalidateBootBackground(): the slot clears ONLY while it still holds the
seeded value (a trusted answer that overwrote it is authoritative), OSC-10
gets first claim in the same startup batch, and a short settle pass re-
derives from the live fallback chain if nothing answered.
The cache is also pin-coherent now: commitTheme persists the config mode
pin (display.tui_theme) alongside the resolved theme + physical background.
Previously "/theme light" on a dark terminal cached a light theme next to a
dark background — the next launch painted light, flipped dark when the skin
resolved against the seeded background, then flipped light again on config
hydration: the exact multi-stage flash the cache exists to eliminate. The
seeded pin counts as config-owned (bootSeededPin), so a later 'auto' can
still clear it instead of mistaking it for a user shell export.
Tests cover the review's sequences: stale-light cache vs current dark
terminal (invalidate → fallback chain), stale-dark vs ambiguous light,
pinned light on a dark physical background across restart (and the
inverse), trusted-overwrite protection, and the explicit-signal guards.
* fix(ui-tui): stacked dialog gets input before the grid under it
Review on #20379, finding 3 (Medium). /grid-test's `d` opens a dialog on
top without clearing the grid, but the grid's input branch ran FIRST — so
Esc/q/Enter mutated the hidden grid (close/unzoom/promote) instead of
closing the visible dialog, contradicting its "Esc/q/Enter close" hint.
Input routing now follows visual stacking: the dialog/grid dispatch is
extracted into handleStackedModalInput() with the dialog branch first, the
hook consumes through it, and tests drive the real dispatch against the
overlay store — each advertised close key closes only the dialog (grid
byte-identical), grid keys don't leak through while the dialog is up, and
the same keys route to the grid again after it closes.
* fix(ui-tui): make `npm run visual` portable off POSIX — zero new deps
Review on #20379, finding 4 (Medium). Three portability defects in the
visual verification harness:
- `FORCE_COLOR=3 COLORTERM=truecolor tsx ...` POSIX env assignment does not
work under the Windows npm command shell. The script is now a plain Node
launcher (scripts/visual/run.mjs) that sets the env itself and spawns tsx
via require.resolve('tsx/cli') — no cross-env, no shell syntax.
- Hardcoded /tmp/tui-visual.{html,png} resolve to a drive-root path like
C:\tmp on native Windows (and fail when that directory doesn't exist).
Both scripts now derive the output directory from a shared paths.mjs
helper: os.tmpdir()/hermes-tui-visual (created recursively;
HERMES_TUI_VISUAL_DIR overrides for CI or side-by-side runs).
- electron was undeclared by ui-tui and only worked via hoisting luck. The
launcher now resolves it EXPLICITLY from the install tree the desktop
workspace already provides (require('electron') in plain Node returns the
binary path), with an ELECTRON_BIN override and a clear error pointing at
the repo-root install when it's absent — instead of declaring a second
~100MB dependency on a TUI workspace for a dev-only harness.
Also fixes the trailing-whitespace line in render.tsx that `git diff
--check` flags. Verified end-to-end: render writes the HTML scene sheet
and the electron shot step produces the screenshot from the tmpdir path;
the missing-electron error path prints the guidance message.
* perf(ui-tui): shimmer loaders share one clock and stop after a bounded period
Review on #20379, finding 5 (Perf). Every ShimmerRows mounted its own 90 ms
setInterval — the session panel can show lazy skills AND lazy tools at
once, and a lazy watch session stays lazy indefinitely, so an otherwise-
idle TUI ran ~22 React state updates per second forever.
All shimmer compositions now subscribe to a single module-level clock: one
interval regardless of how many skeletons are on screen, updates delivered
in one timer callback so React batches them into a single render pass, and
the interval is torn down with the last subscriber. Each mount's animation
is also bounded (SHIMMER_ANIMATE_MS, 30 s): after the budget the skeleton
freezes in place — it still reads as "loading" — and stops costing renders
entirely.
Tests: fake-timer coverage that N subscribers share one timer in lockstep,
the interval stops with the last unsubscribe, and a late subscriber
restarts the clock cleanly.
* fmt(js): prettier pass on the new review tests
* chore: gitignore node_modules symlinks, not just directories
Worktrees symlink node_modules to the main checkout; the dir-only
node_modules/ pattern doesn't match symlinks, so one slipped into a
commit and broke npm ci on CI (ENOTDIR). Dropping the trailing slash
matches both.
* fix(desktop): stop long-session transcript from drifting to old turns (#69019)
content-visibility:auto on turn groups (perf: off-screen turns skip
style/layout/paint) pairs with contain-intrinsic-size:auto, which only
remembers a turn's size after it renders. A turn that finished streaming
near the bottom had its smaller mid-stream size remembered; once it
scrolled off the top edge and got skipped, it collapsed to that stale
height. With overflow-anchor:none the viewport can't self-correct, so the
stick-to-bottom lock drifts and the view creeps up over older turns — the
'long session eventually shows old responses' visual glitch.
Exempt the newest turns (live tail) from virtualization so a turn is only
ever skipped after its layout has settled at its final size (remembered ==
real -> skipping changes no height). Off-screen older turns still skip, so
the dialog/popover whole-document recalc win on long transcripts is kept
(it scales with the hundreds of old turns, not the small tail).
* ci: retrigger (transient setup-uv manifest fetch flake)
* feat(ui-tui): widget-app SDK — registry, host, dispatch; demos become apps
The SDK the desktop app already has, ported to the TUI: a WidgetApp contract
(id/help/mode/init/reduce/render/usage), a registry, and a host that owns the
active widget, routes input to its reducer, and renders it. The grid-test and
dialog-test debug surfaces are reimplemented as widget apps instead of bespoke
overlay state, and slash commands are generated from the registry. Input for an
open widget is owned by the active app (supersedes the demo-only stacked-modal
routing) — the single active widget enforces topmost-owns-input structurally.
* feat(ui-tui): weather reference app — the async-data contract, themed ASCII art
/weather [location]: wttr.in current conditions behind a Dialog, art bucket
table-driven off WWO weather codes, every tint a theme family tone (sun =
primary, rain = shell blue, thunder = warn). Proves the async story the
demos don't: init returns a loading phase and fires the fetch; results land
through the new host.updateWidget, which patches state ONLY while the app
is still active — a late resolution can never resurrect a closed app or
clobber a different one. `r` refetches; Esc/q/Enter close.
Four async-contract tests (loading→ready via updateWidget, late-resolution
guard, error phase, keymap). 1253 TS tests green.
* feat(ui-tui): ambient widget mode — registry-driven slash catalog + in-flow dock
Widgets can render as ambient (glanceable, non-blocking) instead of modal,
docked in the normal layout flow above/below the status bar rather than taking
over the screen. The slash catalog is generated from the widget registry so new
apps surface automatically, and /ticker lands as the first live-animation
ambient demo.
* feat(ui-tui): self-authored widgets — user-widget loader, hot-load, skill
Hermes can write its own widgets: a loader discovers $HERMES_HOME/tui-widgets/*.mjs,
fs.watch hot-loads them the moment they land (no restart), and a tui-widgets skill
teaches the agent the contract and the openWidget-at-register auto-open recipe.
Load/error/remove events announce themselves in the transcript; a lazy intro
skeleton covers the first paint.
* feat(ui-tui): widget primitives — charts, accordion, shimmer, stable streams
Reusable render primitives the SDK exposes to widget authors: sparkline/gauge/
hbars chart helpers (dimension-stable so live updates never resize the card),
an Accordion for expand/collapse sections, animated shimmer loaders, and a
streams demo that no longer reserves a phantom icon column on unfocused titles.
* docs(skill): tui-widgets — auto-open recipe (openWidget at end of register)
* feat(ui-tui): ambient zone system + widget crash boundary
A full placement grid so the agent can put a widget where it asks — dock-top/
bottom and corner zones, with corners as reserved rails that take real space
instead of floating over content. A per-widget error boundary plus lenient
ShimmerRows means generated widget code can't crash the TUI.
* refactor(ui-tui): host placement router + grid-test width-floor fix
host.tsx collapses to one placement router over a shared render context, and the
grid-test app drops its width floor too (carrying the #20379 review rule). Final
formatting pass folded in.
* feat(themes): cross-surface theme SDK — one skin themes CLI, TUI, and desktop
Make the Python skin engine the single source of truth for a canonical theme
shape consumed by every surface, so a skin authored in $HERMES_HOME/skins/*.yaml
(by a user or by Hermes from a prompt) themes the CLI, TUI, and desktop GUI at
once — the theme analogue of the plugin SDK.
- @hermes/shared: canonical `HermesSkin` token shape + `SKIN_COLOR_TOKENS` enum,
consumed by both TS surfaces (TUI `GatewaySkin` and desktop dedup onto it).
- Desktop: `skinToDesktopTheme` resolver (skin → CSS-var palette, VS Code-style
derive-from-seed) + `backend-sync` that registers backend skins into the theme
registry (Appearance/Cmd-K/`/skin`) and applies on a real change. Seeds on
gateway.ready (never stomps a persisted pick), applies on skin.changed and the
post-turn `config.get skin` poll (catch-all for agent-edited config.yaml).
- TUI: `fromSkin` now maps the status bar + `background` keys it was dropping.
- Gateway: `config.get skin` also returns the full resolved palette (additive).
- Skill: `hermes-themes` teaches the agent to author + activate a skin.
Each surface keeps its own normalizing resolver (ansi for the TUI, CSS vars for
the desktop, prompt_toolkit/Rich for the CLI).
* fix(themes): activate skins via `hermes config set`, never a config.yaml hand-edit
The skill told the agent to `patch` display.skin into config.yaml; a stray indent
corrupts the file and breaks the live gateway (the reported "/ menu broke"), and
a raw file edit never live-applies in a running CLI/TUI ("nothing happened").
Route activation through the safe writer (`hermes config set display.skin`), and
state plainly that a tool call can't hot-switch a running CLI/TUI — the user runs
`/skin <name>` (desktop still auto-repaints on the next turn).
* feat(themes): agent-authored skins switch live via a gateway skin watcher
A skin Hermes activates (`hermes config set display.skin X`) or recolors in
place now goes live on every surface (CLI, TUI, desktop) within ~half a
second, on its own — no `/skin`, no tool-hook timing, no user action.
A gateway daemon polls the resolved skin signature `(name, active-file mtime)`
every 0.5s and broadcasts `skin.changed` on any real move — a name switch OR a
live color edit to the active skin. It routes through the SAME path `/skin`
uses, so all surfaces repaint identically. The watcher seeds its baseline at
gateway.ready (stdio + ws) so it only fires on a real change; the `/skin` RPC
seeds the baseline too so it never double-broadcasts.
Subsumes the desktop's post-turn `config.get skin` poll (its skin.changed
handler already applies).
* feat(themes): TUI paints its own background from the skin (OSC 11)
The TUI inherited the terminal's background; now a skin's `background` paints the
whole surface via OSC 11 when a skin is applied, and clears back to the terminal
default (OSC 111) on revert and on exit (ridden in through resetTerminalModes).
Opt-in: a skin with no `background` leaves the terminal untouched, and the
restore only fires if we actually painted. Desktop already themed its own bg;
this closes the loop so Hermes owns its background on every surface.
* feat(themes): element tokens (ui_tool, ui_thinking) + skinnable diffs
Theming was semantic-only: the gold tool `●` was `accent`, shared with
headings/links/chevrons, so "recolor tool calls" was impossible and the agent
had no key to point at. Add `ui_tool` (● + tool spinner) and `ui_thinking`
(reasoning body) tokens that fall back to accent/muted — defaults unchanged,
but now independently settable. Make diffs skinnable too (`diff_*`), which
fromSkin previously hardcoded. Document the full element→key map in the skill so
Hermes knows which knob turns what.
* fix(themes): tweak the ACTIVE skin in place, never fork default
Changing one color ("make the tool ● cyan") forked `default` — which has no
`background` — so applying it reset the terminal to its own (black) default and
dropped the active skin's palette. Teach the skill to edit the active skin's file
in place for a tweak (watcher repaints on the mtime bump), and to fork a built-in
only by carrying its full palette. Hard pitfall: never fork `default` for a tweak.
* feat(themes): `hermes skin set` — deterministic one-color tweak, bg untouched
Changing a single color kept wrecking the rest because the agent hand-authored a
new skin (often from `default`, which has no `background`, resetting the terminal
to black). Add `hermes skin set <key> <hex>`: edits the ACTIVE skin's one key in
place (a built-in is forked into an editable copy carrying its full palette), so
everything else — background included — is preserved. Plus `skin use` / `skin
list`. The skill now points tweaks at this command instead of hand-authoring.
* feat(themes): dedicated code-syntax palette keys
Code highlighting reused brand tokens (accent/text/border/muted), so it couldn't
be themed independently. Add syntax_string/number/keyword/comment skin keys →
syntax* theme tokens (defaulting to those brand tokens, so defaults are
unchanged) and point the highlighter at them. Documented in the element→key map.
* test(themes): E2E live skin switch — config write → skin.changed broadcast
* fix(themes): reconcile element/syntax tokens with main's derive+adapt pipeline
Element tokens (ui_tool/ui_thinking), skinnable diffs, and code-syntax keys
flow through buildPalette → adaptColorsToBackground instead of a hand-mapped
color block, so they inherit #20379's contrast/polarity machinery. thinking
and syntaxComment track the EFFECTIVE muted (banner_dim override included);
the skin's `background` feeds the surface (it also paints the terminal via
OSC 11); statusFg falls back through ui_text/banner_text. Tests assert the
routing/independence contracts rather than pre-adaptation hexes.
* fix(themes): apply a runtime switch back to default on the desktop
ingestBackendSkin returned early for name === 'default' even when
apply=true, so a real runtime switch to the default skin (/skin default
on CLI/TUI, or config.set display.skin=default) emitted skin.changed but
never repainted the desktop. 'default' is no-opinion on the PALETTE (the
desktop keeps its own nous default, so we still never register a converted
theme under it), but it IS a valid apply TARGET: setTheme normalizes
'default' -> nous, so switching back repaints to the desktop default.
Skip only the registry step for 'default' and let it flow through the
apply guard. Addresses Copilot review.
* fix(tui_gateway): serve candidate-inclusive display on warm/live resume
#65919 persists verification candidates (finish_reason=verification_required
/ verify_hook_continue) to state.db but collapses them out of the in-memory
model history via repair_message_sequence. The eager session.resume + REST
paths read the verbatim display lineage (candidate present), but the
warm/live-reuse payload (_live_session_payload) built its user-visible
messages from the collapsed in-memory model history — so switching to a
still-live session dropped the substantive verification answer that a cold
resume of the SAME session showed. That divergence is the cross-session
"substantive text vanishes on switch" class, and the direct sibling of the
resume-duplication regression fixed in #68149.
Reconcile the persisted display lineage (candidate-inclusive, the same
get_messages_as_conversation(..., include_ancestors=True) read the eager
resume + REST paths use) with the fresh in-memory tail in
_live_visible_history, so all three surfaces agree by construction while a
not-yet-flushed live turn is still shown. Extracted
_reconcile_display_with_live as a pure, DI-testable function (anchors on the
last persisted row's (role, text); appends only the uncovered in-memory tail;
trusts the DB display when the tail can't be anchored).
Tests: unit coverage for candidate-inclusion, freshness, empty/raising-DB
fallback, and the combined candidate+fresh-tail case. The existing freshness
guard (test_session_resume_live_payload_uses_current_history_with_ancestors)
stays green.
* fix(tui_gateway): candidate-inclusive display on child-watch resume + E2E
Complete the #65919 warm/live-payload fix across its sibling path and add
real-SessionDB cross-builder coverage.
- Child-watch (lazy) resume: the delegated-subagent watch window served
_history_to_messages(repaired_history) for its user-visible messages, which
collapses out persisted verification candidates just like the warm-payload
path did. Build the visible messages from the verbatim child-only display
projection (repair_alternation=False) while the repaired history still feeds
live replay; fall back to the repaired history if the display read fails.
- E2E cross-builder consistency (real SessionDB, not mocks): a persisted
verification candidate is collapsed out of the model projection but kept in
the display projection, and _live_visible_history now equals the eager
session.resume display projection (candidate present). Adds the combined
candidate + fully-flushed-second-turn case and a lazy child-watch handler
test that asserts the candidate survives in resp["result"]["messages"].
* fix(cli): add skin to _BUILTIN_SUBCOMMANDS for plugin gating
The new hermes skin subcommand must be declared so startup plugin
discovery can skip when the user targets it.
* fmt(js): `npm run fix` on merge (#69048)
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
* feat(desktop): Billing page revamp — current-plan card, in-app plans view, tier art (#68722)
* feat(desktop): revamp Billing page — plan card, in-app plans view, tier art
Reshape the desktop Billing settings per wayfinder ticket 09. New page order:
Plan → Payment → One-time top-up → Automatic refill → Usage, with the at-a-glance
summary strip unchanged at the top.
- CurrentPlanCard replaces the old Subscription row: tier name + price + renewal
and at most one button — "View plans" (free/no-sub + can_change_plan),
"Change plan" (subscriber + can_change_plan), or none for teams / non-changers.
Teams keep the portal "Adjust plan ↗" link so they are not stranded. The button
navigates in-app to the plans sub-view.
- bview=plans sub-view mirrors the settings pview/kview pattern (useRouteEnumParam,
default overview). BillingPlansView renders a grid of PlanCard from live tiers[]
(is_enabled, sorted by tier_order, free tier included).
- PlanCard: tier art + name + $/mo + monthly credits as dollars ("$110 credits/mo").
Current tier is highlighted + inert; higher/no-current tiers get "Choose ↗"
(opens portal with plan=<tierId>); lower tiers are a DISABLED "Downgrade" with a
caption — downgrades move in-app in ticket 11 (gateway pending-change flow), so
this PR intentionally links them out/disabled rather than wiring the money path.
- buildManageSubscriptionUrl gains an optional third arg (tierId) → appends
plan=<tierId>. Signature kept identical to draft PR #68666 for a trivial rebase;
NAS #748 validates the param server-side.
- Tier art: four NAS hero webps rendered as ~40px thumbnails over a Nous-blue well
with per-tier blend modes (the only place Nous blue appears). Keyed by lowercase
tier NAME (free/starter→connect, plus→memory, super→automation, ultra→sandbox);
unknown name → text-only card. Imported via vite static imports for packaged
file:// + webSecurity.
- Top-up vs auto-refill disambiguated by section label + first sentence: "One-time
top-up" / "Buy credits now" vs "Automatic refill" / "Refill when low" (configured
copy reads "Charges $X automatically when your balance falls below $Y.").
- Variant-A auto-refill editing: Manage swaps the row's left side (caption → the two
$ fields with a pre-allocated error line) and the action column (Manage → Save/
Cancel) in place, with the row height reserved for the tallest state so the Usage
section never shifts. Fixes the spurious on-open validation error (errors now show
only after an edit or a save attempt). Save/disable API calls + confirm-disable
flow unchanged.
- Remove subscriptionTierChips and the subscription-row chips; reshape (not delete)
deriveBillingView to expose plan + tiers. Buy-credits row keeps the chips seam.
- Dev fixtures: add free-personal and subscriber-personal (personal orgs, full
4-tier Free/Plus/Super/Ultra catalog) so the plans view is exercisable.
Tests: update/extend index.test.tsx + use-billing-state.test.ts, add tier-art.test.ts;
delete the old chips tests. Desktop billing suite 70/70 green, typecheck clean.
* fix(desktop): mark the free/lowest tier current (not an upgrade) when there is no subscription
Visual verification caught a spec-fidelity bug: in the plans grid, an account with
no active subscription rendered the Free tier ($0/mo, tier_order 0) as a "Choose ↗"
upgrade — clicking would deep-link the portal to "subscribe to Free".
Ruling: current-card = tier.is_current OR (subscription.current == null AND the tier
is the lowest-order / $0 tier). derivePlanTiers now falls back to the lowest-order
tier as the stand-in current plan when there is no subscription, so the free card
renders exactly like is_current (inert, "Current plan") and — being the lowest order
— no tier can be a downgrade; every paid tier is a "Choose ↗" upgrade.
CurrentPlanCard is unaffected (still "Free" + "View plans"); subscriber-personal is
unchanged (Free stays a disabled Downgrade below the current Plus tier).
Tests: free-personal grid now asserts Free = current/inert, no downgrade state, three
Choose buttons; text-only unknown-tier test gains a free tier so the unknown paid tier
is unambiguously an upgrade. Billing suite 70/70 green, typecheck + lint clean.
* chore(desktop): shrink bundled tier art to 128px thumbnails
The plan-card wells render the art at ~40px; shipping the full landing
images added 2.7 MB to the repo for no visible difference. 128px covers
2x displays; total is now 26 KB.
* fix(desktop): address 6 adversarial-review findings on the Billing revamp
1. Grandfathered current tier (BLOCKER). NAS marks a grandfathered current tier
is_enabled:false; the enabled-only filter dropped it, leaving currentOrder
undefined so every lower tier rendered as an actionable "Choose ↗". derivePlanTiers
now resolves current identity/ordering against the UNFILTERED tiers and keeps the
grandfathered current tier in the grid as the inert "Current plan" card; downgrades
classify against its tier_order. (Non-current disabled tiers are still dropped.)
2. Dead plan-card button. derivePlanCard offered "View plans"/"Change plan" purely on
can_change_plan, but the grid could be empty / current-only and showPlans refused,
so the button no-oped. It now offers the in-app action ONLY when the grid has ≥1
actionable (non-current) tier; otherwise it falls back to the portal link.
3. Deep-link bypass. showPlans now gates on the same capability that renders the button
(view.plan?.action), so a team / non-changer deep-linking bview=plans always falls
back to overview instead of a grid of live Choose buttons.
4. Lost portal escape hatch. Whenever the card has no in-app action (teams, non-changers,
refused subscription, empty catalog) it now ALWAYS carries the "Adjust plan ↗" portal
link built from subscription?.portal_url ?? billing.portal_url — the refusal caption
no longer promises a portal the UI didn't render.
5. Choose URLs dropping org_id/plan. (a) derivePlanTiers now threads billing.portal_url
as the fallback base for the Choose URL. (b) buildManageSubscriptionUrl treats the
hard-coded FALLBACK_PORTAL_BILLING_URL as a last-resort ORIGIN (applying org_id/plan)
instead of a bare return, so a null portal_url never strips the routing params.
6. Zero-shift on narrow panes. Replaced the magic min-h-28 (under-reserved once the two
inputs stack below @2xl) with exact reservation: the edit form is always rendered and
both states share one grid cell ([grid-template-areas:'stack']), invisible+aria-hidden
when not editing — the row equals the tallest state at every width, no breakpoint math.
The refusal stays inside the reserved layer.
Tests: +12 (grandfathered current, no-dead-button + empty-catalog portal link, team &
personal deep-link fallback to overview, billing.portal_url-backed Choose URL, fallback
org_id/plan, reserved-form-mounted); updated the two portal-link expectations for §4.
Billing suite 78/78 green; typecheck (app/electron/e2e) + lint clean.
* refactor(desktop): reuse the shared openExternalLink helper in the plans view
* fix(desktop): honor the auto_reload wire contract — null card + disable amounts
A full-stack contract sweep (desktop ↔ shared types ↔ gateway ↔ NAS) surfaced two
real desktop bugs in the auto-refill row:
A. auto_reload.card can be null. The gateway's _parse_auto_reload_card returns None
for a missing/unknown-kind card and _serialize_billing_state emits `card: null`,
but the shared BillingAutoReload.card union had no null arm and use-billing-state
dereferenced `autoReload.card.kind` bare — a crash on the enabled path. Add `| null`
to the shared union (contract honesty) and guard the read (`card?.kind`); null now
falls through to the default enabled path, same as a canonical card.
B. Disable was rejected by the gateway. billing.auto_reload unconditionally requires
threshold + top_up_amount, so `updateAutoReload({ enabled: false })` came back
invalid_request. (The TUI always sends both; desktop fixture mode stubbed it.)
disable() now sends the current threshold_usd/reload_to_usd from the autoReload
prop alongside enabled: false, matching the TUI.
Tests: enabled auto_reload with card:null renders the normal enabled row (derivation
+ render, no crash); disable call carries both current amounts. Billing suite 80/80
green; typecheck (app/electron/e2e) + lint clean.
* fix(tui): guard the nullable auto_reload card in the auto-reload screen
The shared BillingAutoReload.card union gained its honest null arm (the
gateway emits card: null for a missing/unknown card); the TUI's only bare
dereference follows the same default path as a canonical card.
* fix(desktop): align billing inputs to the sm control height
The three billing inputs used an ad-hoc h-8 (32px) next to size=sm
buttons (24px). They now use the control system's size=sm with a
py-[3px] compensation for the input's real 1px border — buttons draw
theirs as an inset shadow, so sm alone still sits 2px taller. All five
controls in the buy row now measure 24px.
* fix(desktop): plan-card actionability + billing view-model hardening
Code-quality review of the Billing revamp (PR #68722).
BLOCKING — a top-tier subscriber (only downgrades/current below them) opened a
plans grid with zero enabled actions AND no portal link. The plan card gated its
in-app button on `tiers.some(state !== 'current')`, which counts the (disabled)
downgrade tiles. It now gates on an actual UPGRADE being present
(`capable && tiers.some(state === 'upgrade')`); with no upgrade the card falls back
to its "Adjust plan ↗" portal link, and the bview=plans deep link (gated on the same
plan.action) falls back to overview.
Reviewer structural items:
- One "plans capability" verdict (personal + can_change_plan + subscription ok) is
derived once in deriveBillingView and threaded to BOTH derivePlanCard and
derivePlanTiers; the grid only mints upgrade actions when capable, so the invariant
lives in one place.
- BillingPlanTierView is now a discriminated union (`current` | `downgrade` w/
disabledCaption | `upgrade` w/ required action), and BillingPlanCardView is an
action-XOR-link union — deleting the `tier.action?.url ?? ''` and `plan.link?.url`
defensive branches in the consumers.
- `findCurrentTier(subscription)` replaces the repeated is_current||id predicate at
its three sites (plan card price, grid ordering, summary plan line).
- BillingView exposes named `paymentRow` / `topupRow` / `refillRow` instead of an
`accountRows[]` + three `.find(id)` lookups.
- The auto-refill row that edits in place carries an explicit `manageInApp: true`;
AutoReloadRow keys off it instead of sniffing the action label/url.
- tier-art header comment no longer cites an internal repo path; dead `?.` removed
from RowValue (via a destructured const) and the plan-card link handler.
Behavior is identical except the blocking fix. Billing suite 82/82 green; typecheck
(app/electron/e2e) + lint clean.
* refactor(desktop): adopt inline-review nits on the billing plan card
Resolves the inline suggestion threads:
- plan-card gate reads a named `hasActionableTier` = "a tile carries an action"
(union-safe `'action' in tier`, equivalent to the old upgrade-only check).
- re-narrow link/action inside the click callbacks (`plan.link && …`,
`tier.action && …`) rather than relying on outer narrowing.
Behavior unchanged; billing suite 82/82 green, typecheck + lint clean.
* fmt(js): `npm run fix` on merge (#69050)
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
* feat(cli): plan catalog on Free + plan= deep link + top-up/auto-refill copy split (#68689)
* feat(cli): plan catalog on Free + plan= deep link + top-up/auto-refill copy split
Bring the plain (non-TUI) CLI billing surface to parity with the desktop/TUI
billing changes:
- /subscription on Free (admin/owner, interactive) prints the plan catalog
(name · $/mo · $credits/mo, from the same tiers[] data the TUI uses; monthly
credits render as dollars). A numbered pick opens the manage-subscription
deep-link directly with plan=<tier_id> appended.
- subscription_manage_url(state, tier_id=...) appends plan=<tier_id> (the stable
tiers[] id) when a tier was picked, org_id first — mirrors the TUI's ?plan=.
The paid change flow's blocked/unknown-preview portal fallback carries plan=
for upgrades only; downgrades stay generic/native.
- /topup overview splits one-time top-up from automatic refill, the distinction
stated in each first sentence ("Add funds now — a single charge…" vs "Refill
when low — charges … automatically …"), keeping "credits" out of the
dollars-only surface.
- Downgrades remain native (chargeless scheduled change), unchanged.
Updates the CLI-parity section of docs/billing-lifecycle.md and tests under
tests/hermes_cli + tests/agent.
* refactor(billing): share plan-catalog helpers + harden manage-url builder
- subscription_manage_url now preserves unrelated portal query params (parse_qsl,
popping only the contract-owned org_id/plan) and restricts to http/https schemes,
matching the desktop URL builder — the function owns the contract.
- Lift the plan-catalog derivation into agent/subscription_view.py so the CLI Free
catalog and the paid picker/blocked-preview branch share one implementation:
selectable_tiers (enabled paid, not current, sorted), format_tier_row (name · $/mo
· $credits/mo — thousands-grouped like the TUI's toLocaleString, credits suffix
hidden when absent/zero), and is_upgrade(state, tier_id).
* fix(cli): numbered pick, canonical guarded browser opener, partial auto-refill copy
- Free catalog: accept a bare digit as a pick (the shared normalizer only knows the
confirm-dialog digit aliases, so `1` used to resolve to None → "Cancelled"). The
Nth digit maps to the Nth printed row.
- Extract one _open_url_in_browser used by every "open the portal" path, applying the
device-code flows' console-browser / remote-session guard (webbrowser.open returns
True even for lynx/w3m over SSH) and returning whether a real browser opened.
- Consume the shared selectable_tiers / format_tier_row / is_upgrade helpers from the
Free catalog, the paid picker, and the blocked-preview branch.
- /topup auto-refill copy: the concrete "charges $X … below $Y." sentence only when
both amounts are present and finite; otherwise the generic sentence.
* docs(billing): correct CLI-parity rows (drop cross-repo ref, downgrade invariant)
Remove the other-repo PR reference from the manage-URL row, and state the real
downgrade invariant: a blocked downgrade may print the generic manage URL but never
carries plan=<tier_id> — selected-tier deep-links are reserved for new subscriptions
and upgrades.
* feat(desktop): native in-app downgrade — chargeless preview → schedule → undo (#68761)
* feat(desktop): native in-app downgrade (chargeless preview → schedule → undo)
Ticket 11, stacked on the Billing revamp (ticket 09). Downgrades no longer bounce
to the portal — picking a lower tier runs the gateway pending-change flow in-app;
the scheduled state renders on the plan card with an undo. Upgrades keep the portal
deep link.
- api.ts: add previewSubscriptionChange / scheduleSubscriptionChange /
resumeSubscription wrappers over subscription.preview|change|resume
({subscription_type_id} / {}), typed via SubscriptionPreviewResponse +
BillingMutationResponse (now re-exported from types.ts).
- use-subscription-change.ts (new): useDowngradeFlow (preview → confirm → schedule,
refetch + onScheduled on success; typed refusals surface via the shared
BillingRefusalInline, so insufficient_scope drives the existing step-up exactly
like the auto-reload save, retried in place) and useResumeFlow (confirm-less undo).
Both accept a `simulate` switch so DEV fixtures click through with canned success.
- plans-view.tsx: downgrade tiles are now an actionable "Downgrade" that opens an
in-card preview → confirm panel (mirrors the TUI confirm copy: "…takes effect
<date>. No charge now; you keep your current plan until then."). The scheduled
downgrade target renders an inert "Scheduled" marker; other lower tiers stay
actionable (picking one reschedules).
- CurrentPlanCard: when a downgrade is pending, the caption reads "Changes to
<tier> on <when>." with an inline Undo → resume → refetch. One line, no jumps.
- use-billing-state.ts: BillingPlanTierView gains a `scheduled` state (and drops the
ticket-09 disabled-downgrade caption); derivePlanTiers matches the pending target
by name (NAS sends no id for it) before the downgrade branch; BillingPlanCardView
gains `pending`, derived from current.pending_downgrade_* .
- inline-feedback.tsx (new): extracted openExternal / BillingRefusalInline /
StepUpInlineAction / InlineMessage so the plans view reuses the step-up-aware
refusal renderer without a circular import; openExternal now delegates to the
canonical @/lib/external-link opener.
- dev-fixtures.ts: add `pending-downgrade` (subscriber-personal on Plus with a Free
downgrade scheduled for Aug 15) for the plan-card pending state + grid marker.
Tests (+16 → 94 green in the billing suite): api wrappers (preview/change/resume +
insufficient_scope refusal); view derivation (pending plan-card state, scheduled
grid marker); confirm flow (preview shown, change called with the right tier_id,
refetch on success, schedule refusal → step-up affordance); undo flow; the
use-subscription-change hooks (preview-refusal retry, cancel, simulate path).
Updated the ticket-09 downgrade tests for the now-actionable tile. typecheck
(app/electron/e2e) + lint clean.
PR (later): base sid/desktop-billing-revamp; retarget to main after #68722 (09) merges.
* fix(desktop): format downgrade credits delta as signed dollars
The downgrade preview rendered the raw wire string ("Monthly credits change:
-88."), violating the "monthly credits are DOLLARS" ruling. NAS sends
monthly_credits_delta as a bare decimal; format it as signed dollars through the
same money formatter ("−$88/mo", sign preserved, abs value formatted). Zero /
absent still hides the line.
Adds formatMonthlyCreditsDelta (exported) + unit tests (negative/positive/zero/
absent) and asserts the rendered "Monthly credits change: −$88/mo." in the confirm
flow. Billing suite 99/99 green; typecheck + lint clean.
* fix(desktop): downgrade flow hardening — concurrency guard, a11y, DEV-gated sim
Addresses the adversarial review of the native-downgrade diff.
- Concurrency: useDowngradeFlow exposes `mutating` (true only while the schedule
RPC is in flight). While a change commits, every other Downgrade tile and the
Back button are disabled; the active panel's Confirm/Cancel already lock. The
plan-card Undo blocks on its own resume via `busy`. (The server also 409s
overlapping per-org mutations — this is UI honesty, not the only defense.)
- Accessibility: the confirm panel is role="status" aria-live="polite" and takes
focus on open (tabIndex=-1 container); closing it returns focus to the tile card,
so keyboard focus is never stranded and the async preview text is announced.
- DEV-gated simulation: the canned preview/change/resume seam is ignored unless
import.meta.env.DEV, so a production build never takes the simulated branch even
if a `simulate` prop leaks through.
- Comments: documented the deliberate manual-retry-after-step-up (no auto-replay,
matching auto-reload) and that name-matching the scheduled target is safe because
SubscriptionTypes.name is @unique in NAS.
Tests (+5 → 104 green in the billing suite): mutating exposed only during schedule;
simulate ignored outside DEV; other downgrade tiles + Back disabled mid-schedule;
Undo disabled mid-resume; confirm panel role + focus on open. typecheck
(app/electron/e2e) + lint clean.
* fix(desktop): scheduled cancellations, downgrade-flow concurrency, inline nits
Addresses the native-downgrade review threads.
Scheduled cancellations were invisible (NEW review item). subscription.current
carries cancel_at_period_end + cancellation_effective_* and subscription.resume
clears cancellations exactly like downgrades, but the pending-transition helper only
read pending_downgrade_*, so a portal/TUI-scheduled cancellation rendered as a plain
renewal with no Undo. The pending state is now a union — { kind:'downgrade', tierName,
when } | { kind:'cancellation', when } — computed once in deriveBillingView and
threaded to BOTH the plan card and the grid. The card reads "Cancels on <date>." with
the same Undo (resume); the grid shows a Scheduled marker only for downgrades (a
cancellation has no target tier). Precedence: a downgrade wins if both fields are set
(it names a concrete target — the stronger signal), commented at the helper. Adds a
`pending-cancellation` fixture + tests (card copy, undo wiring, no grid marker,
downgrade-wins precedence).
Concurrency: confirm() takes a synchronous scheduling ref (mirroring useResumeFlow)
so two same-tick clicks — before React commits busy='schedule' — cannot fire two
schedule RPCs; the ref clears on every exit (simulated/stale/refusal/success).
useResumeFlow reorders its unlock: a refusal releases immediately, a success holds
runningRef/busy THROUGH the refetch so Undo never re-enables against the still-pending
card. Test: a synchronous double-activation fires one schedule RPC.
Inline nits: re-narrow link/action inside the click callbacks (`plan.link && …`,
`tier.action && …`) instead of relying on outer narrowing / `?? ''`.
Billing suite green (109); typecheck (app/electron/e2e) + lint clean.
* refactor(desktop): move DEV billing simulation behind the api seam
The fixture simulation lived as `simulate` / `simulateResume` prop drills and
`if (simulated)` branches inside the flow hooks, and it could not actually produce
the state it advertised (a simulated schedule never showed the pending card).
Replaced with `createSimulatedBillingApi(fixture)` — a fully in-memory BillingApi
built once, DEV-gated, in BillingSettingsWithDevFixtures where the fixture is known,
and supplied to the whole subtree via a new `BillingApiProvider` (context override on
`useBillingApi`; `null` = the real gateway api). It serves fetches from a mutable copy
of the fixture and its subscription-change mutations WRITE that copy's pending state:
schedule sets a pending downgrade, resume clears a pending downgrade OR cancellation.
Fixture mode now flows through the SAME react-query path (fetch short-circuit deleted;
queries always enabled; an effect refetches on fixture switch), so the click-through
genuinely progresses — schedule → pending card + Undo + Scheduled marker, undo → cleared.
Deleted `SubscriptionSimulation`, `simulationEnabled`, both prop drills, and every
`if (simulated)` branch — the hooks are now production-pure. Added a test driving the
full simulated loop (schedule → pending appears → resume → cleared), plus cancellation
undo and no-shared-mutation coverage. Removed the now-obsolete simulate hook tests.
Billing suite green (110); typecheck (app/electron/e2e) + lint clean.
* refactor(desktop): extract billing row/card components out of index.tsx
Purely mechanical, no behavior change: split the settings billing route file
(1065 → 593 lines) into focused siblings now that the downgrade feature has settled
their final shape.
- billing-amounts.ts — the dollar parse/format/validate/clamp helpers.
- account-row-value.tsx — RowValue (shared by AccountRow + AutoReloadRow).
- current-plan-card.tsx — CurrentPlanCard.
- auto-reload-row.tsx — AutoReloadRow (the in-place auto-refill editor).
index.tsx keeps the page shell, AccountRow dispatch, BuyCredits flow, and the fixture
wiring. Billing suite green (110); typecheck (app/electron/e2e) + lint clean.
* refactor(desktop): tighten the downgrade flow — phase union, previewMessage, tidy shared modules
Polish that composes with the api-seam rework:
- ActiveDowngrade's four nullables become a `DowngradePhase` discriminated union
(previewing | previewFailed | ready | scheduling | scheduleFailed). Impossible
combinations (a preview AND a refusal, "ready" with no quote) can no longer be
represented; the hook and panel branch on one `kind`, and `mutating` is simply
`phase.kind === 'scheduling'`.
- The five-way ternary in DowngradeConfirm is replaced by a pure `previewMessage(phase,
fallbackTierName)` helper; the misnamed `caption` className local is renamed `captionCn`.
- inline-feedback.tsx now holds ONLY the shared refusal/step-up pieces: `openExternal`
moves to its own `open-external.ts` (a thin wrapper over `@/lib/external-link`'s
`openExternalLink`), and `InlineMessage` moves back into its sole consumer
(auto-reload-row.tsx).
No behavior change. Billing suite green (110); typecheck (app/electron/e2e) + lint clean.
* fmt(js): `npm run fix` on merge (#69067)
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
* fix(gateway): hard-exit CLI runner after graceful teardown
* test: update gateway run stub for hard-exit helper
* fix(gateway): hard-exit on KeyboardInterrupt path too
The KeyboardInterrupt handler in run_gateway() was the only exit path
that still used bare 'return' instead of _hard_exit_after_gateway_teardown().
While less common than service-managed restarts, a console Ctrl+C still
leaves the process vulnerable to the same Python finalization hang on
non-daemon worker threads (cron ThreadPoolExecutor jobs). Route it through
the same backstop, with a 'return' guard for test stubs that don't raise
on code 0 (production os._exit never returns).
* fix(cli): pass conversation_history on /new /resume /branch flush
Closes#68454
Root cause: cold-resumed transcript rows lack _DB_PERSISTED_MARKER until a
normal turn flush stamps them. Immediate /new,/resume,/branch flushed with
no history boundary, so every restored row was re-appended to the old session.
Fix: pass conversation_history=self.conversation_history at all three sites
(mirrors #68205). Add offline regression coverage for noop + tail-only write.
Verification: pytest tests/agent/test_session_rotation_flush_cold_resume_68454.py (4 passed)
* test: drop source-grep change-detector from #68480
The three behavior tests (control proves dup, boundary is noop, tail-only
write) fully cover the flush semantics. The source-grep test reading
cli.py + cli_commands_mixin.py as text and asserting a string appears is
a change-detector that breaks on benign refactors without adding coverage.
* test: update mock assertions for conversation_history kwarg
The /branch and /resume flush tests asserted the old positional-only
call signature. Update to match the fix from #68480.
* fix(gateway): make adapter fatal-error handoff cancellation-proof; exit if a platform is stranded
The fatal-error notification runs on the failing adapter's own polling
task, and adapter.disconnect() inside the handler can cancel that task
(its current-task guard misses because _safe_adapter_disconnect runs the
close in a wrapper task). The CancelledError killed the handler between
the fatal log and the reconnect queue, leaving the platform permanently
dead inside a live gateway process. #68447 fixed this for telegram at
the adapter layer; this hardens the shared gateway dispatch so every
platform gets the same protection (qqbot #25505/#29005, photon #68693).
- _handle_adapter_fatal_error now runs the real handler in a detached
task, awaited through asyncio.shield() so caller cancellation cannot
tunnel into it (Task.cancel() also cancels the task's _fut_waiter).
- If a retryable platform still ends up neither reconnected nor queued,
the gateway exits with failure so launchd/systemd KeepAlive restarts
it instead of running indefinitely with a dead platform (#68693).
Fixes#68693
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* chore: map anoop.mehendale@gmail.com -> anoopmehendale-cue
For PR #69007 salvage (#69112).
* fix(update): isolate systemctl timeouts per gateway unit during fleet restart
A TimeoutExpired from one hermes-gateway*.service used to abort the whole
per-scope restart loop, leaving later profile gateways on pre-update
in-memory code after hermes update. Catch timeouts per unit, continue the
fleet, warn with the exact stale units, and exit non-zero when any remain
unrestarted (#68523).
* test(update): cover fleet restart timeout isolation (#68523)
* fix: refresh vulnerable npm lockfile entries
* chore: AUTHOR_MAP for tinetwork
* fix(compression): prevent stale-budget retry loops
* fix(compression): harden startup route scoping
* fix(providers): align custom route scoping
* test(compression): cover overflow after blocked preflight
* test(providers): cover route URL identity boundaries
* test(providers): cover query path slash identity
* test(providers): complete hermetic route coverage
* test(providers): cover URL whitespace route identity
* fix(providers): fail closed on missing active route
* fix(providers): scope route-owned runtime settings
* fix: restore base_url rstrip, extract should_clear_context_pin helper
Salvage follow-up for PR #68899:
- Restore .rstrip('/') on base_url in _swap_credential (both anthropic
and OpenAI paths) to match every other assignment site. The route
identity comparison still uses normalize_route_base_url which handles
trailing slash correctly.
- Extract should_clear_context_pin() into hermes_cli/route_identity.py,
consolidating 7 copy-pasted call sites across cli.py, gateway/run.py,
gateway/slash_commands.py, and hermes_cli/model_switch.py into a
single fail-closed helper.
C1 (anthropic path TLS re-application): pre-existing gap — the Anthropic
adapter (build_anthropic_client) has no TLS customization support at
all, so this is out of scope for this salvage.
* fix(compression): ignore assistant handoff summaries in tail anchor
Assistant-role compaction summaries were treated as the last visible assistant reply after head protection decayed. That pulled the tail boundary back to the summary itself and left zero new turns to summarize.
Exclude internal context summaries from both the visible-reply search and the assistant fallback, mirroring the existing user-role summary exclusion.
* chore: AUTHOR_MAP for McHermes
* fix(telegram): group authz fallback + command sender identity
- authz_mixin: add config.extra fallback for group_allowed_chats
when observe-unmentioned mode strips user_id from env-var check
- authz_mixin: check adapter allow_from/group_allow_from for
user authorization from config.yaml without env vars
- telegram/adapter: separate group_allow_from for group chats
vs allow_from for DMs
- telegram/adapter: preserve sender source for command messages
so admin-only slash commands work in groups
- telegram/adapter: add _telegram_extra fallback for
group_allow_from config reading
* fix(telegram): address review findings from PR #67816
- Update test_observed_group_context_preserves_slash_command_text_for_dispatch
to assert user_id is preserved for COMMAND messages (new correct behavior)
- Add _coerce_allow_set helper to handle both list and comma-separated
string allowlist inputs (prevents character-by-character iteration bug)
- Include 'channel' in chat_type checks for group-scoped authorization
- Add _telegram_extra fallback for group_allowed_chats (consistent with
group_allow_from fallback)
- Add AUTHOR_MAP entry for nyaruko@hermes -> tsuk1nose
* fix(telegram): update auth check tests for group_allow_from split
Update test_telegram_auth_check.py to use group_allow_from for group
messages (matching the PR's intentional behavior split: allow_from for
DMs, group_allow_from for groups). Add test_is_user_authorized_from_message_group_allow_from
to cover the new group path.
* fix(openviking): recover pending session commits
* docs: clarify OpenViking local setup
(cherry picked from commit a6807170f1)
(cherry picked from commit 6fb4e9aa8a)
* fix(openviking): serialize orphan session recovery
* fix(openviking): chunk structured session sync
Preserve ordered structured turns across OpenViking's 100-message batch limit and resume retries from the first unconfirmed message.
Based on the OpenViking batching work from commit 1a567f7067 in #58981.
* refactor: cleanup follow-up for salvaged PR #58871
- Remove dead current_sid parameter from _recover_pending_sessions
- Remove dead cleanup parameter from _release_owner_run_claim (always True)
- Set _run_lock_path after flock succeeds, not before
- Collapse redundant BlockingIOError branch (covered by OSError+errno check)
- Track _pending_marked_sids to skip re-writing marker file on every sync_turn
* fix(openviking): inject session-start memory context
(cherry picked from commit 18b474d0bd)
* fix(openviking): align session context with shared profile contract
* fix: discard both session IDs on compression for profile re-injection
The _profile_prefetched_sessions set stores whichever session_id was
passed to prefetch(), which may differ from self._session_id. On
compression, only old_session_id (self._session_id) was discarded,
missing the case where the stored key was the prefetch session_id
parameter. Discard both old and new IDs to cover all cases.
* chore: add kshitij@kshitij.dev to AUTHOR_MAP
* fix(secrets): fall back to stale disk cache when bws live fetch fails
Without this, a single DNS hiccup or BWS outage at gateway startup leaves
the whole fleet running with an empty credential pool — every model call
fails until someone restarts after the network recovers. When a previous
successful fetch already populated the disk cache, return those secrets
with an explicit warning instead of raising RuntimeError.
`use_cache=False` (explicit opt-out) still raises so manual flows like
the setup wizard surface the original error. The disk cache is not
re-written on the fallback path so a process restart still triggers a
proper TTL re-check.
Fixes#41925
* fix(secrets): port stale-cache fallback to current DiskCache API + gate by error kind
The stale-fallback branch called _read_disk_cache(), a helper removed in
db495b0fba when disk-cache logic moved to the
shared DiskCache class — every fallback attempt raised NameError instead of
serving cached secrets, silently defeating the PR's whole purpose. Port to
_DISK_CACHE.read().
Also tighten the fallback per DiskCache's TTL contract and the secret-source
error taxonomy:
- Gate on cache_ttl_seconds > 0 so a caller that opted out of caching
entirely (ttl=0) never gets a secret value that didn't come from a live
fetch, even on the failure path.
- Gate on _classify_bws_error(str(exc)) being NETWORK or TIMEOUT, reusing
the existing classifier — an AUTH_FAILED or malformed-output failure must
still raise, since serving stale secrets there would mask a real
credential/config problem instead of a transient outage.
Ported the test helpers off the removed _write_disk_cache to a direct JSON
write (matching this file's existing disk-cache test convention) and added
tests for the auth-failure, malformed-output, and zero-TTL gates. Reverting
the fix and re-running confirms 7 of 8 stale-fallback tests fail with the
original NameError.
* fix(secrets): fold OP_CONNECT_HOST/OP_CONNECT_TOKEN into 1Password auth cache-key
_auth_fingerprint() built the 1Password secret cache-key from the
service-account token, OP_ACCOUNT, and OP_SESSION_* vars but omitted
OP_CONNECT_HOST/OP_CONNECT_TOKEN, which are in _OP_ENV_ALLOWLIST and are
forwarded to the op child (the Connect-server auth path). Rotating
OP_CONNECT_TOKEN or re-pointing OP_CONNECT_HOST at a different Connect
identity left the fingerprint unchanged, so both the in-process and disk
caches kept serving secrets resolved under the old Connect credentials for
the full TTL (default 300s, disk-persisted across invocations). This
contradicts the function's own docstring invariant that a value cached
under a previous identity is never served under a new one; it closes the
gap for the Connect path, matching the OP_SESSION_*/service-account paths
that are already protected.
* fix(secrets): pass OP_LOAD_DESKTOP_APP_SETTINGS through to the op child env
The 1Password secret source builds a minimal allowlisted environment for the
`op read` child process. The allowlist omits OP_LOAD_DESKTOP_APP_SETTINGS, so a
user who exports it (shell, .env, or service unit) sees it silently stripped
before it reaches `op`.
That var is `op`'s documented switch to skip the desktop-app integration probe.
When the 1Password desktop app is installed, `op` probes its settings/socket at
startup *before* evaluating service-account auth. If the desktop app's group
container is wedged (e.g. macOS 'Interrupted system call' on the 1Password group
container), that probe blocks with no timeout, so `op read` hangs indefinitely
even with a valid OP_SERVICE_ACCOUNT_TOKEN present. Setting
OP_LOAD_DESKTOP_APP_SETTINGS=false is the intended escape hatch — but stripping
it means it has no effect on exactly the headless boxes that need it.
Fix: add OP_LOAD_DESKTOP_APP_SETTINGS to _OP_ENV_ALLOWLIST so the documented
var reaches the child. No behavior change when it's unset. Adds a focused test
alongside the existing allowlist test.
Repro: on a machine with a wedged 1Password desktop container + a valid SA
token, `op read` hangs 600s+ without the var and returns in ~4s with it — but
only if it actually reaches the op process, which this allowlist entry ensures.
Co-authored-by: Minh Nguyen <menhguin@users.noreply.github.com>
* fix(mcp): pass secret-source-injected env vars to stdio servers
Surgical reapply of PR #37523 onto current main (the original branch
predates the SecretSource registry refactor). _build_safe_env() now
forwards env vars tagged in env_loader._SECRET_SOURCES — widened from
Bitwarden-only to any registered secret source (Bitwarden, 1Password,
plugin backends), since the provenance map is source-agnostic.
Explicit server env: config still wins; untagged secrets stay filtered.
Fixes#37499.
* fix(env): stop printing Bitwarden secret names
* fix(secrets): validate bitwarden status token
Keep the env-presence row, but add a real Bitwarden probe so revoked or malformed tokens no longer look healthy in hermes secrets bitwarden status.
Also document the new status behavior and lock it in with a dedicated regression test.
Refs: NousResearch/hermes-agent#40275
Tested: ./scripts/run_tests.sh tests/hermes_cli/test_bitwarden_status.py tests/test_bitwarden_secrets.py
Tested: .venv/bin/python -m ruff check hermes_cli/secrets_cli.py tests/hermes_cli/test_bitwarden_status.py
* fix(secrets): mark _APPLIED_HOMES only after a real fetch attempt (#40597) (#69056)
_apply_external_secret_sources() added the home to _APPLIED_HOMES before
loading config, so a malformed config.yaml, a missing secrets section, or
all-sources-disabled permanently disabled secret loading for the process
— even after the user fixed the config. Long-lived processes (gateway)
never recovered without a restart.
Now the home is marked only after apply_all() actually ran with at least
one enabled source. Fetch errors still mark the home (so import-time
load_hermes_dotenv() calls don't re-fetch and re-print the same failure
3-5x per startup); the cheap early-exit paths stay retryable.
Fixes#40597.
* fix(secrets): fall back to os.environ on scope miss when multiplexing is off
fdab380a1 wraps every cron job in a <home>/.env secret scope regardless of
deployment mode. get_secret() treats any installed scope as authoritative,
so in single-profile deployments where provider keys live only in the
process environment (systemd Environment=, pass-cli/op run wrappers, shell
exports) every cron credential read returns empty, the OpenAI client is
built with the no-key-required placeholder, and each scheduled job 401s —
while interactive turns keep working. Scope-miss reads now fall through to
os.environ when multiplexing is off; multiplexed scopes stay authoritative.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* test(gateway): activate multiplex flag in cross-profile env isolation test
The test installs a secret scope and asserts a scope miss does NOT fall
back to the default profile's env — that isolation guarantee only holds
under multiplexing, which the real gateway activates at startup via
set_multiplex_active(). With the #67827 overlay fallthrough (scope miss
→ os.environ when multiplex is OFF), the test needs to model the
multiplexed runtime it is actually testing.
* feat(secrets): orchestrator-level preserve_existing + profile aliasing (#69058)
Fixes the profile-clobber bug cluster at the apply_all() chokepoint so
every secret source — bundled and plugin — gets both behaviors for free:
- secrets.preserve_existing (#58073): env var names whose existing .env /
shell value always wins, even against a source with
override_existing: true. Escape hatch for per-profile platform
secrets while everything else rotates centrally.
- Profile aliasing (#51447): under a named profile, an applied
FOO_<PROFILE> var (credential-shaped suffixes only) also hydrates the
canonical FOO, so adapters/plugins that read fixed env names see the
profile's value. Direct supply beats alias; protected/claimed/
override guards all apply; secrets.profile_alias: false disables.
Reimplements the intent of PR #58085 (tianma-if, preserve_existing on the
legacy Bitwarden apply shim) and PR #51616 (LeonSGP43, profile aliasing
inside the Bitwarden backend) on the SecretSource orchestrator that
superseded those code paths.
Fixes#58073. Fixes#51447.
Co-authored-by: tianma-if <5895871+tianma-if@users.noreply.github.com>
Co-authored-by: LeonSGP43 <154585401+LeonSGP43@users.noreply.github.com>
* test(state): accept FTS5-specific corruption message in read-probe test
With the v23 external-content FTS layout, corrupt shadow segments surface
as 'fts5: corrupt structure record for table ...' instead of the generic
'database disk image is malformed'. Same corruption class, same variance
already documented and handled by _is_fts_write_corruption_error — widen
the test assertion to match.
---------
Co-authored-by: ethernet <arilotter@gmail.com>
Co-authored-by: brooklyn! <brooklyn.bb.nicholson@gmail.com>
Co-authored-by: Gille <4317663+helix4u@users.noreply.github.com>
Co-authored-by: yoniebans <jonny@nousresearch.com>
Co-authored-by: Rod-fernandez <rodrigo@nxtlevelsaas.com>
Co-authored-by: David Andrews (LexGenius.ai) <david@lexgenius.ai>
Co-authored-by: xxxigm <54813621+xxxigm@users.noreply.github.com>
Co-authored-by: hermes-seaeye[bot] <307254004+hermes-seaeye[bot]@users.noreply.github.com>
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: SHL0MS <SHL0MS@users.noreply.github.com>
Co-authored-by: HexLab98 <liruixinch@outlook.com>
Co-authored-by: kshitijk4poor <82637225+kshitijk4poor@users.noreply.github.com>
Co-authored-by: PRATHAMESH75 <prathamesh290504@gmail.com>
Co-authored-by: x7peeps <xtpeeps@qq.com>
Co-authored-by: Imgaojp <6065749+Imgaojp@users.noreply.github.com>
Co-authored-by: Siddharth Balyan <52913345+alt-glitch@users.noreply.github.com>
Co-authored-by: Ben Barclay <ben@nousresearch.com>
Co-authored-by: sbe27 <283218367+sbe27@users.noreply.github.com>
Co-authored-by: TARS <tars@users.noreply.github.com>
Co-authored-by: Austin Pickett <pickett.austin@gmail.com>
Co-authored-by: Rudimar Ronsoni <rudimar@outlook.com>
Co-authored-by: Austin Pickett <austinpickett@users.noreply.github.com>
Co-authored-by: Joshua <joshua@amokk.net>
Co-authored-by: alelpoan <155192176+alelpoan@users.noreply.github.com>
Co-authored-by: liuhao1024 <sunsky.lau@gmail.com>
Co-authored-by: Cossackx <121278003+Cossackx@users.noreply.github.com>
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Co-authored-by: Jaret Bottoms <jaretbottoms@gmail.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: Enough1122 <chenjin@hermes.local>
Co-authored-by: Frowtek <frowte3k@gmail.com>
Co-authored-by: Aniruddha Adak <aniruddhaadak80@users.noreply.github.com>
Co-authored-by: eazye19 <eazye19@users.noreply.github.com>
Co-authored-by: frohsinnllc <231045016+frohsinnllc@users.noreply.github.com>
Co-authored-by: hanyu1212 <28571259+hanyu1212@users.noreply.github.com>
Co-authored-by: 雨哥 <hanyu1212@users.noreply.github.com>
Co-authored-by: yungchentang <46495124+yungchentang@users.noreply.github.com>
Co-authored-by: trevorgordon981 <trevorbgordon@gmail.com>
Co-authored-by: Börje <borje@dqsverige.se>
Co-authored-by: fazerluga-creator <fazerluga@gmail.com>
Co-authored-by: yoma <yingwaizhiying@gmail.com>
Co-authored-by: Kennedy Umege <kenmege@yahoo.com>
Co-authored-by: pierrenode <298902573+pierrenode@users.noreply.github.com>
Co-authored-by: Sora-bluesky <sora.bluesky.dev@gmail.com>
Co-authored-by: iamwongeeeee <wykim777@naver.com>
Co-authored-by: web3blind <264741654+web3blind@users.noreply.github.com>
Co-authored-by: Bartok9 <danielrpike9@gmail.com>
Co-authored-by: anoopmehendale-cue <anoop.mehendale@gmail.com>
Co-authored-by: tinetwork <martin@tinetwork.com>
Co-authored-by: cucurigoo <241698038+cucurigoo@users.noreply.github.com>
Co-authored-by: McHermes <mchermes@edu.dreamcatcher.ai>
Co-authored-by: Nyaruko <nyaruko@hermes>
Co-authored-by: Hao Zhe <haozhe4547@gmail.com>
Co-authored-by: Naveen Fernando <90832919+NPFernando@users.noreply.github.com>
Co-authored-by: Chris Korhonen <ckorhonen@gmail.com>
Co-authored-by: Flownium <157689911+itsflownium@users.noreply.github.com>
Co-authored-by: kshitij <kshitij@kshitij.dev>
Co-authored-by: JackJin <1037461232@qq.com>
Co-authored-by: briandevans <252620095+briandevans@users.noreply.github.com>
Co-authored-by: menhguin <17287724+menhguin@users.noreply.github.com>
Co-authored-by: Minh Nguyen <menhguin@users.noreply.github.com>
Co-authored-by: SeoYeonKim <28585885+westkite1201@users.noreply.github.com>
Co-authored-by: andrexibiza <84248988+andrexibiza@users.noreply.github.com>
Co-authored-by: izumi0uu <izumi0uu@gmail.com>
Co-authored-by: Soju06 <qlskssk@gmail.com>
Co-authored-by: tianma-if <5895871+tianma-if@users.noreply.github.com>
Co-authored-by: LeonSGP43 <154585401+LeonSGP43@users.noreply.github.com>
Surgical reapply of PR #37523 onto current main (the original branch
predates the SecretSource registry refactor). _build_safe_env() now
forwards env vars tagged in env_loader._SECRET_SOURCES — widened from
Bitwarden-only to any registered secret source (Bitwarden, 1Password,
plugin backends), since the provenance map is source-agnostic.
Explicit server env: config still wins; untagged secrets stay filtered.
Fixes#37499.
Builds on diffen77's #66547 (cherry-picked as the previous commits).
Extend _is_session_expired_error's iterative traversal to follow
__cause__/__context__ in addition to ExceptionGroup .exceptions — SDK
wrappers often raise a generic RuntimeError *from* the message-less
ClosedResourceError, leaving the transport signal reachable only via
the chain. The identity-visited set guards chain cycles (handlers
re-raising previously seen exceptions), and a bounded node budget
(_EXC_TRAVERSAL_MAX_NODES) caps pathological acyclic graphs.
Adds regression tests: cause/context chain detection, interruption
precedence through chains, cyclic cause/context termination, and
budget-bounded termination.
Builds on trevorgordon981's #50589 (cherry-picked as the previous commit).
The #50394 cooldown reset only ran inside the async _shutdown coroutine,
which is skipped on the empty-_servers fast path — the most common state
when a server failed to connect (failed servers are never recorded in
_servers). It was also skipped when the MCP loop wasn't running.
Clear _server_connect_retry_after/_server_connect_failures on the fast
path and in a final unconditional sweep so a full shutdown/restart always
re-attempts every configured server immediately. Adds regression tests
for both paths.
Log-storm hygiene for the reconnect machinery (#65673, #66092):
- State transitions carry exactly one WARNING each:
connected → degraded (keepalive failure), degraded → parked
(budget exhausted / rapid-drop park / permanent error),
parked → connected (revival proven healthy, via
_mark_session_proven).
- Per-attempt retry logs (initial-connect attempts, connection-lost
reconnect attempts, per-cycle rebuild notices, park-wake probes)
demoted to DEBUG.
- Backoff sleeps get +/-20% uniform jitter (_jittered) so a herd of
servers that lost the same backend doesn't retry — and log — in
lockstep.
- Add module-level _unwrap_exception_group (ported from
hermes_cli/mcp_config.py, adapted): handles nested groups, prefers
non-cancellation leaves over the CancelledErrors anyio sprays across
sibling tasks, and re-raises KeyboardInterrupt/SystemExit leaves.
- Add _classify_mcp_failure(exc) -> 'permanent'|'transient'.
Permanent: auth 401/403, NonMcpEndpointError, InvalidMcpUrlError,
FileNotFoundError/ENOENT on the stdio command. Transient: everything
else (network/EOF/ClosedResource/TaskGroup drops). Permanent failures
park immediately without burning the retry ladder — every retry
against a missing binary or a revoked credential hits the same wall.
- Every log site in run() and the keepalive failure log now logs the
unwrapped root cause as 'TypeName: message', so a dead stdio pipe
says 'BrokenPipeError' instead of the opaque
'unhandled errors in a TaskGroup (1 sub-exception)' (and empty
str(exc) dead-pipe errors are no longer blank).
Harden the immediate-reconnect path from #66271:
- _reconnect_or_reraise_group re-raises when the transport TaskGroup
carries a KeyboardInterrupt / SystemExit leaf — fatal signals must
propagate to the interpreter, never be converted into a reconnect.
- Rapid-drop budget: a completed handshake alone no longer clears
_reconnect_retries/backoff on clean transport return. A session is
UNPROVEN until it demonstrates real health — survived >=1 full
keepalive interval (keepalive success path) or served >=1 successful
tool call (_mark_session_proven). Unproven clean returns are charged
against _MAX_RECONNECT_RETRIES, so a flapping transport that
handshakes fine and drops moments later still reaches the park
instead of respawning forever (#62212: 6212 spawns in 63h).
Proven sessions keep the #57604 behaviour: budget clears, transient
blips over a long-lived session never accumulate toward parking.
Streamable-HTTP / SSE MCP transports run their stream pump inside an anyio
TaskGroup. A transient stream drop (idle timeout, brief backend blip,
server-side TCP close) surfaces as a BaseExceptionGroup escaping the
transport context manager. It reached run()'s error path, which applied
exponential backoff (1s..16s) and eventually parked the server for 300s and
deregistered its tools — turning a sub-second glitch into a multi-minute
tool outage even though the POST path was still healthy.
_run_http now wraps all three transport branches (SSE, new + deprecated
Streamable-HTTP) and routes a BaseExceptionGroup through a new
_reconnect_or_reraise_group() helper that returns 'reconnect' (immediate
rebuild, no backoff/park/deregister). It re-raises instead of masking when
shutdown is in progress, when the group carries a real CancelledError
(cancellation must propagate, cf. #9930), or when no live session was
established this attempt (a connect/handshake failure that should fall
through to run()'s backoff rather than hot-loop).
Fixes#66092
Co-authored-by: 雨哥 <hanyu1212@users.noreply.github.com>
Adds an allow_session flag to the gateway approval payload so adapters
can render the session tier independently of the permanent tier. Matrix
gains a session reaction (🌀) and a reaction legend; pure-tirith prompts
now offer once/session/deny instead of collapsing to once/deny.
Salvaged from PR #67312, adapted to the allow_permanent semantics that
landed in #68597 (Always offered when any dangerous-pattern warning is
persistable; pure-tirith prompts stay session-max).
Three related messaging-approval fixes:
1. approvals.timeout default 60 -> 300. PR #63501 collapsed the gateway
wait onto the canonical approvals.timeout (previously
gateway_timeout=300), silently shrinking messaging approval windows
to 60s. Push-notification approvals routinely arrive later than a
minute; taps landed after the wait had already failed closed.
2. Stale-tap honesty: adapters resolved the approval AFTER rendering
'<checkmark> Approved by <user>' (Telegram/Discord/Slack), or ignored a zero
resolve count (WhatsApp Cloud/Feishu). A tap on an expired prompt
claimed approval while the command had already been denied. All
button paths now resolve first and render 'Approval expired -
command was not run' when nothing was waiting.
3. Mixed-warning prompts (dangerous pattern + tirith finding) now offer
Always: the persistence layer already permanently allowlists the
pattern key and downgrades the tirith key to session scope, but the
UI hid Always whenever ANY tirith warning was present. Pure-tirith
prompts still withhold Always (content findings are session-max by
design), and Smart-DENY overrides remain once-only.
Follow-up to #67640: move the agent.file_safety import to module top
(stdlib-only, no circular-import concern), replace the over-broad
except Exception + logger.warning with an import sentinel plus
logger.exception so a guard failure is debuggable instead of silently
swallowed. Adds fail-closed tests asserting the diagnostic is emitted.
The live transcripts added with delegate_task write each child's events to
<hermes_home>/cache/delegation/live/<delegation_id>/. Nothing on that path is
redacted, and the rendered events are precisely the secret-bearing surfaces:
tool args, tool results and streamed assistant text.
That location is not incidental — delegate_tool.py:1620 documents cache/
delegation as "mounted read-only into remote backends", so every line lands
in a file readable from inside the sandbox.
Observed on the writer today:
tool | -> terminal(curl -H "Authorization: Bearer sk-ant-api03-...")
result | terminal ok 0.4s: OPENAI_API_KEY=sk-proj-... AWS_SECRET_ACCESS_KEY=wJalr...
The same three values through the canonical redactor:
curl -H "Authorization: Bearer ***" https://api.internal
OPENAI_API_KEY=*** AWS_SECRET_ACCESS_KEY=***
Every other sink for this data already routes through that redactor — search
results via redact_sensitive_text(file_read=True), terminal output via
redact_terminal_output — so the transcript was the one place an operator's
keys reached disk in the clear.
Redact at three points, covering every artefact the dispatch writes into that
directory:
* event() — every typed helper (assistant_text, thinking, tool_start,
tool_result, marker, finalize, the stream flush) funnels through it, so
one call covers them all and a helper added later cannot bypass it.
* the .log header — written directly rather than through event(), and a
caller can paste a key into the task text.
* manifest.json — _write_manifest serialises the same goal, and the manifest
sits in the same mounted directory, so redacting only the header would
have left the credential exposed one file over.
force=True because this is a safety boundary and must redact even with the
global toggle off. On the (impossible-in-practice) import failure the line is
withheld instead of written raw: losing a debug line costs less than writing
a live credential into a sandbox-readable file.
Key names, tool names, statuses, durations and ordinary prose are untouched,
so tail -f stays as useful as before — pinned by its own test, alongside a
whole-directory sweep asserting no file under live/<id>/ carries the raw key.
register_credential_file() takes a skill-declared relative path from
required_credential_files frontmatter and bind-mounts it read-only into the
remote sandbox the skill's own code runs in. It validates that the resolved
path stays inside HERMES_HOME — the docstring names the threat directly:
so that a malicious skill cannot declare
required_credential_files: ['../../.ssh/id_rsa'] and exfiltrate
sensitive host files into a container sandbox
Containment is the wrong boundary on its own, because HERMES_HOME is exactly
where the master credential stores live. Traversal is blocked; asking for the
keys by name is not:
skill declares mounted? agent may read it?
.env YES DENIED
auth.json YES DENIED
.anthropic_oauth.json YES DENIED
cache/bws_cache.json YES DENIED
mcp-tokens/srv.json YES DENIED
google_token.json YES allowed
../../.ssh/id_rsa no n/a
Every row marked DENIED is refused by the canonical read guard
(agent.file_safety.get_read_block_error) — the agent cannot read_file them —
yet one line of hub-installed skill frontmatter gets them bind-mounted where
that skill can cat them. .env alone is every provider API key.
Reuse the canonical deny-list as the mount bar: what the agent is forbidden
to read is not mountable either, so the mount surface cannot hand a skill
what the read surface denies it. Fails CLOSED — if the guard can't be
consulted the mount is refused rather than risked.
The module keeps doing its job: a skill still mounts its own service token
(google_token.json, skills/*), and a refused entry is reported back through
register_credential_files' missing list instead of failing the batch.
The three prior PRs here (#3946, #3951, #4316) all hardened traversal; this
closes the half that traversal validation never covered.
An orphaned pip descendant holding the probe's inherited stdout/stderr
pipe handles wedged subprocess.run's post-timeout communicate() (which
joins the pipe reader threads with NO timeout on Windows). The warm
probe thread then hung holding the module-level _CACHE_LOCK, so every
new session's prompt build blocked indefinitely (#67964).
Two layers:
- _run(): replace subprocess.run with Popen + communicate(timeout); on
TimeoutExpired kill the process TREE (taskkill /T on Windows) and
reap the direct child bounded — never re-read the pipes, so an
orphaned descendant holding them open can't block us.
- get_environment_probe_line(): the probe now runs in a single
background worker publishing via a threading.Event; callers wait at
most _PROBE_WAIT_TIMEOUT (10s) then fail open with "". After one
full timeout, later callers only peek. If the stuck worker ever
finishes, its line resumes appearing in new prompts.
Regression tests: hung probe with 4 concurrent callers returns bounded;
late recovery publishes; repeat callers skip the wait; _run() returns
promptly despite a pipe-holding descendant (real subprocess E2E).
Fixes#67964
judge_goal() returns (verdict, reason, parse_failed, wait_directive) since
the goals.py wait-directive change, but the kanban goal-mode completion gate
at tools/kanban_tools.py still unpacked 3 values. Every judge call raised
ValueError, the defensive except swallowed it, and the pre-initialized
verdict='done' let every completion through — the acceptance gate was
silently disabled.
Now unpacks all 4 values; the test mock is updated to match the real
contract. The other two judge_goal consumers (hermes_cli/goals.py) already
use 4-value unpacks.
Reported and diagnosed by @bill3wits in PR #57276; reimplemented under
project authorship because the original commit was authored under a
non-existent local identity (bash@hermes.local) that cannot be carried
into history. Also fixes#58066 (duplicate report by @Gibcity).
Follow-up for salvaged PR #39946: file_sync.py already aliases time.sleep
as _sleep specifically to avoid tests mutating the shared stdlib module
object. Apply the same convention to the rate-limit clock (_monotonic)
and point the new regression test at it.
FileSyncManager.sync() is rate-limited to once per _sync_interval via
_last_sync_time, and its docstring promises that on failure "state rolls
back so the next cycle retries everything". But the except handler also
set _last_sync_time = time.monotonic() on failure, so the next non-forced
sync() within the interval hit the rate-limit guard and returned early —
suppressing the retry the rollback had just prepared.
Because the non-forced sync() runs before every command on the SSH, Modal
and Daytona backends, a single transient upload failure (network blip,
dropped channel) left the remote with stale files for the next command
(up to _sync_interval, default 5s). Forced syncs bypass the guard, which
is why it was intermittent.
Remove the failure-path timestamp bump so the clock only advances on a
successful or no-op cycle, matching the documented contract. Add a
regression test that fails before this change and passes after.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The salvaged retention check compared the built-in memory snapshot
before vs after the disk reload. That holds for a long-lived CLI agent,
but on fresh-agent surfaces (gateway per-turn agents, TUI) the cached
prompt is restored from the session DB and can predate mid-session
memory writes that the fresh MemoryStore already absorbed at init: the
snapshot is then identical on both sides of the reload while the prompt
itself is stale, so compression would retain (and re-persist via
update_system_prompt) a prompt missing the new memory for the life of
the session.
Replace the equality check with a containment check
(_cached_prompt_reflects_builtin_memory): retain the cached prompt only
when the freshly-reloaded rendered blocks appear verbatim inside it,
and rebuild when a leftover block header remains for a target whose
entries have since been emptied or disabled. Block headers are shared
via MEMORY_BLOCK_HEADERS in tools/memory_tool.py so the check stays in
lockstep with MemoryStore._render_block.
Adds regression guards for the gateway stale-restore path and the
emptied-memory leftover-block path; verified with a real-MemoryStore
E2E matrix (9 scenarios) against a temp HERMES_HOME.
grok-4.5 is xAI's newest release (their versioning is non-monotonic:
4.5 > 4.20) and is the model xAI's own docs use for the server-side
x_search tool. Users who explicitly pinned x_search.model keep their
choice; everyone else picks up the new default via the config
deep-merge — no _config_version bump needed.
- tools/x_search_tool.py: DEFAULT_X_SEARCH_MODEL
- hermes_cli/config.py: DEFAULT_CONFIG x_search.model + comment
- agent/reasoning_timeouts.py: 300s stale-timeout floor entry for
grok-4.5 (grok-4.20-reasoning entry kept for pinned users)
- docs: x-search.md en + zh-Hans (config sample + troubleshooting)
- tests: default-model assertion + timeout-floor positive case
* feat(delegation): live-viewable subagent transcripts for delegate_task
Each child now streams an append-only, human-readable log to
<hermes_home>/cache/delegation/live/<delegation_id>/task-<n>.log while it
runs, and the dispatch return includes the paths so the caller can tail
them immediately instead of waiting blind for the consolidated summary.
- New tools/delegation_live_log.py: LiveTranscriptWriter (per-event append
+ flush, one-line rendering with truncation, never raises into the agent
loop), wrap_progress_callback (tees the child's existing
tool_progress_callback events into the log, preserves the _flush
contract), dispatch-time creation with pre-headered files so tail -f
attaches immediately, manifest.json (goals/task count/per-task status),
and 7-day retention pruning on new dispatches.
- delegate_task: wraps each child's progress callback with the writer;
sync results and background dispatch responses gain live_transcripts
(+ hint field on dispatch); per-task result entries carry
live_transcript; transcripts finalized with exit-reason markers.
- async_delegation: dispatch_async_delegation_batch accepts an optional
delegation_id so the live/ dir name matches the returned handle; the
completion event carries live_transcripts.
- process_registry: consolidated batch-completion block references each
task's live transcript path.
- Tool schema description documents the live_transcripts return surface;
docs gain a 'Live Transcripts' section with a tail -f example.
Placement under cache/delegation means the logs are mounted read-only
into remote terminal backends for free. Side-channel only: zero changes
to message content, so prompt caching is unaffected. Transcript-OUT only
— no overlap with the subagent control surfaces of PR #66046.
* fix(delegation): label the kickoff transcript line as user — it is the child's one user message
Bug 1 of #55048: when the MCP connection dropped (driver crash / restart),
_lifecycle_coro exited but left _started=True, so the next list_apps/capture
passed _require_started() and then operated on a None session — hanging
forever instead of reconnecting.
- _lifecycle_coro's finally now resets _started=False on ANY exit, so a dead
session is re-enterable (idempotent no-op on the normal stop() path; atomic
bool write, safe from the bridge-loop thread without the lock stop() holds).
- call_tool() re-enters start() when the session isn't active, rebuilding it
before the call. The start_session/end_session handshake (driven by start()/
stop() themselves) is exempted so bootstrap doesn't recurse.
Tests: two cases in test_computer_use_delivery_ladder.py — finally resets
_started, and call_tool restarts a dead session exactly once. Full
computer_use suite green (233).
Refs #55048 (Bug 1). Bug 2 (expose foreground dispatch) is covered by the
delivery_mode work in #67123.
Hermes' computer_use wrapper dropped cua-driver's structured action verdicts,
exposed no delivery_mode, and injected background-only guidance — so the agent
reported unverified no-ops as success and concluded cua-driver 'cannot drive'
Electron/Chromium surfaces (observed live on tldraw offline). Fixes#67052.
Phase A — preserve the result contract:
- ActionResult carries verified/effect/escalation/path/degraded/code/delivery_mode
- CuaDriverBackend._action() reads structuredContent (was data-only); a helper
normalizes it, additive and None-safe on old drivers
- _text_response surfaces the fields additively (ok stays transport-only)
Phase B — bounded, model-reachable foreground:
- delivery_mode (background|foreground) + bring_to_front on the schema, dispatcher,
ABC, and all input methods
- foreground is capability-gated (input.delivery_mode); old drivers get a
structured foreground_unsupported refusal, never a silent background downgrade
- no automatic/hidden foreground retry — the model selects it from the signal
Phase C — guidance + isolation:
- system prompt (prompt_builder) and bundled skills/computer-use/SKILL.md go from
background-ONLY to background-FIRST, teaching the AX→PX→foreground ladder driven
by returned effect/escalation, not predicted from the app being Electron
- foreground approval scoped by (action, delivery_mode): a background approval
never silently authorizes foreground
- approval state keyed per session_id so concurrent gateway runs don't leak unlocks
Tests: tests/tools/test_computer_use_delivery_ladder.py (15) cover confirmed/
unverifiable/suspected_noop/degraded/old-driver verdicts, delivery_mode gating +
foreground_unsupported, and session-scoped foreground approval. Existing 265
computer_use tests still green.
Live E2E (real cua-driver 0.8.3 + tldraw offline on Linux/X11): a background click
returned effect='unverifiable'/path='ax' (no fabricated success), and a foreground
request returned code='foreground_unsupported' — correct on a driver that predates
the input.delivery_mode capability.
The per-turn `_cleanup_task_resources` unconditionally calls
`cleanup_browser`, closing the browser window immediately after each
bot reply. This makes headed mode (`AGENT_BROWSER_HEADED=1`) unusable
— the window flashes up and disappears on every response.
This mirrors the existing VM persistence pattern: skip per-turn
cleanup when headed mode is active and let the inactivity reaper
handle idle sessions instead. Full-session teardown on gateway
shutdown remains unconditional.
Also adds `browser.headed` config.yaml support and passes `--headed`
to agent-browser in local mode when configured, so users don't need
to rely solely on the `AGENT_BROWSER_HEADED` env var.
Closes#11020 (lead bug)
Prevents the curator's LLM consolidation pass from archiving skills the
user placed manually (e.g. via URL install, direct SKILL.md authoring, or
Gitee source). These skills carry created_by=None in .usage.json rather
than created_by=agent, but the _background_review_write_guard only checked
pinned, external, bundled, hub, and protected built-in status — missing
the manual-skill case entirely.
The guard already caught a real case: the user's 'auto-dev' skill
(use_count=50, patch_count=119) was archived 28 seconds after its last
use during a curator auto-run.
Adds a check: if the skill has a usage record and its created_by is not
'agent', refuse the background curator write. Skills with no record at
all (new/unknown) are not blocked.
Use the direct POSIX parent relationship instead of process creation time and pid_exists checks. Remove the dead create-time argument chain while preserving process-group cleanup and signal forwarding.\n\nRefs #62505
A leaf subagent is meant to be denied delegate_task, execute_code, memory,
clarify, cronjob, and send_message. _strip_blocked_tools() only drops a
toolset when EVERY tool in it is blocked, so mixed platform bundles
(hermes-cli, hermes-telegram, and every other gateway bundle) survived
stripping and re-exposed the blocked tools after composite expansion. A
leaf child spawned from any gateway platform could recursively delegate,
run code, and write memory.
Pass exact one-tool deny toolsets into the child's disabled_toolsets so
model_tools subtracts the blocked names AFTER composite expansion, and the
restriction survives later registry/MCP refreshes. Orchestrators regain
only delegate_task.
Salvaged from #66036 by Mason Tanguay (@DictatorBacon); scoped to the
authority fix + its regressions (docs/interrupt changes dropped).
Co-authored-by: Teknium <127238744+teknium1@users.noreply.github.com>
run_agent._dispatch_delegate_task forces background=True for every top-level
delegation, and async_delivery_supported() returns True for any session that
never binds the capability. On runners that cannot receive a completion after
their turn ends, that combination silently discards every subagent result: the
model gets a dispatch handle, ends its turn, and reports 'waiting for results'.
Two such runners never bind the capability:
* hermes -z (one-shot) prints one final response and exits. It bypasses cli.py,
so nothing drains process_registry.completion_queue (only the interactive
process_loop and the gateway watchers do).
* cron run_job clears the HERMES_SESSION_* routing keys, so a completion event
carries session_key="" — _enrich_async_delegation_routing cannot resolve it
and _inject_watch_notification drops it ("no routing metadata"). By then
run_job has already shipped the job's final response via _deliver_result;
there is no turn left to re-enter. Worse, get_current_session_key() can fall
back to the ambient os.environ HERMES_SESSION_KEY, so a cron subagent's output
can be routed into an unrelated user chat rather than merely dropped.
Add declare_stateless_channel() and bind it in both runners, routing
delegate_task to its existing inline/synchronous path — the same fallback the
stateless HTTP adapter already relies on, and the fix suggested in #63142. The
helper binds only the capability: set_session_vars() would also latch
_session_context_engaged, which a pure single-process one-shot must not trigger.
Also correct two agent-facing strings that hardcoded 'stateless HTTP API' as the
only channel without async delivery (delegate_tool, terminal_tool); they now name
the actual condition.
Repro (before): hermes -z 'Use delegate_task to spawn a subagent that replies
BANANA. Report its reply.' -> "Waiting for the subagent's response...", exit 0,
no BANANA. After: BANANA is returned in-turn.
Fixes#53027Fixes#63142
Two internal comments in delegate_tool.py still described the superseded
"N independent handles, no combined wait" model, contradicting the
authoritative batch contract (one async unit, one consolidated result
when all children finish). Aligns the comments with the runtime path in
_execute_and_aggregate / dispatch_async_delegation_batch.
A root-launched CLI session can leak /root into the terminal cwd state a
non-root gateway/cron process later resolves (#65583). os.path.isdir('/root')
is True for a non-root user — stat only needs search permission on / — so
_resolve_safe_cwd returned it and subprocess.Popen(cwd='/root') died with
PermissionError: [Errno 13], failing EVERY cron job's terminal/file/search
tool on every command until restart.
_resolve_safe_cwd now requires X_OK (new _cwd_usable helper) and climbs to
the nearest enterable ancestor, logging a WARNING that names the leak class
when an existing-but-denied cwd is skipped. Missing-cwd recovery (#17558)
behavior unchanged.
E2E-verified: LocalEnvironment constructed with an unenterable cwd now runs
commands from the fallback directory instead of raising.
Per the MCP spec the cursor is an opaque string; anything else
(including MagicMock auto-attributes in tests) means no more pages.
Fixes test_mcp_tool_session_expired mock-session runaway.
Port from anomalyco/opencode#35439/#35500: preserve full MCP catalogs
across paginated tools/list responses.
The MCP spec allows servers to paginate tools/list, resources/list, and
prompts/list via an opaque nextCursor token. The Python SDK's
ClientSession.list_* methods fetch exactly one page per call, and hermes
never passed the cursor back — on a paginated server every tool,
resource, and prompt past page 1 was silently invisible to the agent.
Adds _paginate_full_list() (cursor-draining helper with a 50-page
runaway cap and spec-correct opaque-string cursor validation) and
applies it at all three discovery sites: _discover_tools(), the
tools/list_changed refresh handler, and the list_resources/list_prompts
utility handlers. The keepalive probe intentionally keeps its
single-page call (liveness only).
E2E: real stdio MCP server serving 3 pages (2/1/1 tools) — old code
discovered 2 tools, new code discovers all 4.
Port from openai/codex#31494: user-visible history replay must strip CSI
sequences and control characters. Stored conversation history can carry
raw terminal escapes (pasted content, gateway-origin text, model output
echoing injected tool results). Replaying it via /resume's recap panel or
build_recap (/status on CLI + gateway) wrote those bytes straight to the
terminal — an injected message could clear the screen, retitle the window,
move the cursor, or restyle the recap UI. Rich's Text() does not neutralize
raw escape bytes.
- tools/ansi_strip.py: add sanitize_display_text() — strip_ansi() plus
bare C0/C1 control removal, preserving \n and \t, normalizing \r to \n
(adapted to Python from Codex's sanitize_user_text; reuses the existing
ECMA-48 stripper instead of transcribing their char-walk)
- hermes_cli/cli_agent_setup_mixin.py: sanitize user + assistant text in
_display_resumed_history() before building the Rich recap panel
- hermes_cli/session_recap.py: sanitize preview lines in build_recap()
(_truncate choke point) so /status recaps are clean on every platform
- tests: 10 new sanitize_display_text cases (incl. the exact codex#31494
fixture), recap + resume-display leak assertions
The previous commit fixes the canonical agent/skill_utils.parse_frontmatter.
Six more modules reimplement the '---' fence check locally and had the
same bug:
- tools/skill_manager_tool.py _validate_frontmatter — rejected BOM'd
skill_manage create/edit content outright
- tools/skills_hub.py GitHubSource._parse_frontmatter_quick and
OptionalSkillSource._parse_frontmatter — hub browse/install metadata
- hermes_cli/skills_hub.py — local skill install validation
- gateway/run.py — skill slug discovery for disabled-skill hints
- agent/prompt_builder.py _strip_yaml_frontmatter — BOM'd context files
(AGENTS.md) leaked raw frontmatter into the system prompt
- tools/blueprints.py _split_frontmatter — str.lstrip() does not strip
U+FEFF (not whitespace), so the existing lstrip never covered it
Sibling-surface regression tests added.
Bug class also fixed upstream in cline/cline#12218 (found by the weekly
Cline PR scout).
* feat(browser): store full snapshots on truncation; make eval denylist opt-in
Two harness fixes motivated by BU_Bench results where fixed-verb + lossy
observation cost Hermes heavily vs code-driven browser agents:
1. Snapshot truncation no longer loses content. When a snapshot exceeds
the 8000-char threshold, the complete accessibility tree is saved to
cache/web (same truncate-and-store pattern as web_extract) and the
truncated view / LLM summary includes the file path plus a ready-made
read_file call. Element refs beyond the cut are recoverable without
re-snapshotting. Stored copies are force-redacted and capped at 2MB;
content-hash filenames dedupe repeated snapshots of the same page.
2. The browser_console(expression=...) sensitive-primitive denylist is
now opt-in via browser.restrict_evaluate (default false). The
names-based denylist blocked legitimate DOM extraction — any selector
or expression containing 'fetch', 'cookie', 'input', etc. — which
crippled the agent's only programmatic page-inspection path. The
SSRF/private-URL egress guards in _browser_eval are independent of
this policy and remain always-on. browser.allow_unsafe_evaluate keeps
its meaning (bypass the denylist) for configs that already set it.
* test: update None-guard test for stored-snapshot pointer in _extract_relevant_content
test_normal_content_returned pinned the exact return value; the summary
now carries a pointer to the stored full snapshot. Assert the summary
passes through and the pointer is present instead.
* feat(browser): align snapshot threshold with web_extract's 15k char budget
SNAPSHOT_SUMMARIZE_THRESHOLD 8000 -> 15000, matching
web_tools.DEFAULT_EXTRACT_CHAR_LIMIT so the snapshot and web_extract
truncate-and-store paths give the model the same per-page budget.
_truncate_snapshot's default max_chars now follows the constant.
Invariant test added; docs (en+zh) and CLI tip updated.
Keep each xAI OAuth auth-add login as an independent manual device-code pool entry and recognize xAI personal-team spending-limit 403 responses as billing exhaustion. Preserve the structured top-level error message so the failed credential is quarantined and the next healthy account is selected without attempting a pointless token refresh.
Route direct xAI HTTP consumers through the credential pool as well. Proactive and 401-reactive refreshes update the exact issuing manual entry, preserve validated xAI base URL overrides, and serialize single-use refresh-token rotation across concurrent pool instances.
terminal_tool reads all settings from TERMINAL_* env vars, bridged from
config.yaml by the CLI, gateway, and TUI-PTY launchers. Processes that
skip every launcher bridge — hermes serve / the Desktop app backend's
in-process agents, the desktop cron ticker — saw an unset TERMINAL_ENV
and silently ran every command on the host even when config.yaml selects
terminal.backend: docker. A user who configured Docker isolation got
unsandboxed host execution with no warning.
Two layers:
- _ensure_terminal_env_bridged() in _get_env_config(): when TERMINAL_ENV
is unset, backfill TERMINAL_* from config.yaml via
apply_terminal_config_to_env(override=False). Explicit env always wins
(honor explicit choice; only fix the accidental fallback). One-shot,
fail-open to the historical local default.
- cmd_dashboard/serve: run the same bridge at startup so every consumer
in the backend process (in-process agents, desktop cron ticker,
tui_gateway cwd resolution) sees the bridged env directly.
Fixes#63141, #54449, #61115, #65696.
_download_url_with_cap called urlopen() after only a scheme check, so a
model-controlled URL could reach loopback services, RFC1918/CGNAT hosts,
or cloud metadata endpoints (169.254.169.254), and a public host could
302 to any of those unvalidated.
Route the fetch through the repo's canonical SSRF guard instead:
validate every hop with tools.url_safety.is_safe_url() and follow
redirects manually (httpx, follow_redirects=False, 5-hop limit) so each
Location target is re-checked before it is fetched — the same pattern
as tools/skills_hub._guarded_http_get. The streaming size cap is
unchanged. Local-fixture tests opt in via HERMES_ALLOW_PRIVATE_URLS
(the guard's documented escape hatch); new tests pin rejection of
loopback, cloud-metadata, and private-range URLs, a mocked
public→loopback redirect, and a mocked public happy path.
The salvaged attachment-toolset commit predated main centralizing the
25 MB cap as kanban_db.KANBAN_ATTACHMENT_MAX_BYTES and re-introduced a
private _MAX_ATTACHMENT_BYTES alias. Drop the duplicate: kanban_db's
store_attachment_bytes(), the dashboard upload endpoint, and the
kanban_attach_url tool all reference the one shared constant now, and
the tests monkeypatch that same name.
The kanban board has had full attachment storage and a dashboard HTTP
API (upload/list/download/delete) since #35338, but there was no agent
toolset tool and no `hermes kanban` CLI verb for attachments. Agents and
scripts that don't go through the dashboard server (or can't touch the DB
directly) had no way to create or read real attachments — only links in
comments.
Close that gap by mirroring the existing comment surface:
- `kanban_db.store_attachment_bytes()` — one shared write path (validate
name, enforce the 25 MB cap, write the blob under the per-task dir with
collision-free naming, insert the metadata row, clean up an orphan blob
if the insert fails). `_MAX_ATTACHMENT_BYTES`, `_safe_attachment_name`,
and a new `_collision_free_path` move here so the dashboard, the tool,
and the CLI all share one implementation and can't drift.
- Tools (`tools/kanban_tools.py`): `kanban_attach` (inline base64),
`kanban_attach_url` (server-side http/https fetch with the same cap),
`kanban_attachments` (list). Write tools respect worker task-ownership;
list is read-only. Registered in the `kanban` toolset.
- CLI (`hermes_cli/kanban.py`): `attach <id> <path>`, `attachments <id>`,
`attach-rm <attachment_id>`.
- Dashboard `upload_task_attachment` now imports the shared helpers and
uses `_collision_free_path` — behavior identical (still streams to disk
with the cap, still 413 on overflow).
- Docs (AGENTS.md, kanban-worker skill) and toolset membership updated.
Tests: tool round-trip + oversize + bad base64 + ownership; attach_url
against a local HTTP fixture incl. oversize-mid-stream and non-http
scheme rejection; CLI attach/attachments/attach-rm; shared-helper unit
tests; dashboard parity preserved.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Same bug class as the salvaged #65305/#65307: hmac.compare_digest (and
secrets.compare_digest) raise TypeError when given a str containing
non-ASCII characters, and these call sites feed it raw request input.
Compare as UTF-8 bytes everywhere:
- gateway/platforms/msgraph_webhook.py: clientState from request body
- gateway/platforms/whatsapp_cloud.py: hub.verify_token query param +
X-Hub-Signature-256 header (comment claimed 'works on str' — it
doesn't for non-ASCII)
- plugins/platforms/feishu: verification token + x-lark-signature
- plugins/platforms/raft: bridge token header
- plugins/platforms/line: X-Line-Signature
- plugins/platforms/sms: X-Twilio-Signature
- tools/code_execution_tool.py: sandbox RPC token (both loops)
Regression tests for the two gateway-core sites (msgraph, whatsapp).
_wait_for_callback still read the legacy module-level _oauth_port, so
with two concurrent OAuth flows, flow A's callback wait bound flow B's
port while A's redirect URI pointed at A's port — the callback-side
half of the cross-flow collision that #65622 fixed on the redirect
side. _make_callback_waiter(port) closes over each flow's resolved
port; both provider construction sites (build_oauth_auth and
MCPOAuthManager._build_provider) now wire per-flow waiters. The legacy
_wait_for_callback delegates for backwards compatibility.
Direction credit to @LeonSGP43 (#34280) and the #34260 analysis.
_wait_for_callback catches OSError on bind with a comment claiming the port is
held by a server build_oauth_auth started, and promising to fall back to polling
it. build_oauth_auth never starts a callback server (this is the only listener),
so there is nothing to poll: the branch just raised a misleading "OAuth callback
timed out" when the real cause is a busy port (a concurrent login, a leftover
listener, or a fixed oauth.redirect_port that collided).
Fix the stale comment and raise an accurate, actionable message (names the port,
suggests freeing it or setting a free oauth.redirect_port), chained from the
original OSError. Behavior is otherwise unchanged. Adds a regression test.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Reclaim.ai's AWS API Gateway WAF 403s any /oauth2/authorize request whose
query string contains a literal 127.0.0.1, so the SDK's hardcoded
redirect_uri made the browser flow impossible. New optional oauth config
key redirect_host (default 127.0.0.1, unchanged behavior) lets a server
entry use localhost instead.
Integrated into _resolve_redirect_uri so it composes with redirect_uri:
an explicit redirect_uri wins; redirect_host only rewrites the loopback
default's hostname.
The PR added a configurable `redirect_uri` (proxy/Funnel callbacks) but
shipped without tests, and the loopback SSH-tunnel hint stayed hardcoded —
actively misleading the exact proxy user the feature targets.
- Extract `_resolve_redirect_uri(cfg, port)` so the client-metadata and
pre-registration paths derive an identical callback (a mismatch makes the
authorization server reject the redirect).
- Make `_redirect_handler` redirect_uri-aware: a configured proxy callback
reaches this machine on its own, so it no longer prints the `ssh -N -L`
loopback guidance. Wired via `functools.partial` — no new global state.
- Document `redirect_uri` in the config block.
- 14 new tests (red/green TDD): helper resolution + empty-string fallback,
metadata + pre-registration for configured/default, AnyUrl normalization,
no-client_id skip, client_secret combo, and both SSH-hint branches.
ruff clean · 81 passed (tests/tools/test_mcp_oauth.py) · ty baseline unchanged
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Per egilewski's security review, WEIXIN_BASE_URL and WEIXIN_CDN_BASE_URL
were still resolved from process-global environment variables, leaving
mixed-scope bypasses in multiplex mode.
Changed files:
- gateway/platforms/weixin.py: Added get_secret import, replaced os.getenv()
with get_secret() for WEIXIN_ACCOUNT_ID, WEIXIN_TOKEN, WEIXIN_BASE_URL,
WEIXIN_CDN_BASE_URL in WeixinAdapter.__init__() and send_weixin_direct()
- tools/send_message_tool.py: Added get_secret import, replaced os.getenv()
with get_secret() for all WEIXIN_* fallbacks in _handle_send()
All runtime Weixin send paths now resolve both credentials and endpoint
configuration from the same profile-scoped source.
_find_free_port() closed its probe socket before HTTPServer re-bound
the port minutes later, leaving a window where another process could
steal it (#22161 by @amathxbt). _reserve_callback_port() now keeps the
selected socket bound (bounded FIFO pool) until _wait_for_callback
adopts it via bind_and_activate=False. Also sets allow_reuse_address
BEFORE binding — the cherry-picked #44872 set it after the constructor
had already bound, where it is a no-op.
Also updates the three #57836 non-interactive-guard tests to the
closure-factory API from #44872.
Two related OAuth fixes:
1. Replace module-level _redirect_handler with _make_redirect_handler()
closure factory that closes over the resolved port. This prevents
cross-server state pollution when multiple MCP servers run OAuth
concurrently (#44588).
2. Set server.allow_reuse_address = True on the ephemeral callback
HTTPServer so the socket doesn't stay in TIME_WAIT after the flow
completes. This prevents 'Address already in use' errors on the
next OAuth flow for the same port (#44590).
Fixes#44588Fixes#44590
execute_code's _resolve_child_cwd() only checked the process-global
TERMINAL_CWD env var and os.getcwd(), ignoring the per-session cwd
override registered via session.cwd.set → register_task_env_overrides.
This caused execute_code to write to the process launch directory
while sibling tools (write_file, read_file, patch, terminal) correctly
resolved the session workspace — two file-writing paths in one turn
silently disagreed on the working directory.
Fix: pass task_id to _resolve_child_cwd() and check
_registered_task_cwd_override(task_id) before falling back to
TERMINAL_CWD and os.getcwd(), matching the lookup order used by
file_tools._resolve_base_dir and terminal_tool._resolve_command_cwd.
Follow-up for salvaged PR #63255: with LocalEnvironment._update_cwd
delegating to the stdout marker parser, the cwd temp file has zero
readers left. Drop the 'pwd -P > file' writes from the bootstrap and
_wrap_command so every command stops paying a pointless file write
(and stops littering temp dirs with hermes-cwd-*.txt).
Follow-up for salvaged PR #26790: on a POSIX host with _IS_WINDOWS
patched (test simulation), os.path.isabs rejects C:\Users\x and the
new relative-cwd recovery would mangle a perfectly absolute native
Windows path. Check ntpath.isabs first on the Windows branch.
The per-session record store is now the ONLY cwd mechanism. Deleted:
- env.cwd_owner stamping + prev_owner threading (terminal_tool): the
shared env no longer carries ownership metadata at all
- _resolve_command_cwd's env/prev_owner params: resolution is
workdir > session record > config/override default
- file_tools._live_cwd_if_owned + _get_live_tracking_cwd: path
resolution never consults the shared env's live cwd
- file_tools._last_known_cwd + _remember_last_known_cwd +
_last_known_cwd_for: the #26211 preserved-anchor registry is
subsumed by the session record, which never lived on the env and
therefore cannot be lost to env cleanup. The _get_file_ops
stale-cache rescue now writes the record instead.
- env recreation (both _get_file_ops and terminal_tool) seeds the
fresh env from override > session record > config
Why no transition fallback: the legacy state was process-local and
in-memory exactly like the record store — after a restart both start
empty, and within a running process every legacy write site has been
dual-writing the record since step 1. There is no populated-legacy/
empty-record state to fall back for.
Tests updated to drive the record store instead of the deleted
mechanism; the cross-session isolation suite now asserts the same
behavior contracts (no leak, cd isolation, #26211 persistence)
against the new architecture, plus a new "session C inherits nothing"
case that the old ownership guard could not express.
Third step of the cwd rearchitecture: _resolve_command_cwd now prefers
the session's own cwd record over the shared env's live cwd.
New resolution order: workdir > session record > legacy env.cwd
(ownership-gated, transition-only) > config/override default.
The record is written after every completed command for the session, so
it IS the session's cd state — another session's cd lands in another
record and cannot affect this session's commands. The legacy env.cwd
branch only fires for a session with no record yet (no command has
completed since this code loaded); it keeps the prev_owner ownership
guard for that transition window and is deleted in step 4 along with
env.cwd_owner stamping and file_tools' _last_known_cwd machinery.
Adds command-path regression tests including the terminal sibling of
the leak-A scenario (unowned shared env cwd vs session record) and an
E2E cd round-trip through terminal_tool.
Flips the read side of the cwd rearchitecture onto the _session_cwd
store introduced in the previous commit.
_authoritative_workspace_root now resolves:
1. the session's own cwd record (get_session_cwd) — per-session by
construction, so one session's cd can never leak into another
session's file resolution, with no ownership heuristics at all
2. registered override (fallback for cleared/never-written records)
3. legacy shared-env live cwd + preserved anchor (transition-only,
for commands that ran before this code loaded)
4. sentinel-free absolute TERMINAL_CWD
delegate_task children get their record seeded from the parent's at
spawn: they keep starting in the parent's directory (current behavior)
but their subsequent cds stay isolated in their own record instead of
bleeding back through the shared env.
The wrong-worktree leak class is now solved structurally on this path —
there is no shared cwd for sessions to inherit. The legacy env-side
tracking (cwd_owner, _live_cwd_if_owned, _last_known_cwd) remains only
as a transition fallback and is deleted in the next step.
First step of the cwd rearchitecture (see PR #65185 for the targeted
leak fixes this will eventually supersede, and
.hermes/plans/cwd-rearch-audit.md for the full audit + sequencing).
The root cause of the wrong-worktree bug class is that cwd lives on the
SHARED terminal env — a global mutable timeshared between sessions.
env.cwd_owner stamping, _last_known_cwd, and file_tools' ownership
ladder are all patches over that misplacement.
This adds the replacement store: _session_cwd, keyed by the raw
session/task key, with record/get/clear accessors. Step 1 is dual-write
only — every site that learns a session's live cwd also records it:
- terminal_tool foreground path: after env.execute() the env's own
post-command tracking has updated env.cwd; mirror it under the
session key that drove the command
- register_task_env_overrides: a registered workspace cwd (ACP/TUI/
desktop) seeds the session record
- clear_task_env_overrides: drops the record on teardown
Readers are untouched — behavior is identical. Later steps flip
file_tools resolution and _resolve_command_cwd to read this store,
then delete env-side tracking, cwd_owner, and _last_known_cwd.
Also hardens terminal_tool's env acquisition with an explicit
env-is-None guard (previously implicitly unbound on an unreachable
branch, flagged by pyright once the dual-write read env post-loop).
Apply positive-proof routing to every addressed notification in the registry and TUI poller while preserving ownerless legacy behavior and TUI delivery for poll-observed completions.
Remove the unused exact-key drain helper and cover ordinary success and failure, origin, compression-lineage, orphan, and poll-observed paths.
Complements NousResearch/hermes-agent#54785.
* fix(computer-use): target Linux app windows reliably
Resolve app filters through the canonical cua-driver MCP app metadata and join running app PIDs back to windows. Preserve an exact selected window across capture_after, support direct capture by pid/window_id, and send the active window ID for coordinate pointer actions on Linux.
Co-authored-by: annguyenNous <annguyenNous@users.noreply.github.com>
Co-authored-by: grimmjoww578 <willies578@gmail.com>
Co-authored-by: ai-ag2026 <261867348+ai-ag2026@users.noreply.github.com>
* fix(computer-use): address review on PR #63725
Address three review comments from @f-trycua:
1. type_text, press_key, and hotkey now carry _active_window_id and
fail closed when it is missing. Previously they sent only the PID,
so CUA Driver fell back to the first window for that PID — input
could reach the wrong window in multi-window apps.
2. Coordinate scroll x/y are now capability-gated behind
input.scroll.coordinates. CUA Driver 0.7.1 Linux schema rejects
x/y on scroll; omitting them when the driver doesn't advertise
support avoids the schema rejection while still routing via
window_id.
3. Windows are sorted by z_index descending (higher = front, per CUA
Driver semantics) instead of ascending. Null z_index (Wayland) is
coerced to 0 in _ingest_windows so it doesn't crash the sort and
sorts to the back instead of being selected as the capture target.
---------
Co-authored-by: LeonSGP43 <cine.dreamer.one@gmail.com>
Co-authored-by: annguyenNous <annguyenNous@users.noreply.github.com>
Co-authored-by: grimmjoww578 <willies578@gmail.com>
Co-authored-by: ai-ag2026 <261867348+ai-ag2026@users.noreply.github.com>
Three-layer companion to the salvaged CLI drain-ownership fix (#64240):
1. restore_undelivered_completions stamps restored=True (in-memory only)
on every durable completion re-enqueued at process start.
2. drain_notifications' legacy unfiltered branch re-queues restored
events instead of consuming them — a fresh process can no longer
adopt a dead session's delegation results (#64484). Same-process
keyless events keep the legacy behavior.
3. delegate_tool's async dispatch now falls back to the parent agent's
durable session_id when the approval-context key resolves empty (the
CLI case), so the CLI's new positive-ownership drain can actually
claim its own completions instead of failing closed on ''.
Review finding on the salvaged collector: _wait_for_process is the shared
drain for EVERY env.execute() consumer, not just the terminal tool. Applying
tool_output.max_bytes there silently truncated file-operation cat reads
(read_file_raw feeds the patch engine — read-modify-write on any file >50KB
would corrupt it), paginated read_file, code-execution RPC reads, and log
reads.
bounded_capture is now an explicit opt-in on execute()/_wait_for_process,
set only by the foreground terminal tool. Default preserves the historical
full-fidelity capture via an effectively-unbounded collector (single code
path). Modal transports accept the kwarg for signature parity.
New regression test: default execute() returns a 200KB payload complete and
untruncated. E2E: 20MB internal read intact; ShellFileOperations
read_file_raw round-trips byte-exact; terminal path still bounded at 50KB.
Follow-ups on top of #64061's salvage:
- ResourceLink markers now point at mcp__<server>__read_resource (the
actual registered tool name via mcp_prefixed_tool_name) instead of a
nonexistent <server>_read_resource the agent could hallucinate-call.
- The isError path now surfaces EmbeddedResource .resource.text blocks
instead of dropping them, so error payloads carried in resources no
longer collapse to a bare 'MCP tool returned an error'. (Same-class
fix flagged in #64061 and independently addressed in #63576 by
@alauer.)
- 3 new error-path tests + updated ResourceLink wire-name assertion.
MCP tool results with non-image binary resources (PDFs, archives, office
docs) were silently dropped: the success path only handled TextContent and
ImageContent, so a PDF-returning MCP tool appeared to return metadata only.
- EmbeddedResource blob contents are decoded (50MB cap), materialized into
the Hermes document cache via cache_document_from_bytes (sanitized
filename, traversal-safe), and surfaced as a local-path marker the agent
can read with file/terminal tools.
- EmbeddedResource text contents are inlined directly.
- ResourceLink blocks preserve the URI and point the agent at the server's
read_resource tool; no arbitrary network fetch outside the MCP session.
- AudioContent blocks are cached via cache_audio_from_bytes as MEDIA: tags.
- read_resource blob contents are materialized the same way instead of
returning '[binary data, N bytes]'.
- Unsupported blocks are logged instead of silently discarded.
- Existing ImageContent MEDIA: behavior unchanged.
Reported by an enterprise customer; reproduced against an HTTP MCP server
returning application/pdf resources.
Slack's June 30 Agent messaging experience changelog lists Bolt Python
1.29.0 / Python SDK 3.43.0 as the Agent View minimums. Bump the
messaging/slack extras and the platform.slack lazy-install pins to
match, and regenerate uv.lock. All adapter API surfaces verified
present against the new versions in a clean venv.
The terminal environment is shared process-globally (collapsed to the
default key), so env.cwd tracks the LAST session that ran a command.
_resolve_command_cwd() trusted env.cwd unconditionally — no ownership
check — so when session A left env.cwd pointing at A's checkout,
session B's first terminal command inherited A's stale cwd and ran in
the wrong workspace.
The file tools already solved this exact shared-env problem with
_live_cwd_if_owned() checking env.cwd_owner. The terminal tool never
got the same guard.
Fix: capture env.cwd_owner BEFORE the current session claims it, and
pass it as prev_owner to _resolve_command_cwd. When the previous owner
was a different session, env.cwd is stale — fall through to default_cwd
(the config/override cwd for this session) instead. Once the session
has claimed the env, subsequent calls in the same session still trust
env.cwd so in-session state survives.
Return actionable errors when HERMES_WRITE_SAFE_ROOT blocks a path instead of
labeling every denial as a protected credential file. Wire the helper through
write_file, patch, delete/move, and the Copilot ACP shim; sync docs examples.
Include the Hermes client name and version with Gemini inference, model and tier checks, and TTS requests. Add focused coverage for the request headers and keep the Gemini-specific context scoped to Google Gemini endpoints.
#63955 made Hermes survive a broken `bash -l` (Ainz's `Directory
\drivers\etc does not exist`) by falling back to non-login `bash -c`.
But a non-login shell never sources /etc/profile, so it never gets
`…\usr\bin` on PATH — and that dir holds every coreutil the file/terminal
tools shell out to (cat, mktemp, mv, wc, head, stat, chmod, mkdir, find).
Result: `write_file` returned bytes_written:0 with an EMPTY error (the
failure text went to a missing binary's stderr) and terminal commands
exited 127. The survive-broken-login-bash fix was only half-done: it
stopped crashing but silently failed every write.
Derive Git Bash's bin dirs (mingw64/bin, usr/bin, bin, …) from the
resolved bash.exe and prepend them to the subprocess PATH on Windows, in
/etc/profile precedence order so coreutils win over same-named System32
tools (find.exe, sort.exe) inside the shell. No-op off Windows and when a
login snapshot is healthy (the snapshot re-exports the full PATH inside
the shell), so this only bites on the broken-login fallback path.
Adds _git_bash_bin_dirs() (derivation, cached) + _prepend_git_bash_dirs()
(PATH merge), plus regression tests for PortableGit/MinGit layouts and
the run-env injection ordering.
`_module_registers_tools()` reads each `tools/*.py` file and fully
AST-parses it to check for a top-level `registry.register()` call.
90 files are scanned on every process start — but only 32 actually
register tools.
Add a cheap text prefilter: after reading the file (which we need to
do anyway for AST), check that both `"registry"` and `"register"`
appear in the source before calling `ast.parse`. A file with a
top-level `registry.register()` call must contain both strings, so
this is a perfect superset — zero false negatives. 50 of 90 files
skip the AST parse entirely.
The `source=` parameter is not threaded through `discover_builtin_tools`;
the prefilter lives entirely inside `_module_registers_tools`, keeping
the public API unchanged.
Benchmark (median of 10 runs, scanning 90 files):
before (read + ast.parse all): 305.9ms
after (text prefilter + ast): 187.8ms
speedup: 1.6x (118ms saved)
Identical module set: 32 modules, same names, same order.
#63621 fixed path quoting, but Ainz's Git for Windows still dies on
`bash -l` itself (`Directory \drivers\etc`). Hermes then fell back to
bash -l *per command*, so every write_file/terminal call failed the same way.
After a failed login snapshot, probe non-login bash -c; if it works, skip
-l for the session. Also skip a stale HERMES_GIT_BASH_PATH that fails a
noprofile probe in favor of %LOCALAPPDATA%\hermes\git portable bash.
#63113 rewrote native drive paths in ShellFileOperations, but init_session
/_wrap_command still embedded C:/... hermes-snap paths from get_temp_dir.
MSYS arg-converts those during bash -l and surfaces Directory \drivers\etc
— including for relative write_file targets, since the wrapper is the fault.
Add _bash_safe_path, override BaseEnvironment._quote_shell_path on
LocalEnvironment (no base→local import), and normalize mixed /c/Users\...
paths in file ops.
Co-authored-by: xxxigm <tuancanhnguyen706@gmail.com>
ShellFileOperations builds bash commands (wc/head/sed/cat/tee ...) with the
target path as an argument. On a Windows/Git-Bash host a native `C:\...` path
has its backslashes eaten by bash (and mangled by the msys runtime even when
single-quoted) — the "Directory \drivers\etc does not exist; exiting — update
your msys package" class of failures. Rewrite a native drive path to forward
slashes in `_escape_shell_arg`, reusing the env layer's `_windows_to_msys_path`.
Both `C:/...` and `/c/...` fix the backslash bug (the MSYS coreutils resolve
either via the POSIX API). We emit `/c/...` purely for consistency: it's the
same form `_windows_to_msys_path` already produces for the terminal `cd`
(LocalEnvironment._quote_cwd_for_cd), so shell file ops and `cd` share one
helper and one path form.
Scoped from #55481, which also patched BaseEnvironment._quote_cwd_for_cd — but
LocalEnvironment already overrides that through `_windows_to_msys_path`, so on a
real Windows host the base branch never ran (the cwd is already `/c/...`).
Co-authored-by: konsisumer <der@konsi.org>
When web_search results are passed directly to web_extract, the URLs
field contains dict objects (e.g., {"url": "...", "title": "..."})
rather than plain URL strings. Two code paths assumed URLs were always
strings and crashed:
- agent/display.py get_cute_tool_message for web_extract: tried to call
url.replace() on a dict, causing AttributeError
- tools/web_tools.py web_extract_tool loop: tried regex search on a dict,
causing TypeError
Both now extract the URL string from dict objects (url or href field) or
fall back to empty string, preserving the cosmetic display and allowing
the tool to process the URLs correctly.
Fixes#61693
Git Bash hands file tools paths like /c/Users/... which Path() on native
Windows treats as relative \\c\\Users\\... under the process cwd. Reuse
local._msys_to_windows_path (extended for /cygdrive and /mnt drive forms)
in _resolve_path_for_task / _resolve_base_dir so read/write/search land on
the real drive. Container/WSL Linux paths are left untouched.
Salvages #50488 (drops unrelated desktop artifact commit); tests adapted
from #46995.
Co-authored-by: Jeff Watts <186512915+lEWFkRAD@users.noreply.github.com>
Co-authored-by: LeonSGP43 <cine.dreamer.one@gmail.com>
`_load_web_config()` is typed `-> dict` but returned `load_config().get("web",
{})`, which is `None` when the config has a present-but-null `web:` section
(YAML `web:` with no body). Every caller then does
`_load_web_config().get(...)` and raises `AttributeError: 'NoneType' object
has no attribute 'get'` — this hits `_get_backend`, `check_web_api_key`, and
the extract-char-limit reader.
Separately, `check_web_api_key()` read the backend as
`.get("backend", "").lower()`; a null `web.backend` value yields `None` (the
`""` default only applies when the key is absent), so `None.lower()` raised.
`check_web_api_key` is the `check_fn` gate for `web_search`/`web_extract`, so
this surfaced as an exception during tool-availability checking.
- Make `_load_web_config()` honor its `-> dict` contract (`... or {}`), fixing
the null-`web:`-section crash at every call site.
- Guard the backend value in `check_web_api_key` with `or ""`, mirroring the
existing guard in `_get_backend`.
Adds regression tests for both the null-backend-value and null-web-section
cases.
Sibling sites of the salvaged #47334 fix: xai/openai/elevenlabs/gemini/
mistral/piper/neutts subsection reads in transcription_tools.py and
tts_tool.py used .get(key, {}) which passes a present-but-null value
through as None. All provider-subsection reads now use .get(key) or {}.
Providers without a DEFAULT_CONFIG entry (e.g. stt.xai) were still
receiving None even after the load_config() deep-merge fix, since the
merge can only fill sections that have defaults.
When stt.local, tts.edge, or other config subsections are explicitly set
to null in config.yaml (which happens by default on a fresh --voice
setup), stt_config.get('local', {}) returns None instead of {} because
YAML null preserves the key. The chained .get('model') then crashes
with 'NoneType' object has no attribute 'get'.
Apply the defensive (x or {}) pattern to every place a config subsection
is read via .get('xxx', {}). Covers local, edge, openai, mistral, and
elevenlabs subsections in both transcription_tools.py and tts_tool.py.
Closes#47318
hermes send "MEDIA:/x.png This Caption" now arrives as one native captioned
bubble instead of a separate text message followed by an uncaptioned bubble.
Root cause: the standalone senders (hermes send / cron / send_message tool)
stripped the MEDIA: tag, sent the remaining text as its own message, and
called the media send with no caption -- even though hermes send's help
advertises the captioned form and the bridges/adapters already support a
caption. Signal already captioned correctly.
- tools/send_message_tool.py: new _media_caption_split() chokepoint decides
caption-vs-separate-body (single captionable non-voice file within the
platform's message-length cap). Wired into the Telegram, WhatsApp and
Discord dispatch paths.
- Telegram/WhatsApp/Discord: when the single captioned file is missing, the
caption text is delivered as a plain message so it is never silently lost.
- Telegram caption send gets a MarkdownV2->plain parse fallback.
- Tests: _media_caption_split unit tests + per-platform caption tests
(ride, multi-file fallback, voice exclusion, over-limit fallback,
missing-file text fallback); updated the 3 tests that asserted the old
text-then-media split.
Closes the gap reported against #58911 (the MEDIA_CAPTION directive PR);
credit to @ferreiraesilva for surfacing the caption behavior.
Review finding: callers mutate the returned dicts in place —
hermes_cli/web_server.py annotates s['enabled']/s['usage'] on the skills
list — so handing out the cached objects poisons the cache for every
subsequent caller (and is a cross-thread shared-mutable hazard in the
gateway). Return [dict(s) for s in cached] on both hit and miss paths;
warm-path cost is negligible (241x speedup retained on a 300-skill
fixture). Regression test mutates a returned list/dict and asserts the
next cached call is clean.
Review findings on the cherry-picked cache (follow-up to #58985):
- The cache key was the max mtime of only the TOP-LEVEL scan dirs.
Adding/removing a skill inside a category subdir bumps the category
dir's mtime, NOT the root's, so the cache served a stale list
indefinitely. Replace with a per-dir signature covering roots +
immediate children (one scandir per dir; mirrors
hermes_cli/profiles.py::_count_skills from d5eee133e).
- The disabled-set is config-driven and changes with no filesystem
mtime bump; fold it into the signature so /skills disable takes
effect without a restart.
- Platform is part of the signature (gateway processes serve multiple
platform scopes; scan results are platform-filtered).
- Add a 30s TTL to bound staleness from in-place SKILL.md edits (file
mtime is invisible to any directory signature).
- The original also keyed dirs off the module-level SKILLS_DIR constant;
the scan itself uses _skills_dir() (live profile HERMES_HOME) — use
the same resolution for the signature.
Mutation-verified: nested-add, disabled-set, and TTL tests fail against
the pre-fix cache and pass with it.
_find_all_skills() re-reads every SKILL.md on every call, which is
wasteful when nothing changed between turns. Cache results keyed by
the max mtime across all scanned skill directories — a skill write
touches the directory, bumping mtime past the cached value and
triggering an automatic re-scan.
skip_disabled True/False are cached separately.
This commit is unstacked from #58984; it carries only the skill
discovery cache change.
(cherry picked from commit cd65673a8f)
Completes the session-binding class on the gateway surface (#55578),
matching the TUI rules:
1. Fail-closed pinning: switch_session() re-opens ended sessions, so
pinning a completion to a spawning session that has since ENDED
(user /new, closed rotation) would resurrect a conversation the user
explicitly ended and inject into it. The injection path now checks
the pinned row's ended_at first and drops the injection with a
WARNING when the spawning session is dead or unknown - the result
stays in the delegation records.
2. /new ends the old conversation's delegations: _handle_reset_command
calls interrupt_for_session() with the expiring durable session id
(matching the parent_session_id pin stamped at dispatch) plus the
routing key as fallback, so a reset can't leave dangling subagents
whose completions have no live owner.
interrupt_for_session() gains the parent_session_id selector because a
gateway chat's session_key (the platform conversation key) survives a
reset while the session id rotates - key-based matching alone could
never sever a gateway conversation's delegations.
Background delegate_task completions only carried session_key. When multiple
active sessions shared a routing peer, get_or_create_session could recover the
latest ended_at IS NULL row and inject the subagent result into the wrong
session.
Capture parent_agent.session_id at dispatch time, include it on async-delegation
completion events, and pin gateway routing via switch_session when the
synthetic completion message is handled.
Fixes#57498
Extends the salvaged session_key filter with the same fail-closed,
compression-chain-aware ownership gate the poller uses (#55578):
- drain_notifications() accepts an owns_event callback; when provided,
an async-delegation event is consumed ONLY on positive proof of
ownership, and a broken callback re-queues (never leaks). Bare key
equality remains for single-session callers (CLI); no filter remains
legacy behavior.
- The TUI post-turn drain passes _session_owns_notification_event, so
it can't adopt another session's (or an orphan's) delegation payload,
while a post-compression session still claims its own pre-compression
dispatches - the gap bare key equality left open.
The completion event already carries the dispatching session's session_key
(captured at dispatch time in delegate_tool.py:2798), but the delivery
router ignored it — results landed in whatever session was active at
completion time instead of the session that dispatched the subagent.
Changes:
- drain_notifications() in process_registry.py: optional session_key
filter. Non-matching async_delegation events are re-queued instead of
consumed, so they remain available for the correct session's drain.
- cli.py process_loop: passes active session_key to drain_notifications()
- tui_gateway/server.py post-turn drain: passes session_key from the
TUI session dict
- gateway/run.py _build_process_event_source: logs warning when routing
metadata is unresolvable (previously silent drop)
- Regression tests verifying session-scoped drain filtering
Fixes#58684
Two invariants layered on the origin-routing commit (#55578):
1. Fail closed on orphaned async-delegation payloads. The poller's
belongs-elsewhere check handles events owned by another LIVE session,
but an event whose owner is gone previously fell through and was
adopted by whichever poller saw it - injecting one chat's delegation
output into another chat. Delegation completions are now injected
only into a session that PROVABLY owns them (origin UI id, or
session-key/lineage match via the compression chain); unowned
payloads are dropped from injection with a WARNING (the subagent's
output is already persisted in the delegation records, so nothing is
lost). The shutdown drain applies the same rule. Non-delegation
events keep the historical adopt-orphans behavior.
2. A session's in-flight async delegations end with the session.
_finalize_session now calls interrupt_for_session(): delegations
commissioned by the closing UI session are interrupted always;
key-matched delegations only when the TUI owns the session lifecycle,
so closing a viewer tab on a live gateway session never kills the
gateway's own background work.
Carry the live TUI session id with async delegation completion events and prefer the commissioning UI session when desktop pollers share the completion queue. Resolve compressed session keys to their continuation before treating events as orphaned, and capture the live parent agent session id for TUI/ACP dispatch.
Three fixes for the silent post-restart ticker stall:
1. _jobs_lock() bounds its cross-process flock: LOCK_NB polled against a
30s deadline instead of an unbounded LOCK_EX taken while holding the
process-wide RLock. On timeout it logs at ERROR and degrades to
in-process-only locking (the existing fallback path), so a sibling
process wedged while holding .jobs.lock can no longer freeze every
cron function - including the ticker's get_due_jobs() and thus the
heartbeat - forever with zero logging.
2. fire_claim/run_claim freshness checks are bounded on both sides
(0 <= age < ttl): a claim stamped in the future (clock/TZ skew across
a restart) was previously fresh forever, making the job permanently
unfireable and every manual run report 'already being fired'.
3. _execute_job_now distinguishes paused/disabled/missing jobs from a
genuinely held claim instead of mislabeling them all as 'already
being fired'.
safe_load() raises ComposerError on multi-document streams (k8s manifests)
and ConstructorError on application-defined tags (CloudFormation !Sub,
Ansible !vault) — both valid YAML syntax. Now that the linter's verdict is
a fail-closed write gate, those false positives would refuse legitimate
writes outright. Switch to yaml.parse() (scanner+parser only), which still
catches real syntax failures.
write_file() previously called _atomic_write() first and only ran the
JSON/YAML/TOML/Python syntax check afterward as an informational lint
delta -- a parse failure never set the top-level `error` key, so a
corrupt structured-data write still landed on disk (and file_tools.py's
files_modified gating, which keys off `error`, silently reported it as
a successful modification).
Move the in-process syntax check for JSON/YAML/TOML ahead of
_atomic_write() and refuse the write outright on a parse failure: no
temp file, no rename, nothing touches disk, and the result carries a
top-level `error` so callers correctly see it as unmodified.
Deliberately scoped to _FAIL_CLOSED_INPROC_EXTS (JSON/YAML/TOML), not
all of LINTERS_INPROC -- .py is excluded because this codebase's own
test fixtures (TestPatchReplacePostWriteVerification et al.) write
arbitrary non-Python text through *.py paths purely to exercise
write-mechanics; a hard block there broke 3 previously-passing tests
during development. Python keeps its pre-existing non-blocking
lint-delta report.
Adds tests/tools/test_write_file_syntax_gate.py: invalid JSON/YAML/YML/
TOML refused with nothing written (new file) and nothing modified
(existing file); valid JSON/YAML still written byte-for-byte; a
non-linted extension with garbage content is unaffected; invalid Python
is confirmed NOT hard-refused (still just reported).
signal.SIGKILL / os.killpg don't exist on Windows. The watchdog is only
spawned on POSIX (wrap site gates on os.name), but guard via getattr with
a plain terminate/kill fallback so an accidental Windows import can't
AttributeError.
Two fixes on top of the salvaged parent-death watchdog:
- Apply the watchdog wrap AFTER the OSV malware preflight so the check
inspects the real npx/uvx package instead of the python wrapper
(the wrap previously made the preflight a silent no-op for every
stdio server).
- The real server runs in its own process group under the watchdog, so
the graceful-shutdown killpg no longer reached it; the watchdog now
forwards SIGTERM/SIGINT to the child's group, keeping wedged servers
killable on clean shutdown.
A stdio MCP server (e.g. `npx -y mcp-remote <url>`) is spawned as a direct
child of the Hermes process. Existing teardown (MCPServerTask.shutdown() /
_kill_orphaned_mcp_children()) reaps it correctly on a clean exit, but a
kill -9 / crash / force-quit of the Hermes process skips that path entirely
-- the child (and its own descendants, e.g. mcp-remote's spawned node
process) is orphaned and keeps running. Repeated ungraceful restarts pile up
N orphaned processes racing to hold the same upstream SSE session, producing
errors like 'Invalid request parameters' on legitimate reconnects.
macOS/Linux have no portable equivalent of prctl(PR_SET_PDEATHSIG) at the
Python subprocess level, so this adds a thin supervisor
(tools/mcp_stdio_watchdog.py) that:
- execs the real command as its own child in its own process group
- passes stdin/stdout/stderr through untouched (MCP stdio protocol
talks directly over those streams)
- polls the original spawning PID with the same orphan-detection
algorithm already proven in tui_gateway/slash_worker.py (ppid
comparison + psutil creation-time guard against PID reuse)
- SIGTERM-then-SIGKILL's the child's process group the moment the
original parent is gone
Wired into _run_stdio via a new _wrap_command_with_watchdog() helper,
POSIX-only (matches the existing killpg-based cleanup's platform scope),
fails open (any error resolving pid/create-time falls back to the
unwrapped command) so this can never be the reason a working MCP server
stops starting.
Verified: reproduced the exact orphan scenario standalone (fake parent
process spawns watchdog + fake long-running MCP child, kill -9 the fake
parent, confirm the watchdog reaps the child within its poll window with
zero leaked processes). Updated test_mcp_tool_issue_948.py's resolved-path
assertion to check the watchdog-wrapped command instead of the raw
resolved binary. Full test_mcp_tool.py + test_mcp_stability.py +
test_mcp_tool_issue_948.py suite: 232 passed. Full -k mcp sweep across the
whole test tree: 1003 passed, 2 skipped, 0 failed.
Sibling sites of the same bug class as the salvaged stdio fix:
- SSE, streamable-HTTP (new + deprecated API) initialize() calls are now
bounded by the same connect_timeout, so an endpoint that accepts the
connection but never answers the handshake cannot park the run() task
forever.
- start() now cancels its ensure_future'd run() task when the caller's
connect timeout cancels start() itself — the orphaned-task leak was
the root mechanism behind #59349, and this closes the class for any
future pre-ready hang.
A stdio MCP server that never completes `initialize` (e.g. emits a
non-JSON-RPC frame and then blocks on stdin) leaks a child process plus its
stdio pipes/pidfd on every discovery-retry cycle — unbounded, until the
gateway hits EMFILE and every new open()/spawn fails (#59349).
Root cause (confirmed by instrumenting the live repro, and different from the
issue's own hypothesis): the spawned child IS captured in `new_pids`, so the
report's "new_pids empty at finally" guess is not it. The real cause is that
`session.initialize()` hangs forever on the garbage stream. `connect_timeout`
only bounds the caller's `.result()` wait on the foreground thread — it does
NOT cancel the `_run_stdio` coroutine on the background MCP loop. So the
coroutine is stuck at `await session.initialize()` permanently, its cleanup
`finally` never runs, the child is never reaped, and it stays invisible to the
orphan-reaper (whose `_orphan_stdio_pids` set never gets populated).
Fix: wrap `session.initialize()` in `asyncio.wait_for(..., connect_timeout)`
so a stalled handshake fails instead of hanging. The TimeoutError unwinds
through the SDK context managers (closing the child's stdin -> EOF -> exit)
and lets the existing `finally` reap any straggler. Cross-platform — no
signals/pgid/proc.
Scope: stdio only. The HTTP path has the same `await session.initialize()`
shape but spawns no subprocess (so it can't cause this leak) and already has
httpx transport timeouts.
Verified: the reporter's repro goes from unbounded growth to draining to zero;
added a hermetic regression test (fake transport whose `initialize()` hangs,
asserts the connect is bounded by connect_timeout) that fails on the pre-fix
code and passes on the fix; 566 existing MCP tests pass; ruff clean.
Repro confirmed on macOS (pipe FDs); the Linux-specific pidfd growth in the
report should be equivalent — the reporter offered to validate on Linux.
Closes#59349
Merge the two cherry-picked reap call sites into one unscoped sweep at
the top of _run_stdio (the unscoped sweep is a superset of the
per-server one), and run it via asyncio.to_thread so the 2s
SIGTERM->SIGKILL escalation cannot stall the shared MCP event loop.
When an MCP stdio subprocess fails to connect (token expiry, port
contention, timeout), the run() reconnect loop retries with backoff.
Each retry calls _run_stdio() which spawns a new process pair, but the
previous failed pair was only detected as orphaned (added to
_orphan_stdio_pids) — never actually killed. This caused rapid zombie
accumulation: 5 failed attempts × 2 procs each = 10 orphans competing
for the same port.
Add a _kill_orphaned_mcp_children() call at the top of _run_stdio(),
before the _snapshot_child_pids() baseline, so any orphans from prior
failed attempts are reaped before a new subprocess is spawned.
Fixes#57355
_run_approval_gate's gateway branch only queued via submit_pending, so
plugin-escalated approvals never sent the interactive embed+buttons on
Discord/Telegram/Slack (#59413) - the user was never notified and the
action stayed silently blocked. Mirror check_dangerous_command's path:
when a session notify callback is registered, run the blocking
_await_gateway_decision round-trip (redacted payload, once/session/
always persistence, deny/timeout produce definitive BLOCKED outcomes);
fall back to submit_pending only when no callback exists.
Fixes#59413.
* feat(trace): upload sessions to HF Agent Trace Viewer
Salvage trace upload as a smaller CLI-first feature: deterministic Claude Code JSONL export, fail-closed redaction, lazy Hugging Face dependency, and no gateway slash-command wiring.
* chore(trace): drop external porting references from docstrings
Describe the trace-upload design in Hermes' own terms.
* feat(sessions): fold trace upload into 'sessions export --format trace'
Integrates the HF Agent Trace Viewer exporter (PR #36145) onto the
unified export surface instead of a separate 'hermes trace' subcommand:
- --format trace: Claude Code JSONL to stdout/file, or one
<id>.trace.jsonl per session for filtered bulk export; defaults to
the most recent session when no --session-id/filters given.
- --upload pushes to the user's private HF traces dataset (--public to
opt out of private); reads HF_TOKEN with guided setup when missing.
- traces are secret-redacted by default (force mode); --no-redact opts
out after review; redaction failure blocks export (fail closed).
- hermes_cli/trace.py + subcommands/trace.py removed; agent/trace_upload.py
is the single engine. Docs EN + zh-Hans; 4 new CLI tests.
BaseEnvironment writes shell snapshots and cwd metadata through the process
umask. With a common 022 umask, snapshot files containing exported environment
state landed at mode 0644 even though they can include env-carried credentials
from the parent process.
Set umask 077 only around Hermes metadata writes: the initial snapshot
bootstrap and the post-command snapshot/cwd refresh. User commands still run
under the caller's original umask, while Hermes-owned snapshot and cwd files
are created owner-only.
This intentionally does not copy the source PR's global orphan sweep; deleting
all matching /tmp snapshot files could interfere with concurrent Hermes
processes. The security-critical local disclosure fix is the file mode clamp.
This is salvageable because the source report still identifies a concrete
credential-disclosure path, but the safe subset is smaller than the original
proposal: clamp only the Hermes-owned snapshot writes and leave process-wide
cleanup, user command umask, and concurrent sessions alone.
Salvages source PR: https://github.com/NousResearch/hermes-agent/pull/20056
Related issue: https://github.com/NousResearch/hermes-agent/issues/48441
Co-authored-by: Andrew Homeyer <andrew@hndl.app>
Same bug class as skills_tool: module-level SKILLS_DIR pinned at import
under the launch HERMES_HOME makes skill_manage() write/edit against the
wrong profile in long-lived multi-profile runtimes. Apply the same
_skills_dir() call-time resolution (honoring explicit test patches of
SKILLS_DIR) to _containing_skills_root, _resolve_skill_dir,
_find_skill_in_other_profiles, and create-result path reporting.
Refs #40677
Port from anomalyco/opencode#31877: JSON Schema type arrays like
["number","string"] (common in MCP tool schemas) were collapsed to the
first non-null type, silently dropping every other branch. Several
tool-call backends reject the array form outright — llama.cpp's grammar
generator and Gemini via OpenAI-compatible transports (e.g. GitHub
Copilot proxying to Gemini) 400 on it.
_sanitize_node now mirrors @ai-sdk/google: a single non-null type stays
type:X (+nullable if null was present), multiple non-null types become
an anyOf of single-type schemas so no branch is lost, and an all-null
array becomes type:null. Single-null collapse is unchanged.
Verified nested (object props, array items) survive the full sanitize
pipeline — combinator stripping is top-level-only and nullable-union
collapse only fires on single-survivor unions, so multi-type anyOf is
left intact.
Port from openclaw/openclaw#91950: normalize LLM-generated URLs like 'https:// docs.example' before web tool safety checks while preserving path and query encoding semantics.
Port from nearai/ironclaw#4547: treat a JSON null memory target as omitted so strict providers that fill optional fields with null use the documented default target instead of failing validation.
A user-approved terminal/execute_code command could be SIGINT-killed
(exit 130 + "[Command interrupted]") by a stale interrupt bit that landed
on the execution thread during the blocking approval-wait, while the
result still carried the "...approved by the user." note. The terminal
tool runs sequentially inline on the execution thread, and nothing
cleared or re-checked the bit between approval-grant and env.execute.
Clear the current thread's interrupt bit once before an approved command
spawns its child (terminal foreground; execute_code local + remote), and
enrich the note to "...approved by the user, then interrupted." on a
genuine post-start interrupt instead of implying success. A genuine
interrupt arriving after execution starts (or during a retry backoff)
still SIGINTs the command; non-approved commands keep current behavior.
Adds regression tests covering stale-bit-clears, genuine-interrupt-still-
kills, the retry-backoff window, natural-exit-130 (not mislabeled), and
execute_code local + remote.
When a bundled web provider (firecrawl, tavily, exa, ...) is listed in
plugins.disabled, its provider never registers and the web_search/
web_extract dispatchers emitted the misleading "No web extract provider
configured. Set web.extract_backend to ..." — even though the backend was
configured correctly. The real fix is to re-enable the plugin.
- web_tools.py + web_search_registry.py: when the configured backend names
a disabled bundled web plugin, both dispatchers now point the user at the
actual cause (re-enable the plugin) instead of a wrong config hint.
- plugins_cmd.py cmd_enable: enabling by canonical key now also clears the
manifest-name alias (web-firecrawl) from plugins.disabled, so the
suggested command actually re-enables the plugin ('explicit disable wins'
matches on the name too).
- plugins_cmd.py cmd_toggle / _run_composite_ui / _run_composite_fallback:
the interactive 'hermes plugins' menu now persists the canonical key
(web/firecrawl), never the bare manifest name — the drift that put the
offending entry in plugins.disabled in the first place.
Follow-up to #59518 (which fixed web credential resolution, a different
cause). Fixes the disabled-plugin symptom reported after that PR.
The Firecrawl provider used os.getenv() to read FIRECRAWL_API_KEY and
FIRECRAWL_API_URL, which only checks the process environment. When
values are supplied through Hermes's ~/.hermes/.env config mechanism
(via hermes_cli.config.get_env_value), they are not guaranteed to be
present in os.environ for every gateway/tool execution path.
Switch to get_env_value() which checks both os.environ and the .env
file, matching the pattern used by other providers (nous_subscription,
setup, discord adapter).
Fixes#40190
_sync_back_once defers a SIGINT that lands mid-sync, then re-delivers it once the
sync completes so the user's Ctrl+C isn't lost. It did so with
os.kill(os.getpid(), signal.SIGINT). That is not graceful on Windows: os.kill
only treats CTRL_C_EVENT(0)/CTRL_BREAK_EVENT(1) as console events; any other
value (SIGINT == 2) routes to TerminateProcess(sig), so a Ctrl+C during a
remote-backend (ssh/daytona/modal) sync-back hard-kills the whole CLI session
(exit code 2) on Windows instead of raising KeyboardInterrupt.
Use signal.raise_signal(signal.SIGINT) (3.8+), which invokes the restored
handler through C raise() on every platform. Verified on Windows: raise_signal
runs the handler (graceful) while os.kill(getpid, SIGINT) TerminateProcess-es
the process. Adds a cross-platform regression test that runs on Windows too (it
stubs the locked sync body, so unlike test_file_sync_back.py it needs no fcntl).
The Skills Hub 'Browse Hub' landing page and index-backed search render
empty on fresh deployments (e.g. Fly.io VPS agents) with no stale cache.
Root cause: the centralized index at /docs/api/skills-index.json is a
large body (~34MB, tens of MB compressed) served with Content-Encoding:
br. httpx's streaming Brotli decoder — backed by brotlicffi 1.2.0.1,
which is pinned so aiohttp can decode Discord attachments — trips over
its own output_buffer_limit on a payload this size and raises:
DecodingError("brotli: decoder process called with data when
'can_accept_more_data()' is False")
_load_hermes_index() catches that (DecodingError is an httpx.HTTPError
subclass) and silently falls back to the on-disk cache. On a fresh box
that cache never existed, so HermesIndexSource.is_available is False,
the index contributes 0 skills, and the hub landing page — which is
built solely from an empty-query index search — is blank. Existing
installs only appear to work because they serve a (possibly weeks-)stale
cached index instead.
Fix: request 'gzip, deflate' on the index fetch so httpx never
negotiates the broken Brotli path, and retry once with 'identity' if a
DecodingError still occurs (defends against a proxy that ignores the
header). Falls through to the stale cache only when both attempts fail.
Verified on a live staging VPS agent: index_available flips False->True
and the featured landing list repopulates from 0 to 12.
Also un-freezes already-deployed images: skills added after an image was
built (e.g. the 'unbroker' optional skill) become reachable again via
the index, which is the whole point of the centralized catalog.
register_mcp_servers now nudges cached entries whose session is None
via _signal_reconnect, so a new agent session recovers a parked server
immediately instead of waiting up to _PARKED_RETRY_INTERVAL for the
next self-probe (#50170). Gate-check idea credit: @izumi0uu (#50184),
@LeonSGP43 (#37772), @Tranquil-Flow (#37899).
_wait_for_server_session_ready used a time.monotonic deadline; the
circuit-breaker tests freeze monotonic, turning the loop into an
infinite spin (300s SIGKILL in CI-parity runs). Bound by iteration
count instead.
Four independent pre-request stalls sat on the critical path between
prompt submission and the first streamed token, measured with cProfile
against a live process:
1. Discord capability detection (~2.0s, worst 5s): get_tool_definitions
-> _get_dynamic_schema made a BLOCKING https call to discord.com
inside AIAgent.__init__ for any user with DISCORD_BOT_TOKEN set, on
every platform, every cold process. Now non-blocking: memory cache ->
24h disk cache -> permissive default + one background detection that
seeds the disk cache for the next process. The permissive default is
pinned per-process so tool schemas never flip mid-conversation
(prompt-cache safety); it mirrors the existing detection-failure
fallback (all actions exposed, 403s enriched at call time).
2. Ollama /api/show probe (~0.3s): get_model_context_length step 5e
POSTed to <base_url>/api/show for KNOWN providers (openrouter etc.),
got a 404, and never cached the miss - so every fresh process paid a
full HTTP round-trip. Known non-Ollama providers now skip the probe;
local/custom/unknown endpoints keep the exact previous behavior.
3. env_probe subprocess sweep (~0.5s): the Python-toolchain probe ran
4-8 subprocess calls inside the FIRST system prompt build. Now warmed
off-thread during agent init; the prompt build hits the cache (same
lock, so a mid-flight warm just joins instead of recomputing).
4. tools.mcp_tool import (~0.4s): the between-turns MCP refresh in
build_turn_context imported the whole mcp package even with zero MCP
servers configured. MCP tools can only exist if tools.mcp_tool was
already imported (discovery/reload paths), so gate the import on
sys.modules membership - no behavior change for MCP users.
CLI additionally pre-imports run_agent + openai off-thread during the
idle banner window (same pattern as the /model picker prewarm), hiding
the remaining ~1.5s of module imports while the user types. Fixes 1-4
apply to every interaction layer (CLI, gateway, TUI, desktop, cron).
Measured cold first turn (submit -> request dispatched, openrouter,
discord token set): 4.3s before -> 0.9s after CLI prewarm (~80%); the
agent-side non-import cost drops 2.9s -> 0.36s (init) + 0.27s (turn
prologue).
Some MCP servers (e.g. Spring Boot apps with a React SPA) serve their
frontend on any unmatched GET route. The MCP endpoint works perfectly
via POST (JSON-RPC), but a GET to /mcp falls through to the SPA
controller and returns text/html. Hermes's preflight content-type probe
sees HTML instead of application/json or text/event-stream and refuses
to connect.
This adds a per-server config option that
bypasses the content-type probe, letting the SDK connect directly via
POST where it works fine.
```yaml
mcp_servers:
stirling-pdf:
url: http://localhost:8090/mcp
headers:
X-API-KEY: <key>
skip_preflight: true
```
Related: #52460 (OAuth redirect preflight), #51600 (skip probe on mcp add),
#40366 (skip probe on reconnect — already merged).
Some MCP servers (e.g. DocuSeal) serve their web UI on HEAD/GET but
speak Streamable HTTP only via POST. The preflight probe now tries a
lightweight JSON-RPC `initialize` POST before rejecting endpoints
whose HEAD/GET returns a non-MCP content type (e.g. `text/html`).
If the POST returns `application/json` or `text/event-stream` with a
2xx status, the endpoint is accepted. Otherwise the original rejection
behaviour is preserved.
Adds 5 new test cases covering the POST probe path:
- POST rescues HTML HEAD with JSON response
- POST rescues HTML HEAD with event-stream response
- POST still rejects when it also returns HTML
- POST still rejects on non-2xx status
- POST not attempted when HEAD already returns valid MCP content type
Parking deregisters the server's tools, which removes the only paths
that could ever set _reconnect_event (circuit-breaker half-open probe
and _signal_reconnect both live inside registered tool handlers). A
parked server was therefore unrevivable short of a manual /mcp reload —
the park comment's promised breaker wake could never fire.
Make the parked wait a timed wait: every _PARKED_RETRY_INTERVAL (300s)
the run task wakes and attempts one revival probe, re-parking on
failure instead of burning the full 5-retry budget each cycle. Explicit
reconnect requests still wake it immediately. Idea credit: @Hellbayne
(PR #38881, earliest never-abandon proposal), reconciled with the
park design from #53599.
_discover_tools only filled self._tools; registry registration happened
only in _discover_and_register_server (initial start) and _refresh_tools.
After parking deregistered a server's tools, a revival rebuilt the
transport but published zero tools — a phantom recovery.
Register freshly discovered tools whenever _ready is set and the
registry entry list is empty. Extracted from PR #54139 by @nicha16
(the remainder of that PR reverses the park design and is not taken).
The local retries variable in MCPServerTask.run() accumulated across
transient disconnections — each transport exception incremented it, but
only clean transport returns (auth recovery / manual refresh) or
park-wake reset it. Five transient blips over a long-uptime gateway
would permanently park the MCP server.
Promote retries to instance attribute _reconnect_retries and reset it
at all 4 session-establishment sites in _run_stdio / _run_http, so only
consecutive failures without successful reconnection count toward the
parking budget.
Fixes#57604
Adds approvals.deny to config.yaml — a list of fnmatch globs matched
against terminal commands. A match blocks unconditionally, BEFORE the
--yolo / /yolo / approvals.mode=off bypass, making it the user-editable
counterpart to the code-shipped hardline blocklist.
- Checked in both command gates (check_dangerous_command and
check_all_command_guards), after the hardline floor and sudo-stdin
guard, before the yolo bypass and permanent allowlist.
- Matching runs over the same normalized/deobfuscated command variants
as the dangerous-pattern detector, case-insensitive.
- Opt-in: empty/absent list is a no-op; behavior unchanged.
Supersedes the trust-engine approach from #21500 with a minimal
config-native design: the only capability the existing stack lacked
was deny-that-beats-yolo. Allow already exists (command_allowlist),
ask already exists (session approvals).
PR #58889 fixed the CLI-fallback transport; review of that fix found the
same leak class at four sibling spawn sites of the third-party cua-driver
binary:
- _resolve_mcp_invocation (cua-driver manifest): no env= at all — full
parent environment inherited
- cua_driver_update_check (check-update --json): telemetry env but no
secret sanitization
- doctor._drive_health_report (<binary> mcp Popen): telemetry env only
- permissions._run (every macOS/Linux permission probe): telemetry env only
All now route through _sanitize_subprocess_env(cua_driver_child_env()),
matching the sanctioned MCP spawn and the #53503/#55709/#58889 strip-by-
default policy for non-terminal spawns. Sanitization degrades gracefully
(falls back to the telemetry env) so doctor/permission probes never break
on an import error.
4 regression tests covering each site.
When a single line exceeds the entire char budget, its tail is
unreachable via offset pagination (offsets are line-granular). Tell
the model so it doesn't assume it saw the full line.
read_file previously hard-rejected any read whose formatted output exceeded
the ~100K char safety limit, returning an error with zero content. A file
with few but very long lines (logs, wide CSV rows, minified data) sails past
the line-count limit and then trips the char guard, so the model gets nothing
and must guess a smaller limit — wasting a full round-trip.
Now the read is trimmed to the last complete line that fits the budget and
returns the partial content plus truncated_by="bytes" and a next_offset, so
the model paginates forward instead of starting over. A single line larger
than the whole budget is clamped on a code-point boundary (never empty) and
the cursor still advances. Applies at both read paths (normal + extracted
documents).
Adapted from IronClaw's Rust dual line/byte cap to hermes's Python tool-layer
char guard, which is the single uniform chokepoint over the gutter-rendered
content for every backend.
Follow-up to the salvaged toggle commit:
- file_tools.py / code_execution_tool.py: carry docker_network in their
container_config dicts so those environment-creation paths honor the
lockdown instead of silently defaulting back to bridge (the probe/exec
asymmetry class reported on #46358).
- docker.py: cross-process reuse now inspects HostConfig.NetworkMode when
docker_network=false and removes a mismatched (networked) container
before starting a fresh air-gapped one. Fails closed when inspect fails.
Default-network config never churns containers, so operators using
docker_extra_args --network=none are unaffected.
- tests: AST invariant that every container_config site carrying
docker_run_as_host_user also carries docker_network, plus three reuse
guard tests (reject bridge under lockdown / keep matching none /
no inspect when network enabled).
- docs: configuration.md gains terminal.docker_network + env var row.
Port from qwibitai/nanoclaw#2713: expose Hermes' existing Docker network isolation primitive through terminal config so operators can opt out of container egress.
_CuaDriverSession._call_tool_via_cli() (the EAGAIN/silent-empty MCP
fallback transport) invokes `cua-driver call <tool> <json>` via
subprocess.run() with no env= argument, so the third-party cua-driver
binary inherits the full, unsanitized parent environment. The primary
MCP spawn site (_lifecycle_coro) already applies
_sanitize_subprocess_env(cua_driver_child_env()) before opening the
stdio client, per the same policy #53503/#55709 established for other
subprocess spawn points — this fallback path, added alongside the
EAGAIN/silent-empty-capture hardening, missed it.
Fix: apply the same env=_sanitize_subprocess_env(cua_driver_child_env())
to the subprocess.run() call in _call_tool_via_cli(), mirroring the
sanctioned spawn site exactly (telemetry policy applied first, then
Hermes-managed secrets filtered).
Port from openai/codex#21069 ("Spill large hook outputs from context").
Both shell hooks and Python plugins can return {"context": "..."} from
pre_llm_call, which gets appended to the current turn's user message on
every subsequent API call. A plugin that emits a large blob inflates
every turn and blows out the prompt cache prefix.
- tools/hook_output_spill.py: shared helper that writes oversized
context to $HERMES_HOME/hook_outputs/<session_id>/<uuid>.txt and
returns a head/tail preview plus the saved path. Never raises.
- agent/turn_context.py: apply the cap at the pre_llm_call aggregation
site (moved here from run_agent.py since the original PR), covering
both Python plugins and shell hooks.
- agent/shell_hooks.py: reserve output_spill as a sub-key under hooks:
so the config block doesn't emit unknown-hook-event warnings.
- Docs: document the cap + config in build-a-hermes-plugin.md.
Config (behaviour-preserving when absent):
hooks.output_spill: enabled/max_chars/preview_head/preview_tail/directory
Tests: 14 unit tests; shell_hooks (56) and plugins (100) suites green.
E2E validated with isolated HERMES_HOME (spill, passthrough, traversal
sanitisation, reserved-key skip).
ruff check --fix --select F541 . on current main. Pure prefix removals;
adjacent-string concatenations keep the f only on interpolating fragments.
No string content or live placeholder altered.
Port from anomalyco/opencode#33533. Native MCP tools now register as
mcp__<server>__<tool> (double-underscore delimiter) instead of
mcp_<server>_<tool>, aligning with the convention used by Claude Code,
Codex, and OpenCode.
The double-underscore delimiter disambiguates the server/tool boundary
even when either component contains underscores (the single-underscore
form was ambiguous, which is why is_mcp_tool_parallel_safe already had to
track provenance in a side-map). It also unifies native registration with
the Anthropic-OAuth wire form (_MCP_TOOL_PREFIX = 'mcp__'), so the
single->double promotion that path performed is now a no-op for native
tools while still handling legacy replayed names.
- tools/mcp_tool.py: add MCP_TOOL_NAME_PREFIX + mcp_prefixed_tool_name()
helper; route _convert_mcp_schema, utility schemas, refresh stale-set,
and the parallel-safe prefix gate through it
- agent/transports/codex_event_projector.py: mirror convention in the
deterministic call_id input for MCP server-executed tool calls
- tests: update produced-name assertions to the new convention
The SOM/AX element list dropped labels for two extremely common cua-driver
render forms, leaving the model unable to target elements by name:
- [79] AXButton (Dark) -> parenthesised label
- [4] AXStaticText = "Wi-Fi" -> = "value" form
- [92] AXPopUpButton = "Automatic" -> = "value" form
The old regex only matched quoted "label" and id=Label, so System Settings
buttons/text/popups all surfaced with empty labels. That's why selecting the
macOS Appearance 'Dark' button by element index required guessing — the
labels weren't available to aim with.
Fix: extend _ELEMENT_LINE_RE to capture all four label forms (= "value",
"quoted", (parenthesised), id=Label), skipping a pure-digit (N) order number
in favour of the id= label. Verified live against System Settings: the
Appearance buttons now surface as Auto/Light/Dark.
Adds a regression test covering all label forms. Full suite: 84 passed.
The first fix handled the EAGAIN McpError path. But the persistent MCP
session (long-running gateway/desktop worker) has a second failure mode:
list_windows or get_window_state 'succeed' over MCP yet return a
degenerate/empty payload (no windows, or no screenshot + blank tree)
WITHOUT raising — typically when the bridge reconnected mid-call and
dropped the heavy response. That surfaced to the model as a silent 0x0
capture with no error and no fallback firing (0.00s empty return).
Fix: detect empty results in capture() and re-fetch over the CLI
transport before giving up:
- empty list_windows -> CLI re-fetch the window list
- empty get_window_state (som/ax) -> CLI re-fetch the AX tree + screenshot
- empty screenshot (vision) -> CLI re-fetch get_window_state for the PNG
Adds 2 regression tests. Full suite: 83 passed.
The cua-driver MCP stdio bridge intermittently (and on some machines
persistently) fails to forward heavier calls like get_window_state to
the daemon with POSIX EAGAIN — 'daemon transport error forwarding
get_window_state: Resource temporarily unavailable (os error 35)'.
The wrapper surfaced this as an empty 0x0 capture, so computer_use
returned blank screenshots even though the display, permissions, and
the daemon were all healthy (the direct 'cua-driver call' CLI path
worked fine throughout).
Fix: when the MCP path raises the transient/transport error, fall back
to the 'cua-driver call' subprocess transport, which talks to the
daemon over a different socket. The CLI fallback routes get_window_state
screenshots to a temp file via screenshot_out_file (tiny JSON response
instead of a multi-MB base64 blob that congests the socket), reads the
PNG back, retries with backoff, and remaps the JSON into the same
{data, images, structuredContent, isError} shape the MCP path produces
so capture()/_action() are transport-agnostic.
Adds _is_transient_daemon_error() classifier and 3 regression tests.
Verified live: captures that returned 0x0 now return full
1567x905 screenshots with the AX element tree.
The 'never reached ready' error (issue #57025) was undiagnosable — doctor
and MCP test pass while the wrapper times out, with no hint where startup
stalled. Track a phase marker through _lifecycle_coro (binary-check →
manifest-discovery → mcp-initialize → capability-discovery → ready) and
include it in the timeout RuntimeError plus a pointer to doctor and the
agent.log phase timings.
Complements the 15s→30s bump + success-path phase timing log from #58760.
On Windows, the cua-driver MCP session initialization can exceed the 15s
timeout due to manifest subprocess discovery + MCP transport setup.
This makes the computer_use tool permanently unavailable even though
hermes computer-use doctor and hermes mcp test both pass.
- Increase _ready_event.wait timeout from 15s to 30s
- Add startup timing instrumentation (manifest + mcp_init durations)
- Log timing at INFO level for diagnosability
Fixes#57025
Follow-up to #56236: the broadened root token /[/.]*\** treats any run of
dots after the root slash as a collapse spelling, so a literal root-level
directory named '...' (rm -rf /...) was unconditionally hardline-blocked
with no approval path. Tighten the token to /(?:(?:\.\.?)?/)*(?:\.\.?)?\**
so each inter-slash segment must be exactly '.' or '..' — all real collapse
spellings (//, /., /./, /.., //*, ///, /../..) stay on the hardline floor
while literal dot-run dirs fall through to the softer DANGEROUS_PATTERNS
rules like every other real path.
* feat(approvals): /deny <reason> relays denial reason to the agent
Port from qwibitai/nanoclaw#2832 (reject with reason).
Gateway /deny now accepts an optional trailing reason (/deny <reason>
or /deny all <reason>). The reason rides on the per-session approval
entry through resolve_gateway_approval -> _await_gateway_decision and is
appended to the BLOCKED tool result the agent receives, so a declined
agent can adapt instead of only hearing 'denied'.
Adapted to hermes-agent's synchronous single-command /deny model: no DB
state, no second-message capture step, no migration. Reason is capped at
280 chars and threaded through both the terminal-command guard and the
execute_code guard. Plain /deny and the approve paths are unchanged.
- tools/approval.py: _ApprovalEntry.reason; resolve_gateway_approval gains
optional reason; _await_gateway_decision returns it; both gateway BLOCKED
messages include it
- gateway/slash_commands.py: parse leading 'all' + trailing reason
- locales/en.yaml: deny.denied_reason_{singular,plural}
- hermes_cli/commands.py: /deny args_hint '[all] [reason]'
- tests: 3 new (with-reason, all+reason, plain-deny regression)
* fix(ci): localize deny-reason keys across all locales + update interrupt-path assertions
CI surfaced two enforced invariants broken by the deny-with-reason change:
- test_i18n catalog-parity requires every locale to carry the same keys as
en.yaml with matching placeholders. Added deny.denied_reason_singular/plural
(with {count}/{reason}) to all 15 non-English locales.
- test_approval_interrupt asserts the exact dict from _await_gateway_decision,
which now carries a 'reason' key (None on the interrupt/timeout paths).
Port from anomalyco/opencode#34529: MCP servers can emit
notifications/message logging notifications (RFC 5424 levels), but the
MCP SDK's default logging_callback silently discards them — server-side
warnings/errors during tool calls were invisible.
- tools/mcp_tool.py: pass a logging_callback to every ClientSession
(stdio, SSE, streamable HTTP old+new API paths via the shared
sampling_kwargs sites), mapping the 8 MCP log levels onto Python
logging levels and tagging entries with [server/logger] origin.
- JSON-serialize non-string payloads, cap at 2000 chars so a chatty
server can't flood agent.log, never raise from the handler.
- Gated on SDK support (_check_logging_callback_support) mirroring the
existing message_handler gate for old SDK versions.
- tests/tools/test_mcp_server_log_notifications.py: 10 tests covering
level mapping, origin tagging, JSON payloads, truncation, and the
never-raise contract.
video_analyze_tool's local-path branch read raw bytes via
_detect_video_mime_type (extension-only, no magic-byte check) with no
call to agent.file_safety.raise_if_read_blocked, unlike the image-gen
and video-gen provider plugins that already route local inputs through
that shared chokepoint (#57698). A model could point video_url at a
credential store (e.g. .env, auth.json) renamed or symlinked to a
video-like extension and have its raw bytes base64-encoded and sent to
the vision provider.
vision_analyze_tool and its native fast path (_vision_analyze_native)
had the same gap in their local-file branches; they were only
incidentally protected by the image magic-byte sniff rejecting
non-image content, not by the intended read guard.
Add raise_if_read_blocked() to all three local-file branches, mirroring
the existing plugins/image_gen and plugins/video_gen call sites.
Extend the pre_tool_call plugin hook return contract with a new directive:
{"action": "approve", "message": "why this needs human confirmation"}
Previously a pre_tool_call hook could only veto a tool call (action: block)
or allow it silently. It could not escalate to the existing human-approval
flow. This unlocks user-defined runtime approval rules on ANY tool (HTTP
writes, file writes to sensitive paths, email sends), enforced at runtime —
resolving #51221 as a pure plugin, with no core approval.py rule schema.
Mechanism:
- get_pre_tool_call_directive() returns (action, message) for block|approve;
get_pre_tool_call_block_message() kept as a block-only back-compat shim.
- resolve_pre_tool_block() is the single dispatch-site chokepoint: fetches
the directive and, for approve, invokes the human gate; fail-closed to a
block on denial, timeout, or gate exception. ALL FOUR tool-dispatch sites
now call it: tool_executor (concurrent + sequential), agent_runtime_helpers,
and model_tools.handle_function_call.
- request_tool_approval() escalates via the SAME machinery as Tier-2
dangerous commands: session/permanent allowlist, prompt_dangerous_approval
(CLI) / submit_pending (gateway), [o]nce/[s]ession/[a]lways/[d]eny,
timeout fail-closed, approvals.cron_mode for cron contexts.
Architecture: extracted the shared decision core into _run_approval_gate(),
called by BOTH check_dangerous_command() and request_tool_approval() so the
fail-closed / cron / gateway / yolo / persist policy lives in ONE place and
cannot drift. Fixed a latent divergence — the plugin path now honors --yolo.
Approval grain: [a]lways is keyed on tool_name + a hash of the reason (an
explicit plugin rule_key overrides), so distinct reasons on the same tool
persist independently instead of one 'always' blanketing the whole tool.
Non-interactive: cron honors approvals.cron_mode (parity with commands); any
other non-interactive non-gateway context fails CLOSED for the plugin path
(the command path keeps its historical fail-open default, unchanged).
No new config schema, no new env vars, no new hook events.
The container exec-read piped the whole file through base64 with no size
guard — the 50MB cap was only enforced host-side AFTER the full payload
had already streamed into host memory. A prompt-injected read of a huge
container file (or /dev/zero) could balloon the gateway process.
head -c (cap+1) bounds the read inside the sandbox; the +1 byte lets the
host distinguish at-cap from over-cap and reject with SourceTooLarge.
Input redirect replaces 'base64 --' (no argv exposure at all for
leading-dash paths). Docker integration tests re-verified live.
The salvaged #52688 rasterizer shell-out predates the TUI subprocess
stdin= guard; a rasterizer that prompts on stdin could hang the tool
under prompt_toolkit. DEVNULL it.
- _normalize_to_supported_image: ensure cache/vision exists before writing
the converted PNG (fresh HERMES_HOME had no dir -> FileNotFoundError).
- resolver: SVG passes through as image/svg+xml instead of erroring; the
call sites rasterize to PNG via the salvaged normalize step (cairosvg /
svglib / rsvg-convert / inkscape, best-effort with actionable error).
- normalization offloaded via asyncio.to_thread at both call sites.
- tests: resolver pass-through + rasterization + no-converter error paths.
- AUTHOR_MAP: jonathan@mintrx.com -> JAlmanzarMint (PR #52688 salvage).
vision_analyze embedded SVG (and BMP/TIFF) tool-results into conversation
history with media_type image/svg+xml. Anthropic only accepts jpeg/png/
gif/webp, so the request fails with a non-retryable 400. Because the image
is baked into immutable history and re-sent every turn, the session is
permanently wedged on resume — retries re-send the same bad bytes.
Add _normalize_to_supported_image(): SVG is rasterized to PNG (best-effort
via cairosvg/svglib/rsvg-convert/inkscape), other non-supported raster
formats are re-encoded to PNG via Pillow, and if conversion is impossible
the tool returns an actionable error instead of a session-wedging payload.
Wired into both the native-vision fast path and the auxiliary-API path so
the whole bug class is covered, not just the one call site.
All 99 existing vision tests pass.
- resolve_image_source: bare cwd-relative filenames ('pic.png') resolve
again (main accepted them; the path-shape gate regressed them — review
by egilewski). Unknown explicit schemes (ftp://, s3://) still rejected.
- Local backend: nonexistent path now raises a clean 'image file not
found' instead of a misleading sandbox-fallback message.
- W4: remove the now-dead path-based _detect_image_mime_type (suffix-trust
SVG acceptance) so future callers can't reintroduce it.
- W3: SVG sources rejected with a dedicated actionable message + test.
- Polish: host read_bytes / temp-file write_bytes offloaded via
asyncio.to_thread (matches the container exec-read); unused
ResolveContext.cfg/extra_roots fields dropped; duplicate policy check
documented as intentional pre-flight short-circuit.
Salvage of #35362, evolved to also close the vision sandbox-escape
(GHSA-gpxw-6wxv-w3qq). The two were the same root cause — vision read image
bytes host-side while every other tool reads through the terminal backend —
so one resolver fixes both the delivery gaps and the escape.
Delivery (from #35362, re-authored against current main since the branch was
4140 commits stale and vision_tools.py had been rewritten on both sides):
- tools/image_source.py: one resolver for data:/http(s)/file/local/container
image sources, returning raw bytes through a single magic-byte-sniff +
50MB-ingest chokepoint. Fixes 'no image attached' / 'Invalid image source'
for every source type (#7571, #25118, #29643, #22328, #32709, #9077).
- tools/credential_files.py: from_agent_visible_cache_path, the container->host
cache reverse-map (inverse of the existing forward twin).
- tools/vision_tools.py: both vision sites route through the resolver with
task_id threaded from the handler; resolved bytes are materialized to a temp
file so main's evolved encode/resize/embed-cap pipeline is reused verbatim
(kept over the PR's older bytes-core resize to avoid touching browser_tool /
conversation_compression callers).
Security (fills #35362's deliberately-stubbed _within_allowed_roots seam):
- Under a non-local terminal backend the file tools are confined to the sandbox
(SECURITY.md 2.2), but vision read host-side — a prompt-injected
vision_analyze('/etc/passwd') exfiltrated host secrets, and read_file even
redirects the model to vision_analyze for image paths. The resolver now
enforces the same boundary: local backend reads any host path (chosen
posture); non-local backend host-reads ONLY the media caches under
HERMES_HOME (where the gateway/download media lives) and routes every other
path to an in-sandbox base64 exec-read — which reads the CONTAINER's file,
the same one 'cat' would, never the host's. Paths are resolve()-d so a
symlink can't escape a cache; fail-closed when no sandbox env exists.
This closes the escape AND delivers container-only images (#32709) with the
same mechanism.
Tests: unified resolver + confinement model (tests/tools/test_image_source.py,
incl. proof a non-cache host path under Docker yields container bytes not the
host secret); existing vision tests updated to the resolver boundary; Docker
integration test verified green against a real daemon (exec-read of a tmpfs
/workspace file, a root-owned mode-600 file, and the host-secret invariant).
Fixes GHSA-gpxw-6wxv-w3qq.
Co-authored-by: banditburai <promptsiren@gmail.com>
The Contents-API fallback's directory listing used a raw httpx.get with
no retry, so a 429/403 rate limit aborted the whole skill download even
though file fetches already retry via _github_get. Route the directory
listing through the same helper (429/reset-aware backoff, 5xx retry,
rate-limit flagging).
Salvage of PR #3033's intent — rerouted through the _github_get helper
that landed after the PR was opened, instead of the PR's ad-hoc retry
loops.
Co-authored-by: 0xbyt4 <35742124+0xbyt4@users.noreply.github.com>
The cache used a '__default__' sentinel as its path key, so switching
HERMES_HOME (profiles, tests) within one process kept serving the stale
policy loaded from the previous home. Key the cache on the actual
resolved default config path instead, so a home/config-path change
naturally misses the cache.
Trimmed from bundled PR #3923 (the other sub-fixes are superseded on
main); authored by @aydnOktay.
Both wired against features the iron-proxy author (@mslipper) confirmed on
PR #30179 — and both verified present in the pinned v0.39.0 source.
Header-auth providers (match_headers):
- New _HEADER_AUTH_PROVIDERS: Anthropic native (x-api-key), Azure OpenAI
(api-key on *.openai.azure.com / *.cognitiveservices / *.services.ai),
Gemini (x-goog-api-key + ?key= query param via match_query).
- TokenMapping grows match_headers + alias_env_names; per-provider header
sets flow into the secrets rules; mappings.json roundtrips them
(legacy files load with the Authorization default).
- GEMINI_API_KEY / GOOGLE_API_KEY collapse into ONE mapping (two
require-rules on the same host would reject each other); the sandbox
gets the token under both names, and the proxy child env mirrors the
alias into the canonical name when only the alias is set.
- Docker backend injects alias env names alongside canonical ones.
- The fail-closed tier is now empty, so fail_on_uncovered_providers and
discover_blocked_providers are deleted (dead toggle otherwise);
_NON_BEARER_PROVIDERS shrinks to genuinely-unswappable signature auth
(AWS SigV4, GCP service-account OAuth) — warn-only, as before.
Management API (hot reload):
- Generated proxy.yaml enables the v0.39 management listener: loopback
only at tunnel_port+2, bearer key from HERMES_IRON_PROXY_MGMT_KEY.
- Key minted at setup (management.token, 0600); start_proxy injects it
(v0.39 refuses to start when api_key_env is empty).
- hermes egress reload -> POST /v1/reload: re-reads proxy.yaml and
atomically swaps the pipeline; 422 leaves the running ruleset
untouched; actionable errors for not-running / pre-management config /
key mismatch. Secrets changes still require restart (daemon env is
read at spawn) — the CLI says so.
Validation: 218/218 unit+CLI+docker tests; 3/3 gated live E2E against the
real v0.39.0 binary (Authorization swap, x-api-key swap, live reload with
token rotation on the same pid). Docs updated.
On X11 a window's PID comes from the optional _NET_WM_PID property, so
cua-driver's list_windows legitimately returns pid: null for windows that
don't set it (desktop root, panels, override-redirect popups). capture()
and focus_app() coerced every entry via int(w["pid"]) inside a list
comprehension, so a single null-pid window raised TypeError and aborted
the whole enumeration before any screenshot — capture was impossible on
any X11 desktop with even one such window.
Route both ingestion sites through a new _ingest_windows() helper that
skips entries lacking a usable pid/window_id (uncapturable anyway) and
coerces the rest.
Adds tests/tools/test_computer_use_null_pid_windows.py covering the
helper's filtering/coercion and an end-to-end capture() regression that
reproduced the crash.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Follow-up to the salvaged SSH-tilde-cwd fix. The predicate
"backend == ssh and (cwd == ~ or cwd.startswith(~/))" was inlined at
each expanduser guard site, which is how the test simulator drifted from
production (it grew an SSH guard on a top-level-alias branch that has no
production counterpart).
- Add tools/terminal_tool._is_ssh_remote_tilde_cwd(backend, cwd) as the
single source of truth (case/whitespace-tolerant).
- Use it in _get_env_config and the gateway config bridge.
- Test simulator imports the real helper instead of re-implementing the
predicate; revert the phantom SSH guard on the top-level-alias branch
(production maps top-level cwd: to a plain env var, not TERMINAL_CWD via
an SSH-guarded path — that branch tested nothing real).
Follow-up to the salvaged #58000 fix.
- Extract _raise_if_non_interactive(lead) so the shared 'hermes mcp login'
next-step wording lives in one place across both OAuth boundaries
(_redirect_handler, _wait_for_callback), rather than two copy-pasted
inline raises. Boundary-specific lead sentences preserved verbatim, so
existing message-match tests stay green.
- Add a positive-control test asserting the guard does NOT over-fire on the
interactive path (valid/refreshable tokens keep working), satisfying an
explicit regression-coverage line from #57836.
A cached-but-unusable OAuth token (expired/revoked, or a refresh the IdP
rejects) makes the MCP SDK fall through to the authorization-code flow even
though build_oauth_auth's guard only checks token-file existence. In a
non-interactive context (systemd gateway, cron, background MCP discovery)
_redirect_handler then printed an auth URL / launched a browser flow no
operator can complete, and _wait_for_callback bound a localhost listener and
blocked for the full 300s timeout — gating gateway adapter startup and, on
retry, colliding on the callback port (OSError: [Errno 98] Address already
in use).
Re-check interactivity at the redirect/callback boundary and raise an
actionable OAuthNonInteractiveError before printing a URL, opening a browser,
or binding a listener. The guard holds regardless of whether a token file
exists (the point the token-file guard cannot cover), and only triggers on
the authorization-code path, so valid/refreshable tokens keep working
non-interactively. Both build_oauth_auth and MCPOAuthManager reuse these
handlers, so the sibling construction path is covered too.
Post-merge follow-ups + several review rounds + a hub-search rework, folded together.
Merge-scuff restores (a stale-base refactor had reverted two live-on-main fixes):
- gateway: SessionStore compression-tip healing + its regression test.
- desktop: messaging session/transcript polling in desktop-controller
(MESSAGING_POLL / ACTIVE_MESSAGING_SESSION_POLL, refreshMessagingSessions,
refreshActiveMessagingTranscript, the richer sameCronSignature) so inbound
platform traffic updates live again instead of freezing until manual refresh.
Profile-switch isolation (epoch/close/guard on every profile-scoped async):
- Hub store clears + in-flight runHubAction bails (and swallows the post-switch
404 instead of a phantom toast); hub preview/scan/search/sources profile-scoped.
- MCP: probe/auth epoch guards, dirty-draft reset, sidebar mutations blocked
until config resettles AND every persist re-checks the epoch post-await;
profilePending clears on config settle incl. error; logs re-key on profile.
- Model settings reload on switch and epoch-guard setModelAssignment /
saveMoaModels / API-key activation.
- Config draft resets + cancels its autosave on switch; skill editor/archive and
star-map node dialogs close on switch; openSkillEditor / star-map openEdit
discard stale fetches; tool-usage analytics loads are profile-guarded/keyed.
Correctness + UX:
- Unique per-skill action names for hub install AND uninstall; hub/catalog rows
flip only on a clean exit_code; catalog install polls the background bootstrap
to completion, reconciles the mcp.json draft (no dropped server), and fails
loudly on non-zero exit; MCP catalog query keyed by profile.
- /test reports needs-auth for anonymous auth:oauth servers; /auth snapshots +
restores tokens on a failed re-auth and clears the full 300s callback window.
- config-settings shows a retry on load failure; CodeEditor/JsonDocumentEditor
go read-only while saving so edits typed mid-save aren't dropped.
- Deep-link highlighter deletes its param only after a successful scroll.
- Restored the PageSearchShell trailing slot → Artifacts refresh button/spinner.
- /settings?tab=mcp redirect keeps server=.
Progressive hub search: fan out one query per backend-searchable source
(index-covered API sources stay unsearchable → no ~70-call GitHub re-hammer),
merge/dedupe by trust as each lands, per-source spinner overlaid on the dimmed
chip — results stream in without blocking on the slowest, no layout shift.
test(web): /api/skills list carries usage + provenance (CI contract).
Self-review follow-up. check_web_api_key() had a hand-rolled 'walk all
registered providers and probe each' fallback that duplicated the registry's
own availability-filtered resolvers (get_active_search_provider /
get_active_extract_provider, backed by _resolve()) — a second resolution path
that could diverge (the hand-rolled walk ignored capability, so a search-only
custom provider was handled inconsistently). Delegate to the registry's
resolvers so there is one authority for 'is a custom provider usable'.
Also: _get_backend()'s tail walk now probes provider.is_available() directly
instead of round-tripping through _is_backend_available(provider.name), which
redundantly re-did the registry get_provider() lookup on a provider object
already in hand. Both fallback loops guard is_available() against exceptions.
Documented that _LEGACY_WEB_BACKENDS intentionally includes 'xai' (probed via
has_xai_credentials, not a registered provider) while the registry's
_LEGACY_PREFERENCE excludes it, so the two built-in sets don't silently drift.
Plugin-registered web providers (registered via agent.web_search_registry)
were invisible to the tool-availability gate: _is_backend_available() was a
hardcoded env-var if-chain that returned False for any name outside the eight
built-in backends. Because check_web_api_key() is the check_fn for both
web_search and web_extract, a working custom provider with no built-in creds
left both tools filtered out of the toolset entirely.
Fix at the single chokepoint: _is_backend_available() now delegates non-legacy
backend names to the registered provider's is_available(), falling back to the
legacy built-in probes for known names and unregistered providers. Because
_get_backend(), _get_capability_backend(), and check_web_api_key() all resolve
availability through this one function, the fix cascades to every caller —
including the per-capability extract selection that produced a dead-end
'search-only' error (#32698). The two remaining hardcoded whitelist
early-returns (_get_backend, check_web_api_key) now also accept registered
names, and both walk registered providers as a final fallback so a custom
backend still resolves when no built-in has credentials.
Built-in backend priority is preserved unchanged: the registry is consulted
only for names outside _LEGACY_WEB_BACKENDS.
Fixes#28651Fixes#31873Fixes#32698
_handlers for vision_analyze and video_analyze read model name from
config.yaml (auxiliary.vision.model / auxiliary.video.model) before
falling back to AUXILIARY_VISION_MODEL / AUXILIARY_VIDEO_MODEL env
vars. Matches the existing config-first pattern for timeout and
temperature in the same file.
Fixes#53749
A user with tts.openai.model set to a direct-OpenAI model (e.g. tts-1-hd)
but no VOICE_TOOLS_OPENAI_KEY/OPENAI_API_KEY (or with tts.use_gateway)
routes TTS through the managed Nous audio gateway, which only proxies
gpt-4o-mini-tts. The request 400s with:
VALIDATION_ERROR: Unsupported managed OpenAI speech model
{'model': 'tts-1-hd', 'supportedModels': ['gpt-4o-mini-tts']}
_resolve_openai_audio_client_config now reports whether it resolved the
managed gateway; _generate_openai_tts coerces the model to a
managed-supported one (logging a warning that points at the direct-key
escape hatch) unless the user redirected base_url to their own endpoint.
Direct-key users keep their tts-1/tts-1-hd preference unchanged.
Source: https://github.com/NousResearch/hermes-agent/pull/52346
Related prior work: https://github.com/NousResearch/hermes-agent/pull/39462
Related prior work: https://github.com/NousResearch/hermes-agent/pull/27426
Maintainer direction: https://github.com/NousResearch/hermes-agent/pull/52346#issuecomment-4854881612
Remove acp_command and acp_args from the model-facing delegate_task schema and
dispatch paths. Child agents can still use ACP subprocess transport when it
comes from trusted delegation config or parent inheritance, but a model tool
call can no longer choose the command or arguments that reach child
construction.
This is salvageable because the risky boundary is model control over child ACP
transport, not ACP itself. The patch follows the maintainer direction from the
source discussion by preserving trusted ACP configuration and prior integration
work while removing the untrusted tool-call fields from both top-level and
per-task delegate inputs.
Reproduced on main by passing acp_command through delegate_task and observing it
reach _build_child_agent. Verified after the fix that model dispatch strips the
hidden top-level fields and per-task hidden fields are ignored before child
construction.
Co-authored-by: Carlosian <claudlos@agentmail.to>
Co-authored-by: ssiweifnag <120658181+ssiweifnag@users.noreply.github.com>
Co-authored-by: nikshepsvn <23241247+nikshepsvn@users.noreply.github.com>
browser_cdp's frame_id (OOPIF) path returned early via
_browser_cdp_via_supervisor before _browser_cdp_private_guard ever ran,
unlike the stateless path a few lines below. A model that navigated a
cloud browser to a private/internal URL could still read page content
by passing frame_id, bypassing the same SSRF/private-page boundary
already enforced on Runtime.evaluate, Page.navigate, and other raw CDP
calls.
Apply the same guard call used by the stateless path before dispatching
to the supervisor, so both routing modes share one boundary.
A Cursor-style MCP manager inside Capabilities, plus the backend it needs.
- Server list with brand/favicon avatars + live status dot and a capability
summary (N tools, M prompts, K resources); Servers | Catalog views.
- Catalog: one-click install of Nous-approved servers with required-env prompts.
- GUI OAuth: Authenticate opens the system browser from the TTY-less backend and
verifies a token actually lands; header/API-key servers are never pushed down
OAuth; a dirty mcp.json can't drop a freshly-persisted auth field.
- Full-width mcp.json editor (ecosystem document format) + pinned stdio/agent
LogTail; probes cached 5m and keyed by (profile, config) so revisiting never
respawns the fleet or shows a stale probe.
- Whole-map persistence (PUT /api/mcp/servers) so deletes/toggles actually stick
(the generic /api/config deep-merge could not remove keys).
- perf: MCP probe/auth no longer hold the global skills lock, so a slow stdio
spawn can't stall every other request into a 15s timeout.
- per-tool include/exclude gating (lib/mcp-tool-filter) mirroring the CLI loader.
A Z.ai desktop user reported thinking reverting to medium after one turn,
burning ~200% of a week's credits in 4 days despite reasoning_effort: false
in config.yaml. Four compounding bugs:
- _session_info reported reasoning_effort "" for disabled reasoning,
indistinguishable from unset — the desktop adopted it after the first
turn, wiping its sticky "thinking off" pick so every later chat
reverted to the default effort.
- config.set key=reasoning always wrote agent.reasoning_effort to global
config.yaml, so every desktop model-menu selection (preset.effort ??
'medium') clobbered the user's configured value. Now session-scoped
like the messaging gateway's /reasoning, landing on
create_reasoning_override so lazily-built sessions keep it too.
- YAML `reasoning_effort: false`/`off`/`no` (boolean False) was coerced
to "" by every loader's `str(x or "")`, silently re-enabling thinking.
parse_reasoning_effort now treats False/"false"/"disabled" as
{"enabled": False}; loaders (tui gateway, gateway, cli, cron,
delegate) pass the raw value through. The desktop config reader also
crashed on the boolean (false.trim()), aborting voice/STT settings.
- The zai provider profile never sent thinking on the wire, and GLM-4.5+
defaults to thinking ON server-side — so disabling reasoning was a
silent no-op on direct Z.ai, the actual token burner. The profile now
emits extra_body.thinking {"type": "enabled"|"disabled"} for
thinking-capable GLM models, mirroring the DeepSeek profile.
Also: /new (session reset) now carries reasoning_config across the
rebuild like model_override; config.get reasoning prefers the session's
live value and maps a config False to "none"; Settings shows "Off"
instead of a blank select for hand-written false.
MSYS_NO_PATHCONV is honored by Git for Windows bash only. _find_bash's
final shutil.which fallback can return MSYS2-proper or Cygwin bash,
which ignore it and honor MSYS2_ARG_CONV_EXCL instead. Set both so argv
path conversion stays disabled regardless of which bash flavor spawns.
Also subsumes the cmd /c mangling in #56147.
Git Bash mangles native Windows command flags (/FO, /TN, /Create) into
bogus paths. Hermes terminal and background spawns now opt out by default
so tasklist, schtasks, and wmic work without manual prefixes.
Fixes#56700.
Follow-up to #56874, which added the Camofox private-page SSRF guard
(_camofox_current_page_private_url) but wired it only into the Camofox
eval path (_camofox_eval). The other Camofox content-read tools —
camofox_snapshot, camofox_get_images, and camofox_vision — still read the
current page's accessibility tree / images / screenshot without the
guard, so on a non-local Camofox backend they can return the content of
an intranet or cloud-metadata page (e.g. 169.254.169.254) that the
terminal itself can't reach.
Apply the same guard, gated on _eval_ssrf_guard_active (non-local
backend, not a local sidecar, allow_private_urls unset) and fail-open on
probe failure, matching the eval-path guard and the main-browser
snapshot/vision guards. camofox_back is intentionally not changed: its
target is unknown until navigation completes, and the subsequent content
read is already guarded.
Adds regression tests covering the three read tools blocking on a private
page, the public-page pass-through, and the guard-inactive no-probe path.
Three CLI reliability fixes:
1. Interrupt reliability: chat() only re-queued the user's interrupt
message when the turn result carried interrupted=True. When the agent
thread raced past its last interrupt check (or finished) before the
interrupt landed, the message was silently dropped — and the stale
_interrupt_requested flag left on the agent instantly aborted the
NEXT turn. Un-acknowledged interrupt messages are now re-queued as
the next turn and the stale flag is cleared (only when the agent
thread actually exited). The clarify-race path also parks the message
in _pending_input instead of dropping it.
2. Slow exit (5+ min): stdlib ThreadPoolExecutor workers are non-daemon
and joined unconditionally by concurrent.futures' atexit hook — even
after shutdown(wait=False). One wedged tool worker (abandoned after
interrupt/timeout) held the process open forever. Promoted
async_delegation's daemon executor to a shared tools/daemon_pool
module and adopted it in tool_executor (concurrent tool batches),
memory_manager (background sync), delegate_tool (child timeout wrapper
+ batch fan-out), and skills_hub (source fan-out). Added a 30s exit
watchdog (HERMES_EXIT_WATCHDOG_S) armed at _run_cleanup start as a
backstop for wedged cleanup steps.
3. Exit jank: after prompt_toolkit tears down the input/status bars the
terminal sat silent for the whole cleanup window, looking hung. Print
'Shutting down… (finalizing session)' immediately at exit start.
E2E: live PTY interrupt of a foreground 'sleep 120' terminal tool now
aborts in ~1s and the typed message runs as the next turn; wedged-worker
+ wedged-cleanup subprocess exits in 5.8s (watchdog) instead of hanging.
Salvage of #2863 by @aydnOktay, reimplemented against current main using the
existing utils.env_var_enabled / TRUTHY_STRINGS helper instead of per-site
tuple edits. Covers the 7 gateway/config.py env-flag sites that still rejected
'on' (WHATSAPP_ENABLED, SIGNAL_IGNORE_STORIES, MATRIX_ENCRYPTION,
API_SERVER_ENABLED, WEBHOOK_ENABLED, MSGRAPH_WEBHOOK_ENABLED,
BLUEBUBBLES_SEND_READ_RECEIPTS) plus HERMES_DESKTOP gating in
read_terminal/close_terminal. The PR's approval.py HERMES_YOLO_MODE portion is
already on main via is_truthy_value.
delegation.max_concurrent_children is now the single cap for both a
batch's parallelism and concurrent background delegation units.
- _get_max_async_children() delegates to _get_max_concurrent_children();
a leftover max_async_children key logs a one-time deprecation warning
- config v32→33 migration removes the stale key, folding a raised
max_async_children into max_concurrent_children (max wins, no lost
headroom)
- capacity error messages now point at max_concurrent_children
- pool-at-capacity sync fallback now attaches an explanatory note so
the model/user know why the call blocked instead of dispatching async
Previously users who raised max_concurrent_children (e.g. to 15) still
hit the invisible default-3 async cap: the 4th background delegate_task
silently ran inline, blocking the turn with no signal.
CLAUDE_CODE_OAUTH_TOKEN is set and owned by the user's Claude Code
install (subscription OAuth), not a Hermes-managed inference
credential — Claude subscription auth is not a working Hermes provider
path. Blocklisting it broke agent-spawned claude CLIs: with no token in
the child env, claude fell through to the shared macOS Keychain /
~/.claude/.credentials.json store and, on auth failure, cleared it —
logging the user out of their interactive Claude sessions and the
desktop app.
Exempt it from _HERMES_PROVIDER_ENV_BLOCKLIST (it arrives via the
anthropic registry entry, so discard explicitly with rationale).
ANTHROPIC_API_KEY / ANTHROPIC_TOKEN and every other provider credential
remain stripped, and the GHSA-rhgp-j443-p4rf fail-closed passthrough
guard is unchanged for everything still on the blocklist.
Fixes#55878
Extends the browser private-network eval guard to the Camofox backend.
On main, _browser_eval() returned early in Camofox mode before running the
shared private-URL literal pre-scan and before re-checking the page URL
after eval, leaving Camofox as a sibling backend that could execute
browser_console(expression=...) against private/internal targets.
- move the eval private-URL literal pre-scan before the Camofox early return
- add a Camofox current-page private-URL probe via the evaluate endpoint
- withhold Camofox eval results when the page is now private/internal
Follow-up to browser private-network hardening in #56173, #56526, #56664.
Salvage of #56764 by @rayjun (rayoo), cherry-picked to preserve authorship.
Vertex AI authenticates via OAuth2 (service-account JSON path / ADC), not
PROVIDER_REGISTRY, and VERTEX_CREDENTIALS_PATH is declared with
password=False (it's a path, not a bare key) under category="provider" —
a category the registry-derived blocklist loop never checks. Both it and
GOOGLE_APPLICATION_CREDENTIALS (the ADC fallback the adapter also reads)
fell through every existing blocklist source and leaked the on-disk
location of a GCP service-account key into every spawned subprocess
(terminal, codex/copilot app-server, browser workers) — the same leak
class already closed for every other provider's credentials in #53503.
Three connected changes that fix kanban notifications in multiplex_profile
gateways and enable event-driven agent collaboration:
1. Session profile propagation
- Add HERMES_SESSION_PROFILE ContextVar (session_context.py)
- Gateway stamps source.profile at dispatch time (run.py)
- _maybe_auto_subscribe reads profile from ContextVar instead of
os.environ which is unset in the gateway main process (kanban_tools.py)
2. Notifier profile-aware routing (kanban_watchers.py)
- Adapter selection: prefer _profile_adapters[sub.notifier_profile]
so each profile's bot delivers its own task notifications
- Relax profile skip-filter: process cross-profile subscriptions when
the gateway has an adapter for the owning profile
- Extend TERMINAL_KINDS with status/archived/unblocked
3. Creator agent wakeup on terminal events (kanban_watchers.py)
- After delivering completed/blocked/gave_up/crashed/timed_out
notifications, inject a synthetic MessageEvent into the creator's
session via adapter.handle_message to trigger their agent loop
- SessionSource built from subscription metadata — no session_store
lookup needed
Self-review follow-up on the salvaged approval-routing fix.
The initial adaptation re-read os.getenv("HERMES_YOLO_MODE") at session-build
time. That diverges from the repo's security invariant: HERMES_YOLO_MODE is
frozen into tools.approval._YOLO_MODE_FROZEN at import time precisely so a skill
running mid-process cannot set the env var and instantly flip the approval
bypass (a prompt-injection escalation path). A live re-read re-opened that hole
for the codex routing path.
- Add tools.approval.is_approval_bypass_active() — the canonical three-source
bypass check (frozen --yolo/HERMES_YOLO_MODE + session /yolo + approvals.mode
off) in one place. This is the 4th inline copy of that OR-chain (the three
sites in approval.py and tui_gateway/server.py:3121 all use the same idiom);
the helper is the shared chokepoint they can collapse onto.
- codex_runtime.py now calls is_approval_bypass_active() instead of the
hand-rolled mode-or-session check plus a runtime env re-read.
- Update the env-yolo test to patch _YOLO_MODE_FROZEN (the canonical test
pattern, e.g. tests/tools/test_yolo_mode.py) rather than setenv, which is
dead-on-arrival against the frozen constant.
Fail-closed default preserved on every branch; 28 integration + 77 session/yolo
tests pass; E2E confirms the real exec decision flips decline->accept only when
bypass is active.
Every other content-returning browser tool entry point
(browser_snapshot/vision/console/eval, and click/type/press via
_blocked_private_page_action) re-checks window.location.href against the
private/internal/cloud-metadata floor after the page could have changed --
because a redirect chain or client-side navigation can land on an address
the initial browser_navigate preflight never saw. browser_back was the one
navigation-triggering entry point missing this: it called
_run_browser_command(..., "back", []) and returned the resulting URL
straight to the model with no re-check.
On a cloud/CDP (non-local) backend, if browser history contains a
private/internal address (e.g. a prior redirect touched an internal host),
browser_back would navigate the live browser there and hand the URL back
to the model with no guard -- the exact class of gap the private-page
guard exists to close, just on the one entry point it hadn't reached yet.
Re-check happens after the navigation succeeds (not before, unlike
click/type/press) since it's the resulting page -- not the one being left
-- whose safety matters. A failed back navigation (no history) skips the
check entirely since nothing changed. Verified live: the new regression
test fails (returns the private URL instead of a blocked payload) on the
pre-fix code and passes after.
The Windows _quote_cwd_for_cd override only reached _wrap_command; the
snapshot bootstrap cd in init_session still used a bare shlex.quote(),
so on Windows the bootstrap cd failed and pwd -P captured the login
shell's dir instead of terminal.cwd. Route it through _quote_cwd_for_cd
too, and add -- for hyphen-safety to match _wrap_command.
On Windows machines with both Linux and Git for Windows installed,
_find_bash() called shutil.which('bash') before checking known
Git-for-Windows install paths. shutil.which() may return a
non-MSYS bash which does not understand Windows-style paths.
This caused all terminal commands to fail with exit code 126
because the cwd prefix (a Windows path) was rejected.
Reorder the search: check Git for Windows install locations
(ProgramFiles/Git/bin/bash.exe etc.) before falling back to
PATH lookup. This matches the intent of the surrounding code
(portable Git preferred, system Git preferred, then PATH as
last resort).
Related: #23846 (same file, same class of Windows path issues)
The model could pass `toolsets` (top-level and per-task) to delegate_task,
letting it choose which toolsets a subagent got. Toolset selection is a
capability-scoping decision the model should not control; subagents inherit
the parent's enabled toolsets, period.
- Remove `toolsets` from the delegate_task() signature, the registry handler,
the top-level + per-task JSON schema, and the live dispatch path
(run_agent._dispatch_delegate_task — this forwarded it on every model call).
- Single-task and per-task child builds now pass toolsets=None so
_build_child_agent resolves to pure parent inheritance.
- Drop the now-dead _SUBAGENT_TOOLSETS / _TOOLSET_LIST_STR schema-hint block.
- _build_child_agent keeps its internal toolsets param + intersection helpers
(internal API; fed the inherited value only).
- Tests: schema assertions flipped to assertNotIn; added a regression test
proving the dispatch path never forwards a smuggled model `toolsets`.
- Docs: update delegate_task signature refs in the autonomous-ai-agents skill.
Adds Vertex AI as a first-class provider for Gemini models via Vertex's
OpenAI-compatible endpoint. Vertex authenticates with short-lived OAuth2
access tokens (service-account JSON or ADC), not a static API key — the
missing piece behind the recurring requests (#13484, #12639, #56259).
- agent/vertex_adapter.py: OAuth2 token minting + refresh-on-expiry
(5-min margin), ADC->service-account fallback, global vs regional
endpoint URLs. Config precedence: env var > config.yaml > default.
- plugins/model-providers/vertex/: provider profile (auth_type=vertex),
reuses Gemini's extra_body.google.thinking_config translation.
- runtime_provider: vertex short-circuit BEFORE the credential pool so a
credentials-file path is never mistaken for a static API key; mints a
fresh token + computes base_url per resolve.
- run_agent + conversation_loop: _try_refresh_vertex_client_credentials()
re-mints the token and rebuilds the client on a mid-session 401, so a
long-lived gateway agent survives token expiry (~1h).
- auxiliary_client: vertex auth_type branch for side-LLM tasks.
- config.yaml: vertex.project_id / vertex.region (non-secret, bridged to
env); credential path stays in .env (VERTEX_CREDENTIALS_PATH).
- setup wizard + model picker: dedicated _model_flow_vertex; curated
google/gemini-* model list; --provider choices.
- pricing/metadata: Vertex prices off the gemini docs snapshot; endpoint
host auto-maps to the vertex provider (no probe spam).
- lazy_deps + pyproject [vertex] extra: google-auth, opt-in only.
- docs: guides/google-vertex.md + providers page; tests for adapter +
runtime resolution.
Salvages and modernizes #8427 by @slawt onto current main: rewired from
the legacy PROVIDER_REGISTRY path to the provider-profile architecture,
moved non-secret config out of .env into config.yaml, and added the
per-turn 401 token-refresh the original lacked.
The patch tool's strategy 7 (unicode_normalized) matches ASCII old_string
against a file containing real Unicode (em-dashes, smart quotes, ellipsis,
non-breaking spaces). Writing new_string verbatim silently replaced the
file's Unicode with the LLM's ASCII equivalents.
_preserve_unicode_in_replacement() diffs old_string->new_string and applies
only the actual edits to the file's original Unicode text, preserving
unchanged characters.
Salvaged from #50540 by @aj-nt. Only the Unicode-preservation half is
carried over; the write_file line-number-strip half was dropped (the
existing _looks_like_read_file_line_numbered_content reject guard already
covers its target case, and the strip's looser threshold risks silently
mutating legitimate pipe-delimited content).
browser_navigate's always-blocked cloud-metadata floor (169.254.169.254,
metadata.google.internal, ECS/Azure/GCP IMDS) was gated on
`not _is_local_backend()`, contradicting both the adjacent comment and the
is_always_blocked_url docstring ("denied regardless of backend"). A default
local headless Chromium on a cloud VM — or an off-host CDP browser — could
navigate to IMDS and read instance credentials into the model context. Make the
floor unconditional on the initial-nav and post-redirect paths.
Also: _is_local_backend() ignored a CDP override while _is_local_mode() honors
it, so an off-host CDP browser was treated as "local" and skipped the broader
private/internal SSRF check too. Treat a CDP override as non-local.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Follow-up on the salvaged #49830 hardening. The contributor's sensitive
query-param set included bare English words (code, key, auth, session,
sig) that double as ordinary page facets — ?code= on promo/challenge
pages, ?key= as a search facet, ?session= on blogs — so web_extract and
cloud browser_navigate would refuse a large slice of normal browsing.
Narrow the set to unambiguously credential-named params (access_token,
authorization, client_secret, password, token, x-amz-signature, ...).
Prefix-based vendor-key redaction (is_safe_url) still catches recognizable
key shapes; this set is the belt-and-suspenders for opaque secrets carried
under an explicit credential-named parameter.
Also fixes two intra-PR-staleness test breakages surfaced by salvaging onto
current main:
- web_extract_tool() no longer accepts use_llm_processing= (signature
changed since the PR was authored) — dropped the invalid kwarg.
- agent.redact now fully masks keyed 'token=<secret>' to 'token=***'
instead of partial 'sk-...'; the console-redaction test now asserts the
real invariant (secret body gone) rather than the exact mask format.
Added a regression test that generic English-word query params are NOT
blocked by the credential guard.
Add policy gates and output redaction for browser/CDP surfaces, strengthen session ownership tracking, and block credential-like query parameters before third-party browser/web backends receive URLs.
Inspired by the agbrowse review: keep local browser magic-link flows possible while preventing cloud reader/browser escalation from receiving opaque token, code, signature, or key query parameters.
Root cause: gateway spawns LSP servers (jdtls/pyright/yaml-ls) and
slash_worker without start_new_session=True, so they inherit the
gateway process group (= TUI parent PID). When mcp_tool
_snapshot_child_pids() races with these spawns during stdio MCP
server startup, non-MCP children leak into _stdio_pgids with the
TUI parent PGID. shutdown_mcp_servers() then killpg(tui_parent_pid,
SIGTERM), killing the TUI itself.
Evidence: tui_gateway_crash.log shows recurring SIGTERM stacks:
shutdown_mcp_servers -> _kill_orphaned_mcp_children ->
_send_signal -> killpg(pgid, sig) -> SIGTERM received
Fix (3 layers):
1. agent/lsp/client.py: add start_new_session=True to LSP server
spawn so each LSP server gets its own process group/session.
2. tui_gateway/server.py: same fix for slash_worker spawn, the
symmetric root-cause patch so no gateway direct child shares
the TUI parent pgid.
3. tools/mcp_tool.py: add _filter_mcp_children() defense-in-depth
that drops non-MCP children (slash_worker, jdtls/eclipse LSP)
from the PID delta before they can poison _stdio_pgids.
git's and sudo's option parsers resolve unambiguous long-flag prefixes, so
`git reset --har`, `git branch --delete --force`, and `sudo --stdi`/`--ask`
execute identically to their full-flag forms while evading the exact-string
DANGEROUS_PATTERNS regexes that gate them. Verified live against real git
and sudo binaries. Widen the patterns to accept unambiguous abbreviations,
scoped narrowly enough to avoid colliding with sibling flags (--help,
--soft/--mixed/--merge/--keep, --shell/--set-home).
Rework follow-up on the Windows destructive-shell detection. The PowerShell
pattern required an explicit -Command/-c before the verb, but PowerShell runs
the verb as the DEFAULT POSITIONAL arg — so `powershell Remove-Item -Recurse
-Force C:\x` (no -Command) slipped through, the exact case the PR body claims
to close. Also missing the canonical `ri` alias.
Anchor the verb to the command position (after the shell name + any leading
-Flag switches + optional -Command/-c) so bare invocations are caught while a
benign path arg containing 'del'/'rm' (e.g. -File c:\del-logs\run.ps1) is not.
Add ri to the verb list. Mutation-verified regression tests for the bare
invocation, ri alias, and the benign-path negative.
The execute_code sandbox exposed its tool-call RPC (AF_UNIX socket and
remote file-poll transports) without any caller check, so any local
process that could reach the socket / rpc dir could dispatch
terminal-capable tool calls through the parent. Mint a per-session
HERMES_RPC_TOKEN, pass it to the sandboxed child, and require a
timing-safe match on every request in both _rpc_server_loop and
_rpc_poll_loop. Empty/missing/wrong token fails closed.
Salvaged from #44073 (per-session RPC token). Added timing-safe
secrets.compare_digest comparison and fail-closed regression tests.
Co-authored-by: Hermes Agent <agent@nousresearch.com>
Review of the #50531 salvage found the cross-session HERMES_SESSION_* leak also
survives on the non-terminal spawn helper hermes_subprocess_env (added by #56202
after #50531 was written), which does os.environ.copy() without the guard. Of
its six callers, five re-bind the session identity explicitly (slash_worker/ACP
via --session-key argv) and are safe by accident; but tui_gateway cli.exec
(server.py) spawns a fresh CLI with NO --session-key under the engaged TUI host,
so it inherits a possibly-foreign HERMES_SESSION_* from the last-writer-wins
global and would stamp Kanban rows / telemetry with another session's id.
Route hermes_subprocess_env through the same _inject_session_context_env
chokepoint, restoring the single-uniform-policy-across-every-spawn-surface
invariant the codebase already claims for the internal-secret filter. Safe for
all six callers: bound ContextVars win (re-binders unaffected), _UNSET strips
(closes cli.exec). Adds 3 guard tests; mutation-checked.
Session vars (HERMES_SESSION_*) have a process-global os.environ mirror written
last-writer-wins as a CLI/cron fallback and never cleared. Under a concurrent
multi-session host (messaging gateway, ACP adapter, API server, TUI) that global
belongs to whichever turn wrote it last. A subprocess spawned from a task whose
session ContextVar is _UNSET (a sibling task that never bound, or one that
inherited another session's context) inherited the FOREIGN global and acted on
another session's identity.
Add a session_context_engaged() latch (set once any host calls set_session_vars)
and route both terminal spawn paths through a single _inject_session_context_env
chokepoint: once engaged, a bound ContextVar (incl. "") is authoritative and an
_UNSET var is STRIPPED rather than inheriting the possibly-foreign global. Pure
single-process CLI/one-shot (never engaged) keeps the inherited fallback.
Salvaged from #50531 (supersedes #49922). local.py hunk re-applied by intent
onto the current hermes_subprocess_env refactor.
Co-authored-by: PolyphonyRequiem <3107779+PolyphonyRequiem@users.noreply.github.com>
_plugin_override_policy is keyed by the plugin package root
(e.g. hermes_plugins.allowed), but the lookup used caller_mod
(the exact leaf module string). A call from hermes_plugins.allowed.cleanup
would evaluate _plugin_override_policy.get("hermes_plugins.allowed.cleanup")
→ False and raise PermissionError even when the plugin registered opt-in
under its package root.
Switch the policy lookup to caller_root (.join of the first two segments)
so submodule callers inherit the package-level allow_tool_override grant.
Adds a focused regression test for the opted-in submodule case.
Wrapping a catastrophic command in a bare subshell or brace group walked
straight past the unconditional hardline floor -- even under --yolo,
/yolo, approvals.mode=off, and cron approve mode. The command-substitution
forms were already caught; the bare paren / brace-group forms were the gap.
Rather than add the paren and brace openers to the flat _CMDPOS pattern
class (which cannot tell a real subshell opener from one sitting inside a
quoted argument, and would false-positive on ordinary prose such as a PR
title that merely mentions the trigger word), teach the existing
QUOTE-AWARE command-start tokenizer (_iter_shell_command_starts) to treat
the paren and brace openers as command starts, then emit a detection
variant that marks each real command start with a newline (already a
_CMDPOS separator). Openers inside quotes never register as starts, so
quoted arguments are left untouched while real subshell/brace bypasses now
anchor. One place covers every _CMDPOS rule (shutdown/reboot/init/
systemctl/telinit and the rm root/home/system floor).
Tests: subshell/brace bypasses added to the hardline-block, root-wipe, and
yolo-bypass sets; a regression set asserts quoted paren/brace prose is NOT
blocked (guards our own gh-pr-create workflow).
3.14.1 is the current patched release on the 3.14 line; both CVE-2026-34993
(CookieJar.load RCE) and CVE-2026-47265 (per-request cookie leak on
cross-origin redirect) are fixed as of 3.14.0, and 3.14.1 rolls up the
subsequent point fixes. Re-locked uv.lock.
The messaging extra and platform.slack pin aiohttp==3.14.0, but several
lazy messaging features listed only their SDK and let aiohttp come in
transitively. Each of those SDKs caps aiohttp loosely enough that a
vulnerable already-installed aiohttp still satisfies the range, so the
eager extras got the patched floor while the lazy paths did not:
- discord.py (aiohttp>=3.7.4,<4)
- mautrix / aiohttp-socks (aiohttp>=3,<4 / aiohttp>=3.10.0) [Matrix]
- microsoft-teams-apps (aiohttp<4) [Teams]
(Teams additionally shipped an explicit but *stale* aiohttp==3.13.4 in
both the pyproject `teams` extra and platform.teams.)
- tools/lazy_deps.py: add aiohttp==3.14.0 to platform.discord, platform.matrix;
bump the stale platform.teams pin 3.13.4 -> 3.14.0.
- pyproject.toml: add aiohttp==3.14.0 to the matrix extra; bump the teams extra
3.13.4 -> 3.14.0 (homeassistant/sms/messaging already at 3.14.0).
- tests/test_packaging_metadata.py: test_security_pins_present_in_mirrored_lazy_features
now covers platform.discord/slack/matrix/teams. The existing agree-guard only
compares packages pinned in BOTH sources, so it can't catch a lazy feature
that omits a pin entirely; this guard is an explicit coverage contract
(security package -> lazy features that must carry it) and fails with
'platform.matrix: aiohttp=MISSING' if a floor is dropped again.
- uv.lock: regenerated, zero drift (aiohttp 3.14.0).
- aiohttp 3.13.4 -> 3.14.0 (messaging/slack/homeassistant/sms extras +
lazy_deps platform.slack) — picks up CVE-2026-34993 (RCE via
CookieJar.load deserialization) and CVE-2026-47265 (per-request cookie
leak on cross-origin redirect). Both are fixed only in 3.14.0; there is
no 3.13.x backport.
- anthropic 0.86.0 -> 0.87.0 (anthropic extra) — CVE-2026-34450 /
CVE-2026-34452. lazy_deps provider.anthropic was already 0.87.0; the
extra pin had drifted back to the vulnerable 0.86.0, so this realigns it.
- cryptography pinned explicitly at 46.0.7 in core deps — CVE-2026-39892,
CVE-2026-34073. It only arrives transitively via PyJWT[crypto]; the
explicit floor keeps the WeCom/Weixin crypto paths from drifting below
the fix.
uv.lock regenerated; only aiohttp / anthropic moved (cryptography already
resolved to 46.0.7). Verified 3.14.0 satisfies discord.py 2.7.1
(aiohttp>=3.7.4,<4) and slack-sdk 3.40.1 (aiohttp>=3.7.3,<4).
rm -rf //, /., /./, /.. and //* all resolve to / in the shell but slipped
past the root-filesystem hardline pattern, whose target group only matched
the literal / and /* tokens. They fell to the softer DANGEROUS_PATTERNS
'delete in root path' rule, which --yolo / approvals.mode=off / cron
approve-mode are designed to bypass — leaving the one unconditional floor
open to a full root wipe under yolo.
Broaden the root token from '/|/\\*|/ \\*' to '/[/.]*\\**' inside
_hardline_rm_path so any root-anchored path whose components collapse back
to / (repeated slashes plus ./.. segments) with an optional trailing glob
is caught. A trailing real segment (/tmp, /home, /.ssh) still fails to
match and stays with the softer rules.
Co-authored-by: kernel-t1 <214165399+kernel-t1@users.noreply.github.com>
auxv leaks AT_RANDOM (stack canary seed) + AT_BASE/AT_PHDR load
addresses — an ASLR oracle on par with maps. pagemap exposes
virtual->physical translation. Both slipped through the endswith
tuple alongside the maps family covered by the salvaged commit.
Adds regression coverage for auxv/pagemap and for the per-thread
/proc/<pid>/task/<tid>/<file> alias form (endswith catches both).
Follow-up on #32238, closes#34430.
PR #4609 blocked /proc/*/maps to prevent ASLR layout leakage, but the
endswith("/maps") check does not match /proc/*/smaps or
/proc/*/smaps_rollup — both expose the same virtual-address layout and
bypass the guard. /proc/*/numa_maps carries the same data with NUMA
annotations and is equally bypassed. /proc/*/mem (raw process memory)
is added as defence-in-depth; it requires address knowledge to exploit
but is blocked for consistency.
Extends the endswith tuple in _is_blocked_device_path() to cover all
four variants and adds regression assertions for all new paths to
test_proc_sensitive_pseudo_files_blocked.
Partially addresses #4427.
## What does this PR do?
Closes a critical bypass of the dangerous-command approval system. The
normalizer that every command passes through before pattern matching
(`_normalize_command_for_detection`) already strips ANSI, null bytes,
fullwidth Unicode, backslash escapes and empty-quote token splits — but
it did nothing about the shell `IFS` variable. In any POSIX shell `$IFS`
and `${IFS}` expand to whitespace, so a command written as
`rm${IFS}-rf${IFS}/` is executed by the live shell as `rm -rf /` while
the detection regexes — which anchor on literal `\s` between a command and
its arguments — never fire.
The impact is severe: this evades BOTH layers at once. It slips past every
entry in `DANGEROUS_PATTERNS` (so `curl${IFS}...|sh`, `sed${IFS}-i`
against `~/.hermes/config.yaml`, sudo privilege flags, etc. auto-run with
no approval prompt) AND the unconditional hardline floor that is
documented as un-bypassable "not even with --yolo" (`rm -rf /`, `mkfs`,
`dd` to a raw block device, `shutdown`/`reboot`, fork bomb). A
prompt-injected or malicious instruction could wipe the host filesystem or
power the box off while the approval system reports nothing. Confirmed at
runtime before the fix: `detect_hardline_command('rm${IFS}-rf /')` returned
`(False, None)`.
The fix mirrors the shell's own expansion: it collapses `$IFS` / `${IFS}`
(including the bash substring form `${IFS:0:1}`) to a single space inside
the existing de-obfuscation block, so the whitespace-anchored patterns
match exactly as they do for the un-obfuscated command. It is deliberately
narrow and safe — a `\b` word boundary keeps it from touching unrelated
variables like `$IFSACONFIG`, so it cannot introduce false positives on
legitimate commands.
## Related Issue
N/A
## Type of Change
- [x] 🔒 Security fix
## Changes Made
- `tools/approval.py`: in `_normalize_command_for_detection`, substitute
`$IFS` / `${IFS}` (and `${IFS:...}`) expansions with a literal space
before dangerous/hardline pattern matching, alongside the existing
backslash and empty-quote de-obfuscation.
- `tests/tools/test_approval.py`: add `TestIFSWhitespaceBypass` covering
the brace, bare and substring IFS forms against both
`detect_hardline_command` and `detect_dangerous_command`, plus
regression guards that a look-alike variable (`$IFSACONFIG`) and plain
safe commands are not flagged. Import `detect_hardline_command`.
## How to Test
1. Reproduce the hole (pre-fix): `detect_hardline_command('rm${IFS}-rf /')`
returns `(False, None)` and `detect_dangerous_command(...)` returns
`(False, ...)`, i.e. a host-destroying command is auto-approved.
2. With the fix applied, both now flag the command: hardline match
"recursive delete of root filesystem" and dangerous match "delete in
root path".
3. Run the suite: `pytest tests/tools/test_approval.py
tests/tools/test_hardline_blocklist.py -q` — the new
`TestIFSWhitespaceBypass` cases pass and nothing else regresses.
## Checklist
### Code
- [x] I've read the Contributing Guide
- [x] My commit messages follow Conventional Commits (`fix(scope):`, etc.)
- [x] I searched for existing PRs to make sure this isn't a duplicate
- [x] My PR contains **only** changes related to this fix (no unrelated commits)
- [x] I've run the relevant tests and they pass (two pre-existing failures
are environmental: missing optional deps in the minimal venv, not
caused by this change)
- [x] I've added tests for my changes
- [x] I've tested on my platform: macOS 15 (Darwin 25.5)
### Documentation & Housekeeping
- [x] I've updated relevant documentation (README, `docs/`, docstrings) — or N/A
- [x] I've updated `cli-config.yaml.example` if I added/changed config keys — or N/A
- [x] I've updated `CONTRIBUTING.md` or `AGENTS.md` if I changed architecture or workflows — or N/A
- [x] I've considered cross-platform impact (Windows, macOS) — the change is a
pure string transform with no platform-specific behavior; footgun gate passes
- [x] I've updated tool descriptions/schemas if I changed tool behavior — or N/A
_strategy_exact advanced its scan cursor by pos+1 instead of
pos+len(pattern), so self-overlapping patterns (e.g. "aa" in "aaaa")
matched at overlapping offsets. _apply_replacements works in reverse
order, so the second replacement operated on already-modified content
using stale offsets — corrupting the file and reporting the wrong count
under replace_all=True. Advancing by len(pattern) matches str.replace()
semantics.
Subprocesses spawned by the terminal tool, execute_code, Docker backend, and
the codex app-server could inherit Hermes-internal secrets that the name-based
`_HERMES_PROVIDER_ENV_BLOCKLIST` can't enumerate, because they're injected into
`os.environ` at runtime under dynamic names:
- `AUXILIARY_<TASK>_API_KEY` / `AUXILIARY_<TASK>_BASE_URL` — per-task side-LLM
credentials bridged from `config.yaml[auxiliary]` by gateway/run.py and cli.py
(vision, web_extract, approval, compression, plugin-registered tasks). Often
separate, higher-spend keys plus base URLs pointing at private endpoints.
- `GATEWAY_RELAY_*_SECRET` / `_KEY` / `_TOKEN` — relay-auth material provisioned
by gateway/relay.
Additionally, agent/transports/codex_app_server.py built its spawn env from a
raw `os.environ.copy()`, bypassing the centralized `hermes_subprocess_env()`
helper entirely — handing every codex subprocess the full Tier-1 secret set
(GH_TOKEN, gateway bot tokens, Modal/Daytona infra tokens, dashboard session
token) unfiltered. This is the #29157 sibling spawn-site gap; copilot_acp_client
already routes through the helper.
Fix — single chokepoint:
- Add `_is_hermes_internal_secret(key)` in tools/environments/local.py as the
single source of truth for the dynamic secret patterns. Matches
AUXILIARY_*_API_KEY / _BASE_URL and GATEWAY_RELAY_*_SECRET/_KEY/_TOKEN; leaves
non-secret AUXILIARY_*_PROVIDER/_MODEL and GATEWAY_RELAY routing hints visible.
- Wire the predicate into every spawn path unconditionally (ignores skill
env_passthrough opt-in AND inherit_credentials — a model-driving CLI never
needs these): `_sanitize_subprocess_env` (both loops), `_make_run_env`
(foreground), `hermes_subprocess_env` (Tier-1), and the Docker forward filter.
- Add the static GATEWAY_RELAY_* names to `_HERMES_PROVIDER_ENV_BLOCKLIST` so the
exact-match path catches them independently of the predicate.
- Add the GATEWAY_RELAY_ID/_SECRET/_DELIVERY_KEY triplet to `_ALWAYS_STRIP_KEYS`
(Tier-1) so it is stripped unconditionally on EVERY spawn surface — including
the codex/copilot `inherit_credentials=True` path that skips the Tier-2
blocklist. `_SECRET`/`_DELIVERY_KEY` are already predicate-matched; `_ID` has
no secret suffix, so enumerating it here is what closes its leak on the
inherit path (self-review W1).
- Defense in depth: env_passthrough.py `_is_hermes_provider_credential()` now
consults the same predicate, so a skill can't register these names as
passthrough and tunnel them into an execute_code / terminal child.
- Route codex_app_server through `hermes_subprocess_env(inherit_credentials=True)`
— strips Tier-1 + dynamic-internal secrets while provider creds (which codex
needs to authenticate) still flow.
Consolidates PRs #53715 (necoweb3 — the _is_hermes_internal_secret backbone +
Docker filter), #53503 (srojk34 — env_passthrough guard), and #55709 (srojk34 —
codex routing). Retires #52348 (claudlos): its copilot half is already on main,
and its codex half used the full-strip `_sanitize_subprocess_env` which would
break codex provider auth — the correct tier is `inherit_credentials=True`.
Tests: TestHermesInternalDynamicSecrets (terminal + predicate + passthrough
override), TestInternalDynamicSecrets (hermes_subprocess_env both tiers),
TestSpawnEnvSecretStripping (codex spawn env), plus env_passthrough
defense-in-depth cases.
Co-authored-by: necoweb3 <sswdarius@gmail.com>
Co-authored-by: srojk34 <286497132+srojk34@users.noreply.github.com>
Co-authored-by: claudlos <claudlos@agentmail.to>
A literal "rm -rf /" carried as DATA inside another command's quoted
argument — a PR title, a git commit -m message, an echo/printf arg —
tripped the unconditional root-filesystem hardline and could not run at
all. `gh pr create --title "block rm -rf / spellings"` was blocked
outright, because the bare rm path branch matched the mid-string "rm"
(via \brm) with the space after "/" satisfying its (\s|$) terminator.
Anchor the shared _RM_FLAG_PREFIX to _CMDPOS so the rm hardline rules
fire only when rm is an actual command word (start of line, after a
separator ; && || |, after a subshell opener $()/backtick, or after
sudo/env/exec wrappers) — not when the string appears as an argument
value. Broaden the bare-path terminator to also accept shell
metacharacters ) ` ; | & so a real wipe inside a command substitution
is still caught.
The quoted-path branch is unchanged, so quoted root/HOME paths stay
blocked. Adds regression tests for both directions: data-arg false
positives must NOT block, real wipes at every command position must block.
The model-facing cronjob tool accepts free-form provider + base_url. On fire,
the scheduler pairs the named provider's stored credential with the job's
base_url, so a prompt-injected job (e.g. provider=anthropic,
base_url=https://attacker/v1) sends the real API key to an attacker endpoint. A
base_url with no provider inherits the default provider's key for the same
effect.
Add a fail-closed guard at the tool boundary: a base_url override is allowed
only for the custom/BYOK sentinel, a configured custom_providers entry, or when
the override host matches the named provider's own endpoint; an override without
an explicit provider is rejected. The trust boundary is the caller, so
operator-configured base_urls for named providers are unaffected.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
echo <base64> | base64 -d | bash (and base32/base16, xxd -r, tr
transforms, openssl base64/enc -d) decode a dangerous command at
runtime — the raw text carries no dangerous keyword, so the denylist
never fired. Adds DANGEROUS_PATTERNS entries for decode-and-execute
pipes into a shell.
eval $(curl ...), source $(wget ...), and . $(curl ...) executed
remote content but were not covered by the existing pipe-to-shell /
process-substitution patterns. Adds a DANGEROUS_PATTERNS entry so these
command-substitution forms consistently request approval.
Original authorship preserved from PR #26965 (bot-authored commit
re-attributed to the human contributor).
Command-name obfuscation bypassed the dangerous-command denylist: the
executable name could be spelled with shell tricks that survive regex
matching but still resolve to a blocked command at runtime —
$(echo rm), ${0/x/r}m, backticks, and printf substitutions.
Adds a non-executing shell-word scanner that deobfuscates only at
command positions (start, after ;|&&||, inside $(...), after
sudo/env/exec/... wrappers) and feeds the resulting variants through
the existing HARDLINE_PATTERNS / DANGEROUS_PATTERNS — no second
blocklist. Scoping to command words keeps ordinary arguments
(echo $(echo rm) -rf /) from being promoted into command names.
Co-authored-by: egilewski <1078345+egilewski@users.noreply.github.com>
`_normalize_command_for_detection` strips backslash-escapes before matching
DANGEROUS_PATTERNS and HARDLINE_PATTERNS, but the strip rule was
`re.sub(r'\\([^\n])', r'\1', ...)` — its `[^\n]` class deliberately skips
newlines. A backslash immediately followed by a newline is a POSIX line
continuation: the shell removes BOTH characters and joins the tokens, so
`rm -rf \<newline>/` executes as `rm -rf /`. With the dangling backslash left
in place, the structured rm/dd/mkfs patterns no longer match because a literal
`\` sits wedged between the tokens they expect to be adjacent.
The worst consequence is on the HARDLINE floor. The dangerous-command layer
still fired here only by accident (the generic `\brm\s+-[^\s]*r` "recursive
delete" rule needs no path), and that layer is bypassed by `--yolo` /
`approvals.mode=off`. The hardline blocklist — the unconditional floor reserved
for catastrophic, unrecoverable commands and meant to hold even under yolo —
anchors the root path directly after the flags, so `rm -rf \<newline>/`,
`rm -r\<newline>f /`, and `rm -rf \<newline>~` all slipped past it entirely.
A yolo session could therefore wipe the root filesystem.
The fix collapses line continuations (`\` + `\n` or `\r\n`) to nothing,
mirroring the shell, before the existing escape strip runs. This was the gap
left by 621bf3a87, which added the escape strip but only for non-newline chars.
## What does this PR do?
Closes a shell line-continuation bypass in the dangerous-command detector.
Before: `rm -rf \<newline>/` normalized to `rm -rf \<newline>/`, so the
hardline root-delete patterns did not match and the command could run under
`--yolo`. After: line continuations are collapsed first, the command
normalizes to `rm -rf /`, and the hardline floor blocks it unconditionally.
## Related Issue
N/A
## Type of Change
- [x] 🔒 Security fix
## Changes Made
- `tools/approval.py`: in `_normalize_command_for_detection`, add
`command = re.sub(r'\\\r?\n', '', command)` ahead of the existing
backslash-escape strip so shell line continuations (`\`+newline, LF or CRLF)
are removed exactly as the shell would, instead of leaving a stray backslash
that breaks the structured patterns.
- `tests/tools/test_hardline_blocklist.py`: add a parametrized
`test_hardline_blocks_line_continuation` covering the root, in-flag, home,
CRLF, and mkfs continuation forms, plus
`test_line_continuation_root_wipe_cannot_bypass_hardline` asserting the
continuation root wipe stays blocked even with `HERMES_YOLO_MODE=1`.
## How to Test
1. Reproduce: stash the `tools/approval.py` change and run
`scripts/run_tests.sh tests/tools/test_hardline_blocklist.py` — the new
line-continuation cases fail (`rm -rf \<newline>/` is not flagged hardline,
and leaks past the floor under yolo).
2. Restore the change and rerun the file — all 106 tests pass.
3. Regression: `scripts/run_tests.sh tests/tools/test_approval.py` (the
existing fullwidth/ANSI/null-byte normalization and multiline cases still
pass).
## Checklist
### Code
- [x] I've read the Contributing Guide
- [x] My commit messages follow Conventional Commits (`fix(scope):`, `feat(scope):`, etc.)
- [x] I searched for existing PRs to make sure this isn't a duplicate
- [x] My PR contains **only** changes related to this fix/feature (no unrelated commits)
- [x] I've run `pytest tests/ -q` and all tests pass
- [x] I've added tests for my changes (required for bug fixes, strongly encouraged for features)
- [x] I've tested on my platform: macOS 15 (Darwin 25.5.0)
### Documentation & Housekeeping
- [x] I've updated relevant documentation (README, `docs/`, docstrings) — or N/A
- [x] I've updated `cli-config.yaml.example` if I added/changed config keys — or N/A
- [x] I've updated `CONTRIBUTING.md` or `AGENTS.md` if I changed architecture or workflows — or N/A
- [x] I've considered cross-platform impact (Windows, macOS) — handles both LF and CRLF line endings
- [x] I've updated tool descriptions/schemas if I changed tool behavior — or N/A
# Conflicts:
# tools/approval.py
The salvaged write-target boundary included `#` in its char class, so a
`#` glued to the redirect/tee path (`echo x > .env#backup`) matched as a
comment boundary and flagged the write as dangerous. But the shell writes
to the distinct file `.env#backup`, not `.env` — a false positive, same
class as the config.yaml.bak case the PR already excluded. Drop `#` from
the boundary; a real trailing comment is always whitespace-preceded (\\s).
Adds regression tests for .env#backup, config.yaml#backup, and
tee .env#backup staying out of the deny.
The dangerous-command approval gate has rules that flag a shell command
when it overwrites a project `.env` or `config.yaml` — these files hold
API keys, DB passwords, and (for `config.yaml`) the approval policy
itself, so a write to them should require user approval. The matching
`write_file`/`patch` deny on the file-tools side was paired with these
terminal-side rules so neither path is an open door.
The redirection and `tee` rules anchored the sensitive path with
`_COMMAND_TAIL` (`(?:\s*(?:&&|\|\||;).*)?$`), which only tolerates the
rest of the line being empty or a command separator. The problem: in
POSIX shell the redirection target is fixed regardless of what trails it.
`echo secret > .env extra` still truncates `.env` (the `extra` is just
another argument to `echo`), and `echo secret > .env # note` does too
(the `#` starts a comment). Because neither tail is a separator, the old
anchor failed to match and the command sailed through approval — a
prompt-injected step could overwrite a project `.env`/`config.yaml`
unprompted. The system-path redirection rule one line above never had
this restriction and already caught these forms.
The fix introduces `_WRITE_TARGET_BOUNDARY`, a lookahead that only
requires the path token to END at a shell word boundary (whitespace,
quote, separator, redirection operator, `#`, or EOL) rather than
demanding the rest of the line be empty. It is applied to the two
stream-write rules (redirection and `tee`) where the sensitive path is
always a write target. The `cp`/`mv`/`install` rule deliberately keeps
`_COMMAND_TAIL`: there the sensitive file is only a target when it is the
LAST argument (the destination), so requiring end-of-line is correct and
keeps `cp config.yaml backup.yaml` (config.yaml as the source) out of the
deny.
## What does this PR do?
Closes a bypass in the dangerous-command approval gate where a trailing
argument or `#` comment after a `>`/`>>`/`tee` write target let a command
overwrite a project `.env` or `config.yaml` without triggering approval,
even though the shell still overwrites the file.
## Related Issue
N/A
## Type of Change
- [x] 🔒 Security fix
## Changes Made
- `tools/approval.py`: add `_WRITE_TARGET_BOUNDARY` (a word-boundary
lookahead) and use it instead of `_COMMAND_TAIL` in the two
project-env/config stream-write patterns ("overwrite project env/config
via tee" and "via redirection"). `_COMMAND_TAIL` is kept and still used
by the `cp`/`mv`/`install` rule, where end-of-line anchoring is the
correct semantics.
- `tests/tools/test_approval.py`: add regression tests for
`> .env extra`, `> .env # note`, `>> config.yaml foo`, and
`tee .env backup` (now flagged), plus `> config.yaml.bak` (must stay
safe — different file).
## How to Test
1. Reproduce: before the fix,
`detect_dangerous_command("echo secret > .env extra")` returns
`(False, None, None)` — the overwrite is not flagged.
2. Apply the fix; the same call now returns the "overwrite project
env/config via redirection" detection.
3. Run `pytest tests/tools/test_approval.py -q` — the new cases pass and
the existing `cp config.yaml backup.yaml` / `config.yaml.bak`
false-positive guards still hold.
## Checklist
### Code
- [x] I've read the Contributing Guide
- [x] My commit messages follow Conventional Commits
- [x] I searched for existing PRs to make sure this isn't a duplicate
- [x] My PR contains only changes related to this fix
- [x] I've run the relevant tests and they pass
- [x] I've added tests for my changes
- [x] I've tested on my platform: macOS 15 (Darwin 25.5)
### Documentation & Housekeeping
- [x] I've updated relevant documentation (README, docs/, docstrings) — or N/A
- [x] I've updated cli-config.yaml.example if I added/changed config keys — or N/A
- [x] I've updated CONTRIBUTING.md or AGENTS.md if I changed architecture or workflows — or N/A
- [x] I've considered cross-platform impact (Windows, macOS) — or N/A
- [x] I've updated tool descriptions/schemas if I changed tool behavior — or N/A
Review follow-up on the private-page action guard:
- Add test_guard_inactive_does_not_block_or_probe: when the SSRF guard is
inactive (local backend / allow_private_urls), click/type/press must proceed
WITHOUT probing the page URL. This is the branch most likely to silently
regress if the guard condition is inverted; a mutation check (flipping the
condition) confirms the test fails as designed.
- Add test_camofox_short_circuits_before_guard: camofox mode returns from the
dedicated camofox_* path before the guard runs; guards never consulted.
- Fix PEP8: 3 -> 2 blank lines before _blocked_private_page_action.
## What does this PR do?
Closes a critical hole in the hardline command floor. HARDLINE_PATTERNS is
the unconditional last line of defense: detect_hardline_command runs BEFORE
every yolo / approvals.mode=off / cron approve-mode bypass, so it is the only
gate standing between the agent (or a prompt-injected instruction) and an
irrecoverable disk wipe. The three rm rules anchored on a bare path token,
and _normalize_command_for_detection never strips shell quotes — so the
ordinary, recommended shell idioms slipped straight through:
rm -rf "/" rm -rf '/' rm -rf "/etc"
rm -rf "$HOME" rm -rf ${HOME} rm -rf "${HOME}"
All of these returned NO hardline match. A leading quote pushes the path out
of reach of the flag group, a trailing quote breaks the `(\s|$)` terminator,
and the `${HOME}` brace form was never listed at all. Under --yolo,
approvals.mode=off, or cron approve-mode the dangerous-command layer is also
skipped, so these commands reached execution with zero gate — exactly the
unrecoverable data loss the floor is documented to make impossible. Because
quoting paths and `${HOME}` are normal shell usage, not exotic obfuscation,
this is a high-severity, easily-triggered bypass.
The fix makes the rm path matcher quote- and brace-tolerant while staying
conservative: a path is matched when it is either fully wrapped in its own
matching quote pair (`"/"`) or bare with a whitespace/end terminator. The
matching-quote requirement is deliberate so the change adds no new false
positives — a dangerous-looking string that is merely an argument to another
command (e.g. `git commit -m "rm -rf /"`) has a closing quote but no opening
quote of its own around the path, so neither branch fires.
## Related Issue
N/A
## Type of Change
- [x] 🔒 Security fix
## Changes Made
- `tools/approval.py`: added `_hardline_rm_path()` (matches a destructive
path either fully quoted or bare-with-terminator), factored the protected
system-dir list into `_HARDLINE_SYSTEM_DIRS` and the rm flag prefix into
`_RM_FLAG_PREFIX`, and rebuilt the three rm `HARDLINE_PATTERNS` on top of
them, adding the `${HOME}` brace form. Kept as plain concatenation so regex
backslashes never land inside an f-string field (Python 3.11 floor).
- `tests/tools/test_hardline_blocklist.py`: added quoted (`"/"`, `'/'`,
`"/etc"`, `"$HOME"`, ...) and brace (`${HOME}`, `"${HOME}"`) cases to the
must-block set, a dedicated `_QUOTED_BRACE_BYPASS` regression parametrization,
no-false-positive guards (`git commit -m "rm -rf /"`), and extended the
yolo-cannot-bypass integration test to cover the quoted/brace forms.
## How to Test
1. Reproduce the bypass on `main`: `detect_hardline_command('rm -rf "/"')`
returns `(False, None)` — the floor lets it through.
2. With this change it returns `(True, "recursive delete of root filesystem")`;
the same holds for `'/'`, `"/etc"`, `"$HOME"`, `${HOME}`, `"${HOME}"`.
3. Run the suite: `scripts/run_tests.sh tests/tools/test_hardline_blocklist.py`
— 125 passed, including the new bypass and no-false-positive cases.
## Checklist
### Code
- [x] I've read the Contributing Guide
- [x] My commit messages follow Conventional Commits (`fix(scope):`, etc.)
- [x] I searched for existing PRs to make sure this isn't a duplicate
- [x] My PR contains **only** changes related to this fix (no unrelated commits)
- [x] I've run the relevant tests and they pass
- [x] I've added tests for my changes (required for bug fixes)
- [x] I've tested on my platform: macOS 15 (Darwin 25.5)
### Documentation & Housekeeping
- [x] I've updated relevant documentation (README, `docs/`, docstrings) — or N/A
- [x] I've updated `cli-config.yaml.example` if I added/changed config keys — or N/A
- [x] I've updated `CONTRIBUTING.md` or `AGENTS.md` if I changed architecture or workflows — or N/A
- [x] I've considered cross-platform impact (Windows, macOS) — pattern-only change, ruff + footgun gate pass
- [x] I've updated tool descriptions/schemas if I changed tool behavior — or N/A
Inside httpx AsyncClient response event hooks, response.next_request is
often None even for a genuine redirect, so guards keyed on
`if response.is_redirect and response.next_request` silently never fire.
A public URL that 302s to http://169.254.169.254/ was followed anyway,
defeating the pre-flight is_safe_url() check.
Resolve the redirect target from the Location header (via urljoin, so
relative Locations work too), falling back to next_request only when no
Location is present. Extracted as tools.url_safety.redirect_target_from_response
and wired into every SSRF redirect guard:
- gateway/platforms/base.py (shared image + audio download for all platforms)
- tools/vision_tools.py (two download hooks)
- plugins/platforms/slack/adapter.py
Original fix by @zapabob (PR #35940), which targeted the since-refactored
gateway/platforms/slack.py; reconstructed onto the current shared sites and
widened to the whole bug class.
Review of the salvage found the timeout-message redaction left the more
common failure mode unguarded: when the first websockets.connect(cdp_url)
fails (bad URI / refused / TLS), the raw websockets exception -- which
embeds the full cdp_url incl. ?token= and user:pass@ -- is stashed as
_start_error and re-raised verbatim by start(), and two reconnect
logger.warning sites log the same raw exception.
Add a module-level _redact_cdp_error_text() chokepoint (delegating to
agent.redact.redact_cdp_url) and route all four supervisor egress points
through it:
- start() TimeoutError message (already covered; kept)
- start() _start_error re-raise -> now raises a redacted RuntimeError with
'from None' so no secret leaks via message OR traceback cause chain
- connect-failed and session-dropped reconnect warnings
Guard tests assert the re-raised message is redacted for both token and
userinfo, the raw cause is suppressed, and the helper preserves non-secret
context (host/reason). Verified with a mutation check: reverting to the raw
'raise err' fails the new tests. Correct the redact_cdp_url docstring to
scope its guarantee to direct-URL redaction and point exception callers at
the supervisor helper.
The session-log fix (browser_tool._sanitize_url_for_logs) and the
supervisor attach-timeout fix (CDPSupervisor.start) both composed the
same three redactors (redact_sensitive_text -> _redact_url_query_params
-> _redact_url_userinfo) to mask CDP endpoint credentials. Two copies of
one policy drift: tune one site (e.g. add fragment masking) and the other
silently re-leaks.
Promote that composition to a single public helper redact_cdp_url() in
agent/redact.py -- the one place the CDP-URL redaction policy lives -- and
route both call sites through it (_sanitize_url_for_logs becomes a thin
wrapper; the supervisor imports the helper instead of re-composing the
private redactors). Add direct unit tests for the seam covering query
tokens, multiple credentials, userinfo passwords, plain-URL passthrough,
non-string/exception coercion, and None.
No behavior change at the call sites; both leak paths remain closed.
PR #54851 added _sanitize_url_for_logs() and wired it into the three log
sites inside _resolve_cdp_override(). A fourth site was missed:
_create_cdp_session() logs the already-resolved cdp_url unconditionally,
and CDPSupervisor.start() interpolates the raw cdp_url[:80] into the
attach-timeout TimeoutError (which _ensure_cdp_supervisor() logs with %s).
Both leak query-string credentials (e.g. ?token=secret from hosted CDP
providers) into Hermes logs.
Sanitize the URL at both remaining sites. The raw URL is preserved
unmodified in the returned session dict and used for the real connection;
only the logged/error representation is redacted.
Salvaged from #55883.
Co-authored-by: srojk34 <286497132+srojk34@users.noreply.github.com>
Salvaged from PR #35130 (the safe subset of jnibarger01's security pass):
- threat_patterns.py: replace unbounded (?:\w+\s+)* filler with bounded
{0,8} + cap scan input at MAX_SCAN_CHARS (64KiB), and bound the .*
runs in the exfil/config-mod patterns. Kills catastrophic backtracking
on adversarial near-misses.
- hermes_state.py: cap FTS5 query length (MAX_FTS5_QUERY_CHARS) and
extract quoted phrases with a linear scan instead of a regex so
pathological quote runs can't induce backtracking.
- acp_adapter/edit_approval.py + agent/tool_dispatch_helpers.py: recognize
'*** Move File: src -> dst' V4A headers so patch-mode edits are
permissioned/traversal-checked (previously only Update/Add/Delete), and
surface a proposal for mode=patch V4A calls (previously replace-only).
Tests: +ReDoS-bound + FTS5-cap + Move-File-target + V4A-approval cases.
The MCP input-schema normalizer in _normalize_mcp_input_schema promotes the
legacy JSON Schema 'definitions' meta-keyword to '$defs' (draft 2019-09+)
so local '$ref' resolution works downstream. The previous walk renamed
*any* key named 'definitions' anywhere in the tree, including inside
'properties' dicts. That turned user-facing parameter names into '$defs',
producing property keys that contain '$', which Anthropic and OpenAI
both reject with HTTP 400 (pattern '^[a-zA-Z0-9_.-]{1,64}$').
Real-world repro: an MCP server that exposes a CI/pipelines tool whose
'definitions' parameter is an array of pipeline-definition IDs. Such a tool
is enough on its own to break every conversation, because the full tools
array is sent on every request.
Fix: when descending into a 'properties' or 'patternProperties' mapping,
iterate property-name -> schema pairs directly, leaving the property names
verbatim. Ordinary JSON Schema semantics resume inside each property's
schema, so a legitimately nested 'definitions' meta-keyword inside a
property's schema is still promoted.
Adds two regression tests:
- test_definitions_as_property_name_is_preserved (the property-name case)
- test_definitions_property_and_meta_keyword_coexist (both forms in one
schema; the property name stays, the meta-keyword promotes)
The gateway/API server rebuilds the in-memory TodoStore by replaying
caller-supplied conversation_history. _hydrate_todo_store previously
accepted any role:tool message containing a "todos" array, so a forged
bare tool result could seed arbitrary todo state and re-inflate context
every turn (GHSA-5g4g-6jrg-mw3g).
Restrict hydration to tool results paired with an earlier assistant
todo tool call (matching tool_call_id, function name == todo, no
user/system boundary between). Reuse the existing _get_tool_call_id/
name_static helpers so dict- and object-shaped tool calls both work.
Add a generous MAX_TODO_RESULT_CHARS payload guard to drop absurd
forged results before parsing; item/content caps already exist on main.
Co-authored-by: Hermes Agent <agent@nousresearch.com>
Two related bugs caused subagent delegation to silently return empty summaries
with 0 tokens when the user configured delegation.provider=bedrock alongside
delegation.base_url=https://bedrock-runtime.<region>.amazonaws.com.
Root cause #1 — misrouting in _resolve_delegation_credentials():
The configured_base_url branch unconditionally forced provider='custom' and
api_mode='chat_completions', only specializing for chatgpt.com, anthropic,
and kimi hosts. Bedrock (and other native-SDK providers) fell through as
'custom' + chat_completions, which then POSTed OpenAI-shaped JSON at
Bedrock's native API. Bedrock rejected the payload and returned nothing,
which looked like an empty LLM response to the child agent.
Fix: when provider is one of {bedrock, vertex, google, google-genai}, skip
the base_url short-circuit and fall through to resolve_runtime_provider(),
which knows how to construct the proper SDK client. base_url can still be
forwarded through that path for regional overrides.
Root cause #2 — '(empty)' sentinel accepted as success:
After N retries of empty LLM responses, run_agent.py emits the literal
string '(empty)' as final_response. _run_single_child then hit
`elif summary:` — '(empty)' is truthy, so status became 'completed' and
the parent surfaced a blank result with no error. Users saw api_calls=4,
tokens=0, duration~0.4s, status=completed.
Fix: treat final_response.strip() == '(empty)' as a failure so the parent
surfaces it instead of silently accepting zero-content 'success'.
Both paths were reproduced in a live Hermes TUI session on us-west-2 Bedrock
(provider=bedrock, model=us.anthropic.claude-sonnet-4-6) and are covered by
new tests in tests/tools/test_delegate.py.
Follow-up to the salvaged #22070. The cron-deny tirith ImportError branch
was unconditionally fail-open; now it honours security.tirith_fail_open:
false by blocking (a cron session has no user to approve), mirroring the
main flow's fail-closed synthesis (#20733).
Adds regression tests: tirith-only content threat blocked in cron-deny,
plus fail-closed/fail-open ImportError behavior.
In check_all_command_guards, the cron-deny path only ran
detect_dangerous_command (regex patterns). The tirith check starts at
line 1017, after the early return at line 1002, so content-level threats
caught only by tirith (homograph URLs, pipe-to-interpreter, terminal
injection) were silently approved in cron sessions even with
approvals.cron_mode: deny.
Add a tirith call inside the cron-deny block, mirroring the same
ImportError guard used in the main flow.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Text-only Matrix messages sent via the send_message engine (hermes send,
cron deliver: matrix) arrived unencrypted (red padlock) in E2EE rooms.
Media sends already routed through the mautrix adapter and encrypted fine,
but text-only sends took the raw-HTTP standalone_sender_fn path, which
never encrypts.
Route ALL Matrix sends through _send_matrix_via_adapter so text is
encrypted too. The adapter reuses the live gateway's E2EE session when
available (#46310) and falls back to an encryption-aware ephemeral adapter
for standalone/cron contexts. The registry standalone_sender_fn stays
registered for the contract; it is simply no longer reached for Matrix.
Salvaged from PR #20259 onto current main (the original patched the
pre-#41112 _send_matrix branch, which had since moved to the plugin's
standalone path).
Co-authored-by: Teknium <127238744+teknium1@users.noreply.github.com>
GNU tools accept unique long-option prefix abbreviations at runtime, so
`chown --recurs root` and `git push --forc` evaded the approval gate's
exact-match `--recursive`/`--force` patterns. Switch those two entries
to prefix matches (--recur[a-z]*, --forc[a-z]*).
The rm/chmod/sed long-flag patterns were left unchanged: every abbreviation
of those is already caught by the sibling short-flag and target patterns
(rm -[^s]*r, base chmod 777, sed -[^s]*i), so prefix-matching them is a
no-op. Only chown (beyond the coincidental case-insensitive r->R catch) and
git push had genuine gaps.
Co-authored-by: Subway2023 <subw3@mail2.sysu.edu.cn>
The dangerous-command approval prompt renders the flagged command so the
user can decide whether to approve. If the agent constructed it with a
credential (curl -H 'Authorization: Bearer sk-...', psql postgres://user:pw@host,
an execute_code script with api_key = 'sk-...'), that secret hit stdout and,
via the gateway notify payload, Discord/Slack messages — which are
screenshottable and forwardable.
Apply the existing agent.redact.redact_sensitive_text() to every user-facing
approval surface. Redaction is display-only: the raw command still executes
after approval, and approval persistence keys off pattern_key (not the command
text), so the allowlist is unaffected. Decision context (URL, flags, command
structure) is preserved; only the secret value masks.
Covers all surfaces, including the execute_code path the original PR missed:
- prompt_dangerous_approval(): callback + stdout fallback
- check_all_command_guards(): gateway approval_data + cron/batch pending fallback
- check_execute_code_guard(): gateway approval_data + no-notifier pending fallback
(script body can embed credentials)
Adds TestApprovalPromptRedaction covering callback redaction, no-over-redaction
of clean commands, and the execute_code pending fallback.
Salvaged from PR #13139 by @sgabel; extended to the execute_code surface.
`handle_401` spawned a dedup'd recovery coroutine via
`asyncio.create_task(_do_handle())` and discarded the returned task
reference. Python's event loop only keeps weak references to tasks, so
the coroutine could be garbage-collected before it called
`pending.set_result(...)`. Every concurrent caller awaiting that future
then hangs forever, and the `finally: entry.pending_401.pop(...)`
cleanup never runs — so subsequent 401s for the same key latch onto the
dead future too. Same pattern the adapter-side fixes address (#11997,
#11998, #12000, #12001, #12006).
Hold the task in a process-wide set on the manager and discard it via
`add_done_callback` once it completes. Regression test covers both the
structural invariant (task tracked, then removed on completion) and a
concurrent dedup path with a forced `gc.collect()` between the handler's
await points.
Tool call ids are used to name persisted large-result files. Treating that id as a raw path segment allowed traversal-like ids to resolve outside hermes-results even though the shell command quoted metacharacters.
Convert ids to single filename stems, preserve normal ids, and add a short hash when normalization is needed so unsafe ids do not collide silently.
Constraint: Avoid new dependencies and preserve existing tool-result paths for normal tool call ids
Rejected: Quote only the path | shell quoting does not prevent ../ path traversal
Confidence: high
Scope-risk: narrow
Reversibility: clean
Tested: source /Users/peter/hermes-agent/venv/bin/activate && pytest tests/tools/test_tool_result_storage.py -q
Tested: source /Users/peter/hermes-agent/venv/bin/activate && python -m compileall tools/tool_result_storage.py tests/tools/test_tool_result_storage.py
Tested: git diff --check
The init snapshot dumped functions with a line-based filter:
declare -f | grep -vE '^_[^_]'
That strips a function's *header* line (e.g. `_foo () `) but leaves the
orphaned `{ ... }` body behind, corrupting the snapshot that is sourced
before every command. Sourcing the torn snapshot runs leftover body code
and breaks subsequent commands (intermittent exit 127).
- Filter private (`_`-prefixed) functions by NAME via `declare -F` and
dump only the wanted whole definitions, so a body is never torn. Guard
against an empty name list (bare `declare -f` dumps everything).
- Treat a non-zero bootstrap exit code as snapshot-init failure, so
execution safely falls back to login-shell-per-command mode.
- Add a regression test asserting snapshot_ready stays false when
bootstrap exits non-zero.
Preserves the atomic-write ($BASHPID temp + mv -f) machinery from #38249.
- tools/skills_hub.py: the per-call resolvers now honor a test-injected real
module attribute (patch.object(hub, 'SKILLS_DIR', ...) / monkeypatch.setattr)
before falling back to dynamic profile resolution. PEP 562 __getattr__ only
fires when no real attribute exists, so an unpatched module resolves the
active profile and a patched one respects the test's value — keeping the
existing skills_hub test seam intact (5 tests had broken).
- tests/test_profile_isolation_runtime.py: real two-profile (no-mock) suite
driving each previously-leaking site under override A then B and asserting
the active profile's path/identity is used: skills_hub paths + derived
constants + default-arg resolution, gateway cache getters (incl. the
monkeypatch-still-wins seam), rich_sent_store path, and thread/executor
context propagation (raw-thread hazard documented; primitive + _run_async
worker proven to preserve the override).
A bare threading.Thread / ThreadPoolExecutor worker starts with an empty
contextvars.Context, so the context-local profile override
(_HERMES_HOME_OVERRIDE) does not cross the spawn boundary. In single-process
multi-profile runtimes (desktop tui_gateway) the worker then resolves
get_hermes_home() to the launch/default profile, leaking one profile's
reads/writes into another. The fix primitive (tools.thread_context.
propagate_context_to_thread, which copies the parent context) already exists;
the leaking spawns simply did not use it.
- model_tools.py _run_async: wrap the worker-thread loop runner. This is the
generic sync->async bridge for every async tool, so wrapping it here fixes
the leak for all async tools at once (verified: an async tool reading
get_hermes_home() under an override now resolves the active profile).
- run_agent.py bg-review thread: wrap so MEMORY.md / skill review writes land
in the spawning turn's profile (#54937 path).
- tools/async_delegation.py: wrap both single + batch executor.submit calls so
detached children resolve the dispatching profile's paths.
Scope: the vision CPU executor is intentionally left unwrapped — it runs pure
in-memory encode/resize and never resolves profile-scoped paths.
In single-process multi-profile runtimes (desktop tui_gateway), profile
scoping is a context-local ContextVar override, not a process env var. Three
subsystems froze their HERMES_HOME-derived paths at import time (or read
os.environ directly), pinning every later profile to whichever profile first
imported the module — a cross-profile data leak.
- tools/skills_hub.py: SKILLS_DIR/HUB_DIR/LOCK_FILE/etc. were module constants
frozen at import. Replace with per-call resolver functions; add a PEP 562
module __getattr__ so external 'from tools.skills_hub import SKILLS_DIR'
callers (all function-local) resolve dynamically with no call-site changes.
Convert default-arg bindings (HubLockFile/TapsManager) and the derived
HERMES_INDEX_CACHE_FILE constant too.
- gateway/platforms/base.py: image/audio/video/document cache-dir getters now
re-resolve via get_hermes_dir() per call, falling back to the module
constant when a test has monkeypatched it (preserves the existing test seam).
Media-delivery safe-roots already enumerate all profiles' cache dirs
(#31733), so per-profile resolution does not break delivery.
- gateway/rich_sent_store.py: _store_path() read os.environ['HERMES_HOME']
directly, bypassing the override entirely; route through get_hermes_home().
The judge gate added for kanban_complete (Issue #38367, PR #38388) only
covers one of the two exit paths out of run_kanban_goal_loop(). The loop
treats status == "blocked" as terminal identically to "done" (and any
other status outside running/ready/done/blocked also stops the loop —
see goals.py's status dispatch). A goal_mode worker that has learned
kanban_complete is gated can simply call kanban_block(reason="anything")
to escape the loop with zero judge involvement, fully defeating the
intent of #38367's fix.
This is Issue #38696, filed as the explicit follow-up by a reviewer on
PR #38388: "kanban_complete is one way out; kanban_block is another...
A worker that learns the complete path is gated can shift to calling
block to escape the loop with the same effect."
Implements the issue's "Option B" (deterministic allowlist, no extra
judge LLM call) using the kind taxonomy that already exists in
kb.VALID_BLOCK_KINDS, rather than inventing a new judge_goal() outcome
type (judge_goal only returns done/continue/wait/skipped — there's no
"is this block legitimate" verdict to hook the issue's "Option A"
pseudocode onto without expanding the judge's contract).
goal_mode tasks may only block with kind in {dependency, needs_input} —
the two kinds that represent a genuine external blocker the worker
cannot resolve itself. `capability`, `transient`, and an unset kind are
rejected with a message directing the worker to kanban_complete instead,
which the judge now gates. Non-goal_mode tasks are completely unaffected.
Builds on the zero-match feedback fix (previous commit) to close the silent-hang
symptom: when memory is at capacity, a failed `add`/`replace`/`remove`
consolidation could loop the whole turn to iteration-budget exhaustion and
deliver no user-facing reply.
#41755 turned the at-capacity overflow error into a *commanded* in-turn retry
("...then retry this add — all in this turn"); combined with the fragile
substring-only `replace`/`remove` matching (LLMs can't reliably re-quote a long
entry verbatim), the model loops add↔replace on inexact guesses until the turn
dies. The existing tool_guardrails halt would catch this, but hard_stop_enabled
is opt-in (off by default), so a default install still hangs.
This fixes it at the memory layer without changing global guardrail behavior:
- MemoryStore tracks per-turn consolidation failures; after a cap (3) it drops
the "retry in this turn" instruction and returns a terminal "leave memory
unchanged, continue your reply" result, so a failed memory side effect can
never block the turn's reply.
- The counter resets on any successful write (progress) and at each turn
boundary (turn_context.reset_consolidation_failures, guarded via getattr so
plugin memory stores without the method are a no-op).
Co-authored-by: liuhao1024 <sunsky.lau@gmail.com>
- replace() and remove() now return entry previews and current_entries
when no entry matches old_text, matching the multi-match and add-limit
error behavior
- add() limit error also now returns previews for consistency
- Agent can self-correct after a failed replace/remove instead of looping
blindly until turn budget is exhausted with no user response
Follow-up to the per-tool availability derivation: `_snapshot_toolset_checks`
and `_evaluate_toolset_check` had no remaining callers once the four
availability surfaces switched to `_toolset_has_exposable_tools`. Remove both,
drop the no-op `quiet` param from the new helper, and document why
`_toolset_checks` is still written (banner.py reads it via TOOLSET_REQUIREMENTS
to classify unavailable toolsets as lazy-init vs disabled).
Doctor and banner used the first check_fn registered for a toolset, so
desktop-only read_terminal gated the whole terminal toolset even though
terminal and process still expose at runtime.
Fixes#54820
egilewski found the prior sink gate was transient: it only applied while
PluginManager executed register(ctx). A plugin could defer a direct
registry.register(..., override=True) to a post-load callback/thread, after
the scope was cleared, and still replace a built-in.
Make authorization durable by binding it to where the handler is DEFINED
(handler.__globals__['__name__']) rather than to call timing. At load, each
plugin's module namespace is mapped to its allow_tool_override opt-in in a
table that is never cleared. The sink resolves the handler's owning plugin
module and rejects an override from any plugin namespace without opt-in,
regardless of when or on which thread the call happens. Plugin namespaces
with no recorded policy are treated as not-opted-in (fail-closed). Built-in
and MCP handlers live outside the plugin namespace and are unaffected.
Adds a regression test for the delayed/post-load direct-registry override.
The opt-in gate lived only in PluginContext.register_tool, so a plugin
could bypass it by importing tools.registry and calling
registry.register(..., override=True) directly. Enforce the same gate at
the sink: during plugin load, the registry rejects an override from a
plugin without operator opt-in regardless of the path taken. Built-in and
MCP registrations (no active plugin scope) are unaffected.
Adds a regression test covering the direct-registry bypass.
The standalone send path (_send_telegram, used by the send_message tool,
cron delivery, and out-of-process callers) chunked the *raw* message on
UTF-16 length, then formatted and sent the result un-rechunked. MarkdownV2
escaping inflates the text (`!`/`.`/`-` -> `\!`/`\.`/`\-`), so a
4096 UTF-16-unit raw message can become ~8192 units once formatted and gets
rejected by Telegram as 'Message is too long'.
Move all text chunking into _send_telegram, after formatting: split the
formatted MarkdownV2/HTML text on UTF-16 length so every send is <=4096,
with per-chunk plain-text fallback and thread-not-found retry preserved.
Media attaches after all text chunks. (#28557)
Defense-in-depth follow-up to the runtime guard added in the previous commit.
Models on headless hosts (Railway / Fly / Docker / fresh VPS) without any ACP
CLI installed occasionally hallucinate ``acp_command="copilot"`` from the
schema description, despite the explicit "Do NOT set" instruction. The runtime
guard prevented the crash but the model still wasted a tool turn and got an
opaque silent fallback.
This commit removes the temptation at its source: ``_build_dynamic_schema_overrides``
now strips ``acp_command`` and ``acp_args`` from both the top-level and per-task
schemas when none of the known ACP CLIs (``copilot``, ``claude``, ``codex``) are
detectable on PATH. The model literally never sees the fields, so it cannot
pass them.
The runtime guard from the previous commit stays in place as defense-in-depth
for internal callers, tests, and any future code path that bypasses the schema.
``_acp_binary_available`` is intentionally NOT cached: ``shutil.which`` is
cheap, and avoiding the cache means the schema reacts to mid-session installs
without requiring a process restart.
Tests:
- ``test_schema_prunes_acp_command_when_no_acp_binary``
- ``test_schema_keeps_acp_command_when_binary_available``
- ``test_acp_binary_available_checks_known_clis``
Full ``test_delegate.py`` suite: 136/136 pass.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
When a model passes `acp_command="copilot"` (or any other binary name) in a
`delegate_task` tool call, `_build_child_agent` unconditionally sets
`effective_provider = "copilot-acp"`, which routes the subagent through
`CopilotACPClient`. That client spawns the named binary via subprocess; if it
isn't on PATH, every retry raises RuntimeError and an asyncio cleanup race
during error delivery can take the entire gateway down.
This is a real failure mode on headless deploys (Railway / Fly / VPS / Docker)
where `copilot` / `claude` / etc. aren't installed. The schema does say
"Do NOT set unless the user explicitly told you an ACP CLI is installed,"
but models occasionally pass it anyway — particularly for X (Twitter) search
prompts where Grok seems to associate ACP with "search assistance."
Reproduction:
- Headless install (no `copilot` binary on PATH)
- Set provider to xai-oauth + model grok-4.3
- Telegram prompt: "Search X for crypto twitter trends"
- Grok decides to delegate and passes `acp_command="copilot"`
- Subagent crashes 3x, gateway crashes on the 3rd retry teardown
Fix: validate the binary exists on PATH via `shutil.which` before honoring
the override. If missing, log a warning and fall through to the parent's
default transport. No behavior change when the binary IS present (covered
by `test_build_child_agent_honors_acp_command_when_binary_present`).
Tests:
- `test_build_child_agent_ignores_acp_command_when_binary_missing`
- `test_build_child_agent_honors_acp_command_when_binary_present`
Verified on Python 3.11 (macOS) and 3.12 (Debian 13 container).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Concurrent ACP sessions run on a shared ThreadPoolExecutor (max_workers=4).
Each _run_agent mutated the process-global os.environ["HERMES_INTERACTIVE"]
and restored it in finally, so one session's restore could clobber another's
set mid-run — dropping the second session onto the non-interactive
auto-approve path, executing a dangerous command without the approval
callback firing (GHSA-96vc-wcxf-jjff).
Replace the env-var flag with a thread/task-local contextvar in
tools.approval. The two HERMES_INTERACTIVE read sites in approval.py now go
through _is_interactive_cli() (contextvar-first, env fallback for legacy
single-threaded CLI callers). The ACP executor sets the contextvar instead
of os.environ; the existing contextvars.copy_context() wrapper isolates each
session's write.
Co-authored-by: Hermes Agent <127238744+teknium1@users.noreply.github.com>
delegate_task's per-task completion display emitted lines like
"✓ [1/3] Research done (17.92s)" via a bare print(). Under ACP (and any
headless JSON-RPC stdio host where AIAgent routes human output to stderr
via a custom _print_fn), these landed on stdout and corrupted the
protocol frame stream, surfacing as "Failed to parse JSON message: ✓
[3/3] …" in the ACP adapter.
Add _emit_parent_console() which prefers parent_agent._safe_print (the
same hook AIAgent uses for every other user-facing print) and falls back
to print() only when no router is wired up or it raises. CLI behavior is
unchanged.
The PR's other fix (preset toolset expansion) is already covered on main
by _expand_parent_toolsets(), so only the stdio-safe printing change is
salvaged here.
Batch delegation returned each subagent's full final_response verbatim
into the parent's context. A fan-out of N children could dump 60k+ tokens
at once, blowing the parent's context window and — on rate-limited
providers — triggering a compression/429 death spiral (429 misread as
context-too-large -> window step-down -> retry loop -> conversation dies).
Cap each summary against the parent's *remaining* context headroom split
across the batch (not a magic char count). When trimming, mirror the
web_extract convention: spill the full text to cache/delegation (mounted
into remote backends via credential_files._CACHE_DIRS) and return a
head+tail window (75/25, line-snapped) plus a footer with the exact
read_file offset to page the omitted middle. Both the subagent's opening
AND its closing (outcomes / files-changed / issues, which live at the end)
survive in-context, and nothing is lost — the parent can read_file the
full version on any backend.
delegation.max_summary_chars (default 24000) is a static ceiling layered
on top as belt-and-suspenders for models that ignore 'be concise'; 0
disables it. Child prompt tightened to lead with outcomes / bullets.
Co-authored-by: rc-int <rcint@klaith.com>
Three independent security-scanner hardenings, re-homed onto the current
shared threat-pattern architecture (tools/threat_patterns.py):
- approval.py: add bash/sh/zsh/ksh heredoc to DANGEROUS_PATTERNS. The
existing heredoc pattern only covered python/perl/ruby/node, so
`bash <<'EOF' ... EOF` ran arbitrary shell — including exfil pipelines
whose inner commands don't individually match a pattern — with no prompt.
- threat_patterns.py: apply unicodedata.normalize("NFKC", ...) before
pattern matching so full-width / compatibility homographs (e.g.
`cat ~/.hermes/.env`) are folded to ASCII and no longer bypass the
keyword scanners. Invisible-char detection still runs on the raw content
first (NFKC can strip those codepoints).
- code_execution_tool.py: add CREDS/BEARER/APIKEY to _SECRET_SUBSTRINGS so
vars like HERMES_LLM_CREDS, API_BEARER, MY_APIKEY are scrubbed from the
sandbox env. PASS was intentionally dropped from the original proposal —
it false-positives on BYPASS_CACHE / COMPASS_DIR / PASSENGER_HOST while
PASSWORD/PASSWD already cover the credential cases.
The original PR also proposed a 'synonym' injection pattern block
(overlook/forget/set aside/bypass/discard + developer-mode); dropped here
because it false-positives on ordinary AGENTS.md/SOUL.md prose ("don't
forget to follow the rules", "run in developer mode"), exactly the
bossy-English class threat_patterns.py is documented to avoid.
Salvaged from #9028.
Co-authored-by: Hermes Agent <agent@nousresearch.com>
patch_tool extracts V4A patch paths so _check_sensitive_path can refuse
writes to /etc/*, /boot/*, etc. before they reach the low-level file ops.
The extraction regex had two gaps:
1. `*** Move File: src -> dst` was never extracted (regex only matched
Update/Add/Delete), so a Move targeting /etc/crontab skipped the
pre-check and fell back on the narrower file_operations deny list.
2. The regex required `\\s+` after `***` but patch_parser uses `\\s*`, so
`***Update File: /etc/hosts` (no space) parsed + applied while
skipping the check.
Loosen the leading whitespace to \\s* and add a Move regex that checks
both endpoints. Move endpoints also run through the same '..' traversal
rejection as the other V4A headers (closes the sibling gap on current
main, which gained that traversal guard after this PR was opened).
The gateway-lifecycle guard's hermes-CLI pattern required `hermes`
and `gateway` to be adjacent, so a profile flag slipped the agent
past it: `hermes -p ade gateway restart` was not flagged. That is the
exact form from the 2026-04-11 ade-profile self-kill loop. Allow an
optional run of global flags (`-p ade`, `--profile ade`, multiple
flags) between `hermes` and the gateway subcommand.
launchctl self-termination is already covered on main by #33071; this
narrows the only remaining real gap.
Ensure Windows desktop and local terminal teardown kill full process trees so Git Bash descendants cannot survive wrapper exits and accumulate across retries.
Two robustness gaps from the #54843 truncate-store path:
- _store_full_text wrote the full clean page to cache/web with no upper
bound (path.write_text(content)); a multi-MB page → unbounded per-extract
disk write. Cap at MAX_STORED_TEXT_CHARS (2MB, the pre-truncate-store
refusal ceiling) with a marker when capped.
- The truncation footer told the model 'read_file ... offset=<line>' — a
literal placeholder it had to guess. Compute the real starting line of the
omitted middle (head line count + 1) so the first read_file lands in the gap.
Follow-up to the judge gate. judge_goal() is fail-open at the source:
when no auxiliary model is reachable it returns a "continue" verdict
that is indistinguishable from a real "not done yet" judgment. The gate
treated any non-"done" verdict as a rejection, so an unconfigured or
degraded auxiliary model would wedge every goal_mode worker — it could
never close its own task. That contradicted the gate's own "fail-open"
comment.
Probe judge availability before enforcing (the same auxiliary client
lookup judge_goal performs) and only gate when a judge is actually
reachable. When none is, completion proceeds.
Also fix the rejection guidance: kanban_create takes parents=[...], not
parent=.
Add test_complete_goal_mode_allows_when_judge_unavailable covering the
fail-open path; update the rejection test to force the availability probe.
Apply naqerl's review comments on PR #38388:
- Hoist `from hermes_cli.goals import judge_goal` to module-level
imports so an import failure surfaces at module init, not lazily
on the first goal-mode completion (no circular import: hermes_cli
package init is trivial and does not load tools.kanban_tools).
- Narrow the fail-open `try` to wrap only the judge_goal() call.
The verdict check and its rejection `return tool_error(...)` now
live outside the handler, so a failure there can no longer be
swallowed by the broad except.
- Pass `exc_info=True` to the logger.warning call per CONTRIBUTING.md.
Update the test mock target to tools.kanban_tools.judge_goal, since
the hoisted import rebinds the name into this module's namespace.
Prevents workers in goal_mode from bypassing the auxiliary judge by
calling kanban_complete before acceptance criteria are met. The tool
handler now synchronously invokes the goal judge against the task's
title/body and the completion summary. If the verdict is not "done",
the completion is rejected with actionable guidance for the agent.
This keeps kanban_db.py as a pure SQLite wrapper while intercepting
the bypass exactly at the agent tool-call boundary, aligning with
Hermes separation of concerns.
Fixes#38367
Co-authored-by: CommandCodeBot <noreply@commandcode.ai>
Add durable public-URL output and URL-based chaining to xAI Grok Imagine:
- Store generated media on files-cdn with permanent public HTTPS URLs
(public_url: true, no expiry by default).
- Chain by URL: generate -> edit -> extend each take a prior result's
public HTTPS URL (or a data URI / local file for inputs).
- Add provider-specific xai_video_edit and xai_video_extend tools.
- Image generation: public-URL/storage output, multi-reference edits,
and ~/ local-path support for image edits.
Credentials use xAI Grok device-code OAuth (separate PR).
* feat(web_extract): truncate-and-store instead of LLM summarization
web_extract no longer runs an auxiliary LLM over scraped pages. The extract
backends (Firecrawl/Tavily/Exa/Parallel) already return clean, boilerplate-
stripped markdown, so we return it directly: pages within a char budget
(default 15000, web.extract_char_limit) come back whole; larger pages get a
head+tail window plus an explicit footer giving the stored full-text path and
the read_file call to page through the omitted middle. The full clean text is
written to cache/web (mounted read-only into remote backends like the other
cache dirs), so nothing is lost.
Inline base64 images are converted to [IMAGE: alt] placeholders (token bombs
dropped) while real http(s) image URLs are preserved as links so the agent can
still web_extract/vision_analyze them.
Removes process_content_with_llm + the chunked summarizer + check_auxiliary_model
+ _resolve_web_extract_auxiliary. context_references._default_url_fetcher is
updated to the truncate path and its stale data.documents shape read is fixed
to results (it was silently returning empty).
Live before/after eval (firecrawl, 4 URLs): 11.7x faster overall (176.6s ->
15.1s); 10-60x on large pages. Quality identical; findability 4/4 (answer
recoverable from stored full text on every truncated page). web_search is
unchanged.
No own scraper added; no changes to web_search.
* fix(web_extract): add char_limit to execute_code web_extract stub
The new web_extract char_limit param must appear in the code_execution_tool
_TOOL_STUBS signature (and doc line) or test_stubs_cover_all_schema_params
fails — the stub schema must cover every real schema param.
Wire the salvaged _safe_command_timeout() guard into the surviving
open-timeout call site. _get_open_command_timeout() feeds the
browser_navigate 'open' path; this closes the last call site that
could observe a None timeout from a torn cache (#14331), since the
original PR's max(_get_command_timeout(), 60) site no longer exists
on main (now routed through _get_open_command_timeout).
The original cap held a process-global slot across the WHOLE vision
analysis (image load + encode + LLM call) with a default of min(CPUs, 4).
That serialized legitimate multi-image workflows — "compare these 6
screenshots", "read this 10-page scan", "analyze every frame" — behind a
4-wide gate, and on the native fast path it even throttled calls that make
no LLM request at all. Excess calls queued (blocking acquire, nothing
dropped), but the latency hit on real fan-out was the wrong tradeoff.
The incident was CPU exhaustion, not call count: concurrent base64/resize
bursts saturated every core and left none to service the shared event loop
serving /api/status. So cap ONLY that:
- A dedicated, bounded ThreadPoolExecutor (_vision_cpu_executor) runs the
encode/resize/dimension-check off the caller's loop, sized to the host's
usable core count with NO fixed ceiling — the cap tracks the actual
exhausted resource (cores), not a magic number. Excess encodes queue on
the executor; cores stay free for the loop.
- The LLM call is deliberately OUTSIDE the executor, so multi-image
workflows keep full request concurrency.
- Override via auxiliary.vision.max_concurrency / HERMES_VISION_MAX_CONCURRENCY
(honored verbatim, including above core count); sub-1 ignored.
- _vision_concurrency_slot() is now a no-op shim for back-compat.
Tests assert: resolver defaults to host cores with no ceiling; env/config
override (incl. above cores); sub-1 rejection; the executor is dedicated and
core-sized; encode runs on a vision-encode thread; and crucially that encode
bursts are bounded to the cap while the analyses themselves stay fully
concurrent (calls_peak > cap).
A single agent turn can fan out N vision_analyze calls at once — the
classic trigger is "analyze every frame of this video", where ffmpeg
explodes a clip into dozens of frames and the model calls vision_analyze
on each. Every call does a CPU-heavy base64-encode/resize burst AND holds
a long-lived LLM stream open. The tool executor runs concurrent tool calls
on a per-session ThreadPoolExecutor (_MAX_TOOL_WORKERS=8), and multiple
agent sessions share one process (the dashboard runs the agent in-process),
so there was no global ceiling. In prod (June 2026) a video-frame fan-out
pinned a worker thread at ~100% CPU and starved the shared asyncio event
loop that also serves the dashboard's /api/status liveness probe, flapping
the instance to UNHEALTHY even though nothing had crashed.
Add a process-global threading.BoundedSemaphore that bounds how many vision
analyses run concurrently across the whole process, held across the entire
analysis (image load + encode + LLM call) in the single _handle_vision_analyze
chokepoint (covers both the native fast path and the legacy aux-LLM path).
It is a threading semaphore, NOT asyncio: each vision call is dispatched
through model_tools._run_async on a per-thread event loop, so an asyncio
primitive bound to one loop cannot coordinate across them. The acquire is
offloaded via run_in_executor so waiting for a slot never blocks the calling
loop.
Default: min(host CPUs, 4), floored at 1 — respect the host's concurrency,
or lower. Override via auxiliary.vision.max_concurrency (config.yaml) or
HERMES_VISION_MAX_CONCURRENCY (env). Values < 1 are ignored so the cap can
never be disabled into an unbounded fan-out.
Tests: bounded-fan-out regression guard + a control proving it would fail
without the cap; resolver tests for host-cpu default, ceiling clamp, low-cpu
host, env override, and sub-1 rejection. Pre-existing handler tests updated
for the now-async _handle_vision_analyze. Verified via the real
registry.dispatch -> _run_async per-thread-loop path (16 concurrent calls,
peak bounded to cap).
When a Camofox browser tab is garbage collected (idle timeout, browser
recycle), the held tab_id becomes stale. The next browser_navigate call
hits /tabs/{stale_id}/navigate -> HTTP 404 -> unhandled HTTPError.
Catch the 404 in camofox_navigate, clear the stale tab_id, and create a
fresh tab via _ensure_tab. The agent recovers transparently without
requiring a session restart.
Other tab operations (snapshot, click, type, etc.) use the same pattern
but only fail if the tab dies between successful calls — much rarer.
The navigate fix covers 95%+ of cases since navigate is always the entry
point.
The Camofox browser backend hardcoded a 30s HTTP timeout via
_DEFAULT_TIMEOUT, ignoring the user's browser.command_timeout config.
The main browser_tool path already reads this config via
_get_command_timeout().
This commit adds an equivalent _get_command_timeout() to
browser_camofox.py that reads browser.command_timeout from config
with caching, and switches all HTTP helper methods (_post, _get,
_get_raw, _delete) to use it as the default timeout.
Fixes#40843
The five HTTP call sites in browser_camofox.py (_ensure_tab, _post,
_get, _get_raw, _delete) did not include Authorization headers, causing
403 Forbidden when the Camofox server has API key auth enabled.
Added _auth_headers() helper and wired it into all five call sites.
The health check endpoint (/health) is left without auth since it is
a connectivity probe, not a browser operation.
Regression test covers: header present when key set, absent when unset,
blank key produces empty headers.
Fixes#20476
The Camoufox REST API server expects `listItemId` in the `POST /tabs`
body, but `_ensure_tab` was sending `sessionKey`. This caused a 400
Bad Request on every `browser_navigate` call.
The parameter name mismatch is visible in the same file: line 283
already reads `tab.get("listItemId")` when adopting existing tabs,
confirming the server-side field name.
Fixes#37960
The supermemory and mem0 memory providers shipped third-party SDKs
(supermemory / mem0ai) that are not core dependencies, but — unlike the
honcho and hindsight providers — they imported those SDKs directly with
no tools.lazy_deps.ensure() preflight and had no LAZY_DEPS allowlist
entry. On the published Docker image the agent venv is sealed
(HERMES_DISABLE_LAZY_INSTALLS=1) and lazy installs are redirected to a
writable durable target (HERMES_LAZY_INSTALL_TARGET). honcho/hindsight
route through ensure() and install fine there; supermemory/mem0 never
called it, so their SDK was never installed on a hosted instance and the
provider silently reported itself unavailable even with the API key set.
Fixes:
- Add memory.supermemory + memory.mem0 to the LAZY_DEPS allowlist
(tools/lazy_deps.py), pinned to current PyPI releases.
- Call ensure('memory.<x>', prompt=False) at each SDK-import chokepoint
(_SupermemoryClient.__init__; Mem0MemoryProvider._create_backend),
mirroring honcho's wrapped try/except shape.
- Drop the SDK-import gate from supermemory's is_available() — it was a
chicken-and-egg trap (provider never loaded on a sealed venv, so
ensure() never ran). Now key-presence only, like honcho/mem0.
- Add matching pyproject extras [supermemory]/[mem0]; update the
lazy-covered-extras contract test (excluded from [all] by policy).
Tests prove each path fails without the fix and the real sealed-venv
durable-target gate accepts both features.
Make the read-only agent terminal mirrors stream in real time and give
the agent a desktop-only way to dismiss its own tabs.
- Stream background output live: the local reader used a blocking
read(4096) that buffered small periodic output until EOF, so agent
tabs only "filled in" at process exit. Switch to buffer.read1(4096)
(decoded) for incremental chunks.
- Route agent.terminal.output / terminal.close to the window that owns
the process (its gateway session) instead of an empty session id, so
events actually reach the desktop renderer.
- Add close_terminal: a HERMES_DESKTOP-gated tool (sibling of
read_terminal) that drops a process's read-only tab WITHOUT killing it
via process_registry.on_close; output keeps buffering and the user can
reopen from the status stack.
- ⌘W now closes a focused agent tab: mark the agent instance
data-terminal and focus it on activation so isFocusWithin routes there.
- ensureTerminal() no longer spawns an extra user shell when a tab
already exists (e.g. opening a background task from the status stack).
_normalize_approval_mode() previously accepted any string, so an unknown
value like 'auto' fell through every downstream mode check (off/smart) and
silently behaved like manual with no signal. Validate against the known
modes (manual/smart/off), emit a warning for anything else, and default to
manual to match the config default and the rest of the function.
Bug 1 from the original PR (/approve & /deny bypassing the running-agent
guard) already landed on main independently, so only the mode-validation
fix is salvaged here.
Fixes#4261
Co-authored-by: Hermes Agent <agent@nousresearch.com>
All three .env parsers use `line.partition("=")` without stripping the
bash-compatible `export ` prefix first. A line like `export API_KEY=sk-...`
produces key `"export API_KEY"` instead of `"API_KEY"`, silently ignoring
the variable and causing auth failures for users who copy-paste from
bash profiles or follow tutorials that include `export`.
- tools/skills_tool.py: `load_env()` for skill environment
- hermes_cli/config.py: `load_env()` for core config
- hermes_cli/main.py: `_has_any_provider_configured()` inline parser
Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
* fix(terminal): require approval for host-bound Docker commands
The Docker terminal backend blanket-skips dangerous-command approval on
the assumption that the container is isolated from the host. That holds
only when nothing is bind-mounted in. Once a host path is exposed (via
TERMINAL_DOCKER_MOUNT_CWD_TO_WORKSPACE or a host-path entry in
TERMINAL_DOCKER_VOLUMES), a command like `rm -rf /workspace` reaches
real host files but is still auto-approved.
Detect host bind mounts and route those sessions through the normal
approval flow. Isolated Docker keeps the fast path. The same gating is
applied to the execute_code guard, which had the identical blanket skip.
Co-authored-by: Hermes Agent <agent@nousresearch.com>
* chore: add AUTHOR_MAP entry for PR #6436 salvage (Kolektori)
* test: accept has_host_access kwarg in _check_all_guards mocks
The host-bound Docker approval fix adds a has_host_access kwarg to the
_check_all_guards wrapper. Six pre-existing tests monkeypatch it with a
fixed (command, env_type) / (cmd, env) lambda signature, which now
raises TypeError when terminal_tool passes the new kwarg. Widen those
mock signatures to accept **kwargs.
---------
Co-authored-by: Kolektori <256073454+Kolektori@users.noreply.github.com>
Co-authored-by: Hermes Agent <agent@nousresearch.com>
On hosts where the cgroup v2 cpu/memory/pids controllers are not delegated
to the docker/podman process (unprivileged Proxmox LXCs, some rootless and
nested setups), --pids-limit/--cpus/--memory cause every container start to
fail with OCI runtime error / exit 126, breaking terminal + execute_code.
- Add _cgroup_limits_available(image): one-shot, host-wide cached probe that
spawns a throwaway container from the sandbox image itself (sleep 0) with
all three flags together, mirroring the existing _storage_opt_supported
probe-and-degrade pattern.
- Remove --pids-limit from static _BASE_SECURITY_ARGS; apply it (default 256
via _DEFAULT_PIDS_LIMIT) in resource_args gated on the probe.
- Gate --cpus and --memory on the same probe.
Behavior unchanged on cgroup-capable hosts; graceful degradation with a
one-time warning where controllers aren't delegated.
Fixes#6568.
(cherry picked from commit c933880b7e)
Co-authored-by: angelos <angelos@oikos.lan.home.malaiwah.com>
Replace the 5s output_tail poll (which often showed nothing) with a real push
stream. The process registry gains an on_output sink called from its reader
threads with each chunk; the tui_gateway wires it to emit agent.terminal.output
{process_id, chunk} (write_json is _stdout_lock-guarded, so emitting from the
reader thread is safe). The desktop routes chunks by process id straight into
the read-only agent xterm via a small writer registry, with a capped backlog so
a tab opened mid-stream (or reopened) replays what it missed.
Drops the fragile poll/tail path: no session-key matching, no truncation, no
lag — full-fidelity ANSI, env-agnostic (local/docker/ssh).
windows_hide_flags() already returns 0 on POSIX (and creationflags=0 is
the no-op default there, exactly how server.py::_list_repo_files does it),
so drop the IS_WINDOWS import + ternary/one-use-dict gating and just pass
creationflags=windows_hide_flags() directly. Tests lose the now-pointless
IS_WINDOWS monkeypatch.
The #54236/#54417 backend git/gh sweep routed git_probe, the repo-file
picker, coding_context, context_references, copilot_auth, and the gateway
process scans through CREATE_NO_WINDOW, but two sibling spawn legs that
also run inside the console-less desktop/gateway backend were missed:
- tools/checkpoint_manager.py `_run_git` (and the one-shot `git init
--bare` in `_init_store`) — when checkpoints are enabled, every
file-mutating turn fires multiple bare `git` calls (status, add,
write-tree/commit-tree, update-ref). Spawned from a parent with no
console (Electron spawns the backend with windowsHide → CREATE_NO_WINDOW),
each one allocates its own conhost window → a flurry of terminal popups.
- tools/skills_hub.py `GitHubAuth._try_gh_cli` — `gh auth token`, the same
bug class as the already-fixed copilot_auth gh probe.
Route both through `windows_hide_flags()` (no-op on POSIX), matching the
established per-site pattern. Tests added to
tests/test_windows_subprocess_no_window_flags.py.
* fix(daytona): quote single-upload mkdir parent path
The single-file _daytona_upload() path shelled out 'mkdir -p {parent}'
with the remote parent interpolated unquoted, so shell metacharacters in
the path could break the command or inject arbitrary commands into the
sandbox. The bulk-upload, bulk-download, and delete paths were already
hardened with shlex-quoting helpers; this single-upload path was missed.
Route it through the existing quoted_mkdir_command() helper and add a
regression test covering a path with shell metacharacters.
Reported by @Gutslabs (#3960); the original branch predated the
file_sync refactor, so the fix is re-applied to the current code path.
* docs(infographic): daytona quote-sync fix
The SSRF cluster (7a6fe9bb, 48f5c425, 7ef04ae7) sealed
browser_snapshot, browser_vision, and _browser_eval against
eval-navigated private pages, but browser_get_images bypasses
_browser_eval and calls _run_browser_command("eval", ...) directly.
An eval-driven navigation to a private address followed by
browser_get_images would leak image src URLs and alt text from the
private page.
Add the same _eval_ssrf_guard_active + _current_page_private_url
recheck before returning image data, matching the pattern established
by the sibling guards.
5 new tests cover: block on private page, allow on public page, skip
for local backend, skip when private URLs allowed, no guard needed on
failed eval.
When a local browser_navigate (or any browser command) fails fast because
Chromium isn't on disk, attempt a one-shot binary download via
`agent-browser install` and retry instead of only printing a hint.
Scope is narrow on purpose:
- binary only, never `--with-deps` (that shells apt/needs root, so missing
system libraries stay a user action)
- gated by `security.allow_lazy_installs` (same opt-out as every lazy install)
- skipped in Docker (Chromium ships in the image)
- attempted once per process
Follow-up to #54353, which made the cold-start failure legible; this closes
the "doesn't actually install the missing browser" gap for the common case.
Local browser_navigate cold-starts the agent-browser daemon and Chromium;
60s was too short on slow Linux hosts and timeouts discarded stderr,
leaving users with a generic failure. Use a 120s floor on first open,
inject --no-sandbox in Docker, include captured daemon output plus install
hints when commands time out, and show "Failed to open" in the desktop
tool chip when navigation returns success=false.
main (cb982ad99) wired windows_hide_flags() into the auxiliary git/gh/wmic/
bash/powershell/taskkill legs but left two it didn't reach, plus the Electron
backend-launch leg it explicitly deferred. Cover them the same way:
- apps/desktop/electron/main.cjs: getNoConsoleVenvPython resolves the BASE
pythonw.exe instead of the venv Scripts\pythonw.exe shim, which re-execs a
console python.exe and flashes a conhost the desktop backend can't suppress.
Both backend creators put the venv site-packages on PYTHONPATH so imports
still resolve under the base interpreter. (main's commit said this Electron
leg "needs a Windows-tested change of its own".)
- tools/tts_tool.py, tools/transcription_tools.py, plugins/platforms/discord:
ffmpeg conversions (voice notes / TTS / STT) via windows_hide_flags().
- plugins/platforms/whatsapp: netstat + taskkill bridge-port cleanup via
windows_hide_flags().
All no-ops on POSIX. Tests assert the base-pythonw preference and the ffmpeg
legs pass CREATE_NO_WINDOW.
OAuth-protected MCP servers (e.g. Hospitable) return 200 text/html on an
unauthenticated HEAD probe — a login/landing page the server cannot substitute
for a real MCP response without a Bearer token. The preflight cannot
distinguish this from a misconfigured URL, so it raises NonMcpEndpointError
before the OAuth browser flow has a chance to run.
Add `and self._auth_type != "oauth"` to the preflight condition in
MCPServerTask.run(). The probe is inapplicable to OAuth servers: their URL
legitimacy is established by .well-known/oauth-protected-resource during the
OAuth handshake, not by a GET content-type check.
Concrete repro: Hospitable (https://mcp.hospitable.com/mcp) returns
`200 text/html` to an unauthenticated httpx HEAD. Without the guard:
✗ NonMcpEndpointError at `hermes mcp test`
With the guard:
✓ Connected (1487ms) — 63 tools discovered
Relation to open PRs:
- #37598 adds a POST probe fallback for POST-only non-OAuth servers (e.g.
DocuSeal), but only passes when POST returns 2xx + MCP content-type.
Hospitable returns 401 on the POST probe (Bearer challenge), so #37598
does not cover this case.
- #49463 extends the POST probe to also pass on non-2xx auth challenges
(making it OAuth-aware), but is labeled duplicate of #37598 and may not
land independently.
This fix is complementary: it handles OAuth servers with zero extra
round-trips rather than adding a POST probe step.
Tests:
- test_oauth_server_html_response_raises_without_skip: documents that
_preflight_content_type raises NonMcpEndpointError for 200 text/html
(the underlying issue), with an OAuth-server docstring.
- test_run_skips_preflight_for_oauth: verifies that run() does NOT invoke
_preflight_content_type when auth_type=="oauth", using class-level
monkeypatching so the gate is exercised without a live MCP transport.
23 passed tests/tools/test_mcp_preflight_content_type.py
When security.redact_secrets is on (default), read_file/search_files/cat
applied redact_sensitive_text(code_file=True) to file content, which still
ran prefix masking. An API key in config.yaml (ghp_..., sk-..., xai-..., etc.)
came back as a head/tail mask like `ghp_S1...Pn2T` — a plausible-looking
truncated key. When an agent read that and wrote it back to config, the masked
value replaced the real credential, silently breaking auth (401). Production
evidence: a config.yaml found containing the exact 13-char masked GitHub PAT.
The two community PRs (#35529, #35534) fixed the corruption by NOT redacting
prefixes for config reads — but that exposes the user's real keys to the agent
context, model, and logs (a security regression). This takes the safer route:
keep redacting, but for file content emit a NON-REUSABLE sentinel.
- New `_mask_token_nonreusable`: prefix secrets -> `«redacted:ghp_…»` (vendor
label preserved for debuggability; zero secret bytes; angle-bracket/ellipsis
wrapper is syntactically invalid as a token so it can't be mistaken for or
written back as a usable key).
- New `redact_sensitive_text(file_read=True)` routes prefix matches through it
(implies code_file=True). Default/log/display mode is UNCHANGED — `_mask_token`
still keeps head/tail (fine for logs, never written back).
- Wired the 3 file_tools.py call sites (read_file / search_files / cat) to
file_read=True.
Fixes both the corruption AND avoids the secret-exposure of the un-redact
approach. 6 new tests (sentinel shape, no-leak, not-a-plausible-key, default
mode unchanged, file_read implies code_file, sk- prefix); 88 redact tests pass;
mutation-verified (reverting to the old mask fails the sentinel/leak tests).
Co-authored-by: liuhao1024 <sunsky.lau@gmail.com>
Co-authored-by: adammatski1972 <289282750+adammatski1972@users.noreply.github.com>
Closes#35519. Supersedes #35529, #35534.
* fix(security): redact secrets in background process + foreground env-dump output
Terminal-output redaction was incomplete (#43025):
- Gap 1: process(action=poll/log/wait) returned background stdout verbatim —
no redaction at all. A background printenv/server/test emitting a key leaked
raw to the model, session.db, and CLI display. Same for the gateway
background-process watcher's completion/progress notifications.
- Gap 2: the foreground terminal path hardcoded code_file=True, which skips the
ENV-assignment pass, so an opaque token (no vendor prefix) from env/printenv
leaked even there.
Adds agent.redact.redact_terminal_output(output, command) as the single policy
for ALL terminal-output surfaces: env-dump commands (env/printenv/set/export/
declare) get the ENV-assignment pass (code_file=False) to mask opaque tokens;
other commands stay on code_file=True to avoid false positives on source dumps.
Wired into terminal_tool, process_registry (_handle_process boundary), and the
gateway watcher. Respects security.redact_secrets (no force) — opt-out preserved.
* docs: add infographic for #43025 terminal-output redaction fix
The snapshot/vision guards re-check the page URL before returning content,
but browser_console(expression=...) -> _browser_eval returns arbitrary JS
results directly, leaving two same-class bypasses open:
1. Direct fetch: fetch('http://127.0.0.1/secret').then(r=>r.text()) reads
a private endpoint and returns the body — the page URL stays public so
the post-eval recheck never sees it.
2. Navigate-then-read: location.href='http://127.0.0.1/' then a later eval
reads document.body.innerText.
Guard _browser_eval on the same condition as navigate/snapshot/vision
(not local backend, not local sidecar, not allow_private_urls):
- pre-scan the expression for private/always-blocked URL literals
- re-check window.location.href after the eval at both success-return
sites (supervisor fast-path + subprocess fallback)
Probe failures fail-open (matching the snapshot/vision guards).
The private-network guard in browser_snapshot() and browser_vision()
blocked all private URLs, including those accessed via local sidecar
sessions (hybrid routing). Local sidecar sessions intentionally access
private URLs — the cloud provider never sees the URL in that case.
Add `_is_local_sidecar_key(effective_task_id)` check to both guards,
matching the existing pattern in browser_navigate().
Fixes#45101 review feedback from egilewski.
The SSRF bypass in #44731 was only patched for browser_snapshot(), but
browser_vision() exposes the same vulnerability — it takes a screenshot
and sends it to the vision model without checking if eval-driven
navigation moved the page to a private/internal URL.
Add the same current-page URL safety check to browser_vision() before
any screenshot is captured, encoded, or forwarded to the vision model.
This covers both the normal screenshot path and the Lightpanda Chrome
fallback path.
7 new tests: blocks private URL, allows public URL, skips in local
backend, skips when private URLs allowed, handles eval failure/empty/exception.
browser_snapshot() now checks the current page URL before returning
content. When browser_console() changes location.href to a private or
internal address (e.g., http://127.0.0.1:8080/), the snapshot returns
an error instead of exposing the private page content.
This closes the SSRF bypass where an attacker could:
1. Navigate to a public page
2. Use browser_console to eval location.href = 'http://127.0.0.1:port/'
3. Use browser_snapshot to read the private page content
The fix reuses the existing _is_safe_url() and _allow_private_urls()
infrastructure, and fails open if the URL check itself fails.
Fixes#44731
The atomic mv approach (kyssta-exe's commit) narrows but does not close the
#38249 race: the temp name used $$ (parent shell PID), which is identical
across &-launched concurrent subshells. Two concurrent writers pick the same
temp file, clobber each other mid-write, and mv then publishes a torn snapshot
— a reader sourcing it absorbs declare-x/export fragments into PATH.
- Use $BASHPID (actual per-subshell PID) so concurrent writers never collide.
- Chain mv on export success (&&) and rm the temp on failure so a partial dump
never replaces a good snapshot; apply the same to the init_session bootstrap.
- shlex-quote the static temp-path portion (Windows/spaces), $BASHPID outside.
- LocalEnvironment.cleanup sweeps orphaned snap.tmp.* temps.
- Regression tests: string-shape + a behavioral concurrent writers/readers test
that proves the snapshot never tears (would still tear with $$).
Fix race condition in terminal environment snapshots that could corrupt
PATH with declare -x entries. When concurrent terminal calls share the
same snapshot file, the non-atomic 'export -p > snapshot.sh' write could
be read mid-write by another process, causing partial/corrupted env vars
to be sourced and mixed into PATH.
The fix uses atomic file replacement:
- Write to a temp file: export -p > snapshot.sh.tmp.303651
- Atomically replace: mv -f snapshot.sh.tmp.303651 snapshot.sh
On POSIX, mv within the same filesystem is atomic, so source() will
either see the old complete snapshot or the new complete one, never a
partial/truncated file.
Fixes#38249
When tools.environments.local can't be imported (partial install,
import-time error), _is_hermes_provider_credential() returned False —
fail-open. A skill could then register a Hermes provider credential
(ANTHROPIC_API_KEY, etc.) as env passthrough; _scrub_child_env lets
passthrough vars bypass the secret-substring net (rule 1), so the
operator's real key would land in the execute_code child. Reopens the
GHSA-rhgp-j443-p4rf bypass.
Fail closed instead: on import failure, treat the name as a protected
provider credential and refuse passthrough. Regression test exercises
the full register -> scrub path under a simulated import failure.
Co-authored-by: Hermes Agent <noreply@nousresearch.com>
Secret redaction is display/output-scoped on main — write_file writes
content verbatim, terminal/execute_code redact only output not the
command/source. The real bug is in displayed tool OUTPUT (read_file,
terminal, execute_code):
_DB_CONNSTR_RE's password group [^@]+ was greedy across newlines, so on a
multi-line block it scanned past the DSN line to the next stray '@' (a
Python @decorator), replacing every intervening character — including line
breaks — with ***. That dropped lines and concatenated the next line onto
the f-string line, making read_file output look corrupted (the file on disk
was always correct). Reported in #33801.
Fix:
- Forbid whitespace in the userinfo/password groups ([^:\s]+ / [^@\s]+) so
the match can never span a line break. A real DSN password never contains
whitespace. This alone kills the catastrophic line-dropping.
- Under code_file=True, preserve a password group that is a pure {...} brace
expression — f"postgresql://{user}:{pass}@{host}" is an f-string template,
not a live credential. Literal passwords are still masked.
- Pass code_file=True at the terminal and execute_code output redaction call
sites (file_tools already did) so code-execution output isn't corrupted by
ENV/JSON/template false positives. Real prefixes, auth headers, JWTs, and
private keys are still redacted.
Verified E2E against the reporter's exact pydantic-settings module: file
written verbatim, read_file shows the DSN f-string + @model_validator intact
with zero *** corruption, while a literal postgresql://admin:pw@host DSN and
a real sk- key are still masked.
Reported-by: koishi70
Reported-by: pfrenssen
The 600s default evicted the gateway clarify entry while users were
still away (meeting/AFK); a later button tap then landed on a dead
entry and the agent hung on 'running: clarify'. Raise the default to
1h in DEFAULT_CONFIG and the get_clarify_timeout() code-level fallback,
documenting the running-agent-guard tradeoff. User overrides still win.
Two follow-ups on top of the salvaged #46365 fix:
1. Tests: the salvaged tests injected the ephemeral MatrixAdapter via
sys.modules["gateway.platforms.matrix"], but Matrix migrated to a plugin
(#41112) and the fallback now imports from plugins.platforms.matrix.adapter.
Point the three sys.modules patches at the current module path so the
ephemeral-fallback tests actually exercise the injected fake adapter.
2. Harden the live-adapter lookup: split the gateway import guard from the
adapter lookup and log (instead of silently swallowing) when a runner
exists but adapters.get() raises. A silent fall-through there would
re-introduce the per-send reconnect/OTK-exhaustion storm this fix exists
to prevent (#46310). Documented that the live adapter is gateway-owned and
must not be disconnected, and why the ephemeral finally never touches it.
When a live gateway adapter is available (i.e. the tool runs inside a
running gateway), reuse the persistent connection instead of creating a
new MatrixAdapter per call. This eliminates per-message E2EE re-init
storms that exhaust recipient OTKs and silently drop messages.
The fix follows the same pattern as _send_to_platform (line 618):
gateway_runner_ref → runner.adapters[Platform.MATRIX]. Falls back to
the ephemeral connect/disconnect cycle for standalone contexts.
Also extracts the shared send logic into _send_via_matrix_adapter()
to avoid duplicating the media dispatch code between the two paths.
Fixes#46310
Fixes#28126. sync_skills() was unconditionally writing bundled skills
into the local <profile_home>/skills/ tree even when the profile's
config.yaml delegated skill resolution to an external directory
via skills.external_dirs. The skill loader then saw two candidates
for the same name (local shadow + external canonical), refused to
resolve on collision, and every worker that auto-loaded such a skill
crashed with 'Unknown skill(s): <name>'.
Changes:
- _build_external_skill_index() indexes skills available in external
dirs (by directory name and frontmatter name)
- sync_skills() skips writing a bundled skill when it finds the same
name in the external index; records the hash in the manifest so
subsequent syncs treat it as already handled
- Self-healing: removes stale local shadows left by prior buggy syncs
(only when origin_hash == bundled_hash == user_hash, i.e. we wrote
it and user didn't touch it)
- New 'shadowed_by_external' key in sync_skills() return dict
3 new tests in TestExternalDirsIndexing (all passing).
All 48 tests in test_skills_sync.py pass.
Closes#28126
The curator's LLM consolidation pass could archive whole clusters of
active skills with zero verified consolidations (#29912): a bare prune
(skill_manage delete with absorbed_into empty/omitted) from the forked
review agent was accepted, removing the skill's name from lookup even
though counts.consolidated_this_run was 0.
- _delete_skill now fails closed during the curator/background-review
pass: a delete is only allowed when it declares a verified
consolidation (absorbed_into=<umbrella>, umbrella must exist). A prune
with no forwarding target is refused; the skill stays active. The
deterministic inactivity prune (archive_skill) is unaffected.
- A verified consolidation delete during the curator pass now routes
through the recoverable archive primitive instead of shutil.rmtree, so
a misjudged consolidation can be undone with hermes curator restore.
The usage record is kept (state=archived) rather than forgotten.
- Foreground, user-directed deletes keep their existing hard-delete
semantics.
Follow-up to the cherry-picked #29212 (#29177):
- Promote the 24h stale-process threshold to config.yaml
(session_reset.bg_process_max_age_hours) instead of a hardcoded
constant. 0 disables the cutoff (legacy: any live process blocks reset).
Wired through GatewayConfig.default_reset_policy in gateway/run.py.
- Bug 2: process(action=list) now resolves the gateway session_key from
the contextvar and surfaces session-scoped background processes (a
forgotten preview server under a different task), flagged
session_scoped — so the agent/user can discover and kill the blocker.
Previously the task-scoped list returned [] and the blocker was invisible.
- Tests: config round-trip for the new field, cross-task list visibility.
- Docs: messaging session-reset section.
Background processes (e.g. http.server preview) that Hermes starts and
forgets about previously blocked session idle/daily reset indefinitely.
The reset guard in session.py checked has_active_for_session() with no
max age — a 3-day-old preview server blocked reset the same as a task
started 30 seconds ago.
Changes:
- Add max_active_age parameter to has_active_for_session() in
process_registry.py. Processes older than this threshold are ignored.
- Add MAX_ACTIVE_PROCESS_AGE constant (24h / 86400s).
- Wire max_active_age into the gateway's session store callback in
run.py so stale processes no longer block session lifecycle.
- Add debug logging when reset is skipped due to active processes.
- Add 3 tests covering recent, stale, and legacy (None) max age.
Fixes#29177
The module-level import broke tests/tools/test_managed_browserbase_and_modal.py,
which loads browser_tool.py via spec_from_file_location against a stubbed
'tools' package that does not include tools.environments.local. Move the import
into a _build_browser_env() helper called at the two agent-browser spawn sites,
matching the lazy-import pattern already used by lazy_deps.py.
Subprocesses spawned outside the terminal/execute_code path (agent-browser,
copilot ACP, dep-ensure, lazy_deps uv install, TUI Node host, cli.exec)
inherited the operator's full credential environment via os.environ.copy().
The terminal path was already scrubbed by _HERMES_PROVIDER_ENV_BLOCKLIST
(#1002/#1264/#32314); these spawn sites bypassed it.
Adds hermes_subprocess_env(inherit_credentials=) in tools/environments/local.py
reusing the existing dynamic blocklist as the single source of truth:
- Tier 1 (_ALWAYS_STRIP_KEYS): gateway bot tokens, GitHub auth, infra
secrets -- stripped even for credential-inheriting children.
- Tier 2 (_HERMES_PROVIDER_ENV_BLOCKLIST): provider/tool keys -- stripped
unless inherit_credentials=True. The opt-in is grep-able for audit.
Browser worker keeps a _BROWSER_PASSTHROUGH_KEYS allowlist (BROWSERBASE/
FIRECRAWL) re-added after the strip. Model-driving children (ACP, TUI Node
host, cli.exec) use inherit_credentials=True so they still get provider keys
while losing Tier-1 secrets. Installers (dep-ensure, lazy_deps) inherit
nothing sensitive. cua_backend already routed through _sanitize_subprocess_env
on main -- left as-is. Gateway adapter utility spawns (gh pr comment, ffmpeg)
are left inheriting env: gh needs GH_TOKEN by design, ffmpeg is a trusted
system binary -- no untrusted-dependency exposure.
This is defense-in-depth (personal-assistant trust model: same-user spawns),
making the existing scrub policy uniform across the spawn surface; the main
real payoff is shrinking the blast radius if a transitive npm dep in
agent-browser is compromised.
Reconstructed on current main from the design in #31959 (Tranquil-Flow);
also credits #39003 (rodboev), #37843 (coygeek), #35769 (egilewski).
Co-authored-by: Tranquil-Flow <tranquil_flow@protonmail.com>
Co-authored-by: rodboev <rod.boev@gmail.com>
Co-authored-by: egilewski <egilewski@egilewski.com>
The durable _last_known_cwd anchor is keyed by the shared 'default' container,
so a non-owning worktree session could inherit the owning session's cwd through
it — breaking the wrong-worktree-routing fix (test_file_tools_cwd_resolution::
test_resolution_routes_to_resolving_sessions_worktree).
Reorder _authoritative_workspace_root so the session-specific registered cwd
override (keyed by raw session id) is checked BEFORE the shared-container
_last_known_cwd fallback. A non-owning session now resolves into its own
registered worktree; the durable anchor only fills in when there's no
session-specific override (the #26211 single-session case). Adds a regression
test covering the owner-mirrors-then-other-session-resolves interaction.
Belt-and-suspenders on top of the cherry-picked cwd-preservation fix:
- Proactively mirror every live terminal cwd into _last_known_cwd on each
successful read, so the durable anchor survives even when the cleanup
thread pops both _file_ops_cache and _active_environments before
_get_file_ops' stale-cache save branch can fire.
- Fall back to _last_known_cwd in _authoritative_workspace_root. write_file_tool
resolves the path (via _resolve_path_for_task) BEFORE _get_file_ops rebuilds
the env, so restoring only the rebuilt env's cwd was insufficient — the
resolution that decides where the file lands runs first. This closes that gap.
The local env's persisted _cwd_file can't serve this role: it's keyed by a
random per-session uuid and deleted on cleanup (the same cleanup that triggers
the bug). The in-memory _last_known_cwd registry is the durable anchor instead.
Adds a real-IO E2E regression (TestSilentFileMisplacementE2E) exercising the
actual write_file_tool path after env cleanup.
Root cause: when the terminal environment (`_active_environments` entry) is
cleaned up and re-created during a long conversation, the new environment
always starts with the default config CWD (typically `~/.hermes/hermes-agent`)
instead of preserving the user's last-known working directory. Subsequent
relative-path writes (`write_file`, `execute_code`, shell commands) silently
land in the default CWD, making files appear to be "created but absent."
Fix: add `_last_known_cwd` dict that preserves the old environment's CWD
before the stale cache entry is invalidated. When a new environment is
created for the same task_id, we check `_last_known_cwd` first and use the
preserved CWD instead of the config default.
Changes:
- tools/file_tools.py: add `_last_known_cwd` dict, save CWD before stale
cache invalidation, restore CWD on env recreation
- tests/tools/test_file_tools.py: add `TestLastKnownCwd` with 2 tests
verifying CWD preservation and fallback behavior
Fixes#26211
A flaky external probe in a tool's check_fn (e.g. check_terminal_requirements
running `docker version` with a 5s timeout, momentarily timing out under load)
would return False for a single get_tool_definitions() call. Because file
tools delegate their check_fn to the terminal check, that one flake silently
stripped read_file/write_file/patch/search_files AND terminal from whatever
agent was being constructed at that instant — most visibly a delegate_task
subagent, which then reported "Tool read_file does not exist". This explains
both the intermittent (~80% success) user-session failures and the
deterministic cron failures in #21658 / #5304.
The existing _check_fn TTL cache made this worse: it cached the transient
False for the full 30s window, poisoning every subagent spawned in that span.
Fix: remember the last time each check_fn returned True; when a fresh probe
fails within a short grace window of that success, treat it as a flake —
serve the last-good True and do NOT cache the failure (so the next call
re-probes). A failure with no recent success, or past the grace window, is
honored normally so a backend that genuinely went down stops advertising its
tools. Probe failures now log at WARNING regardless of quiet mode, making the
previously-silent tool loss diagnosable in subagent (quiet) sessions.
Co-authored-by: Stuart Horner <5261694+djstunami@users.noreply.github.com>
Follow-up to #53791 addressing review feedback: the footgun checker treated
capture_output=/stdout=/stderr=/check_output as proof a subprocess can't pop a
Windows console. That invariant is false — stream redirection controls where a
child's output goes, not whether a console is allocated. From a console-less
parent (Desktop/Electron, pythonw.exe, detached gateway/cron) a console-subsystem
child still flashes a window even when fully captured.
- check-windows-footguns.py: capture/redirect/check_output is no longer a blanket
safe-pass. Added _WINDOWS_FLASHING_PROGRAMS (git/gh/npm/node/python/uv/ffmpeg/
docker/powershell/…); calls to those are flagged even when captured. Non-flashing
programs keep the capture exemption (no 271-site noise). _subprocess_compat.run/
popen calls are inherently safe (wrapper injects CREATE_NO_WINDOW).
- Routed the 35 genuine flashing git/gh/npm/uv/ffmpeg/docker spawns through the
_subprocess_compat.run/popen chokepoint (Brooklyn's wrapper from #53810) — the
durable fix, not per-site annotations. cmd.exe /c start stays # ok (intentional).
- Updated tests + CONTRIBUTING.md rule #17 to the corrected invariant.
* fix(windows): stop terminal-window popups from background spawns
Native-Windows desktop/gateway users saw cmd/conhost windows flash on
gateway restart, image paste, the dashboard Projects tree, voice notes,
and ~5 min after closing the app (detached cron). Two root causes:
- Console-subsystem exes (taskkill, schtasks, wmic, netstat, tasklist,
agent-browser, git, ffmpeg, powershell, git-bash) spawned via raw
subprocess allocate a fresh console when the launching process has
none (pythonw desktop backend / detached gateway) - even with output
captured.
- uv venv pythonw shims re-exec console python.exe, so Python children
get a console regardless of how they're launched.
Fixes:
- Single hidden-spawn primitive (_subprocess_compat.run/.popen) that ORs
CREATE_NO_WINDOW on Windows, no-op on POSIX. Route every Hermes-owned
console-exe spawn through it.
- FreeConsole() catch-all in hermes_bootstrap: any Python child that
exclusively owns an auto-allocated console detaches it at startup
(GetConsoleProcessList()==1 gate leaves shared interactive consoles
untouched).
- Replace PowerShell/wmic gateway PID scans with in-process psutil.
- Skip schtasks queries on non-interactive desktop restarts.
- Prefer native agent-browser .exe over .cmd shims.
- Guard test bans raw subprocess spawns of the Windows-only console
tools repo-wide so the popup class can't regress.
* fix(windows): scope FreeConsole to background entry points; fix merge fallout
Console detach review (per #53810 feedback): GetConsoleProcessList()==1 can't
tell a uv pythonw->python phantom console apart from a user opening the
interactive CLI/TUI in its own fresh console (double-click, shortcut, ConPTY) —
both report a single attached process with a tty. Running FreeConsole() in the
import-time bootstrap therefore risked detaching a legitimately-interactive
terminal.
- Extract FreeConsole into explicit hermes_bootstrap.detach_orphan_console();
remove it from apply_windows_utf8_bootstrap() (import side effect).
- Call it only from known background mains: gateway run, dashboard backend
(start_server, what the desktop spawns), cron standalone, tui_gateway entry,
slash worker. Interactive CLI/TUI never calls it.
- Behavior-contract tests: frees only when solo owner, leaves shared console,
no-op without console / on POSIX, and asserts it's not an import side effect.
Merge fallout from origin/main (#53791):
- local.py: 3-way merge left a dangling **_popen_kwargs (NameError crashing
every terminal init). _subprocess_compat.popen already hides the window, so
drop it.
- discord adapter: merge stacked an undefined windows_hide_flags() onto the
primitive call; drop the redundant arg.
- test_gateway: scan now goes psutil-first (zero spawn); rewrite the
case-variant test to drive that production path.
* test(claw): mock _subprocess_compat.run seam for Windows process scan
claw.py's Windows tasklist/powershell scan routes through the hidden-spawn
primitive; the tests still patched claw_mod.subprocess, so on win32 the mock
was never hit and real spawns returned nothing. Patch the actual seam.
* fix(windows): stop subprocess console-window popups + add CI guard
The single biggest source of Windows 'terminal popup' bug reports was bare
subprocess.run/Popen calls spawning a console window. The compat helpers
(windows_hide_flags / windows_detach_popen_kwargs) already existed but the
footgun checker had no rule to stop new bare calls from reintroducing the flash.
- scripts/check-windows-footguns.py: new AST-based rule flagging subprocess
calls that can create a new console — output-redirection-aware (capture/
redirect/check_output exempt) and POSIX-only-program-aware (launchctl/
systemctl/brew/etc. exempt). Comprehensive on real popups, no annotation
burden on calls that can't flash.
- Swept all genuine window-spawning sites through windows_hide_flags()/
windows_detach_popen_kwargs(); marked intentionally-visible launches
(editor/terminal/foreground re-exec) with '# windows-footgun: ok'.
- tests/scripts/test_windows_footgun_subprocess_rule.py: behavior-contract
tests + full-repo cleanliness invariant.
- CONTRIBUTING.md: documents the rule + the helper pattern.
* test: accept creationflags kwarg in psutil_android fake_subprocess_run
The Windows no-window sweep added creationflags=windows_hide_flags() to
install_psutil_android.py's subprocess.run call; the test's fake stub had a
fixed (cmd) signature and raised TypeError on the new kwarg.
The Hermes gateway runs inside its own venv, so its process environment
carries VIRTUAL_ENV (and possibly CONDA_PREFIX). The terminal tool spawned
subprocesses inheriting those markers. When the agent ran `uv sync`,
`uv pip install`, `poetry install`, etc. in ANY other project directory,
those tools honored the inherited VIRTUAL_ENV and rebuilt/synced that
project's dependencies into the Hermes venv path — wiping Hermes' own runtime
deps (and, when the other project pinned a different Python, replacing the
interpreter), bricking the gateway on the next restart (#23473).
Strip VIRTUAL_ENV/CONDA_PREFIX in both subprocess-env construction points in
tools/environments/local.py — `_sanitize_subprocess_env` and `_make_run_env`
— via a shared `_ACTIVE_VENV_MARKER_VARS` constant. The Hermes venv stays
reachable because its bin dir is already first on PATH, so removing the
active-environment markers is safe and only prevents the cross-project clobber.
Adds TestActiveVenvMarkerStripping: end-to-end (markers in os.environ don't
reach the spawned subprocess) and unit coverage for both functions, plus a
guard on the marker constant.
Also adds the AUTHOR_MAP entry for the salvaged contributor.
Closes#23473
check_all_command_guards() swallowed ImportError from tools.tirith_security
with an unconditional pass, leaving tirith_result["action"] as "allow"
regardless of security.tirith_fail_open. When an operator sets
tirith_fail_open: false they have explicitly opted into fail-closed
behaviour; a missing or broken Tirith module must not silently permit
command execution.
Inside the except ImportError handler, read the live security config.
When tirith_enabled is true and tirith_fail_open is false, synthesise a
"warn"-action Tirith result so the command flows through the normal
approval path (prompt the user, or block in cron/gateway contexts)
instead of bypassing it. The default tirith_fail_open: true behaviour
is unchanged.
Adds three regression tests to tests/tools/test_approval.py:
- fail_open=true + ImportError → silently allowed (no regression)
- fail_open=false + ImportError → approval callback invoked, command denied
- tirith_enabled=false → always allowed regardless of fail_open
Fixes#20733
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
# Conflicts:
# tests/tools/test_approval.py
Cron jobs accumulate large volumes of repetitive vocabulary (recurring
project names, dates, summaries) and out-number a user's interactive
sessions. Under bare BM25 they dominate the top FTS rows, so discover's
early-exit-at-N dedup collects only cron sessions and the user's own
conversations never surface — "recall blindness" (#19434).
- _order_for_recall() stable-sorts FTS rows so interactive sources rank
above cron before lineage dedup; within each class BM25/recency order
is preserved. Cron is demoted, not excluded, so it still surfaces when
it is the only match.
- raise discover scan limit 50 -> 300 so buried interactive matches are
in hand for the demotion pass.
Fixes the cron-flooding sub-bug of #19434. The split-brain sub-bug is
covered by #52798; the child-session sub-bug is superseded by in-place
compaction.
send_message with MEDIA:/path to a WhatsApp target previously dropped the
attachment: the WhatsApp branch never passed media_files, the plugin's
_standalone_send accepted the param but only POSTed text, and WhatsApp was
absent from the media-supported platform list.
- send_message_tool: add a Platform.WHATSAPP media block (mirrors Feishu) that
routes media_files through the whatsapp plugin's standalone_sender_fn, and
add whatsapp to the supported-media list strings.
- whatsapp adapter: _standalone_send now sends text first (skipped when the
chunk is media-only), then uploads each file via the bridge /send-media
endpoint with a mediaType derived from extension/is_voice/force_document, so
images/videos/voice arrive as native bubbles instead of documents.
- _bridge_media_type classifier maps ext -> image|video|audio|document.
Closes#19105 (remaining send_message gap). Other items in the report
(inbound video paths, image_generate auto-deliver, history dedup, native
gateway bubbles) already landed on main.
A long-running gateway session could permanently lose an MCP server: once a
stdio subprocess died (or transient drops accumulated over the session), the
run loop exhausted its reconnect budget and returned, orphaning the task. With
no listener for _reconnect_event, the circuit breaker's half-open probe could
never revive the server — every probe hit a dead/absent session, re-armed the
60s cooldown, and looped forever until a full gateway restart (#16788).
Root cause was split ownership of transport liveness between the run loop and
the tool handler, plus a permanent give-up path. Fixed by one invariant: a
non-shutdown server task is always reconnectable.
- run loop parks (deregisters phantom tools, then awaits _reconnect_event)
instead of returning when the reconnect budget is exhausted, so the task
stays alive as a dormant listener
- retry budget resets on every successful (re)connect, so a healthy
long-lived server can't accumulate lifetime drops into a death sentence
- half-open probe with no live session signals a reconnect (reviving a
parked/dead task and respawning a dead stdio subprocess) and returns a
clean 'reconnecting' error instead of writing into a dead pipe
- breaker resets on successful session init across all transports
(stdio/HTTP/SSE) — fully transport-agnostic, no PID/pipe polling
Builds on the closed-PR cluster for this issue: keeps #49255's deregister-on-
exhaustion insight and #21006's signal-don't-probe insight, discards the racy
os.kill PID machinery.
Co-authored-by: LeonSGP43 <LeonSGP43@users.noreply.github.com>
Co-authored-by: srojk34 <srojk34@users.noreply.github.com>
TELEGRAM_HOME_CHANNEL set to an @username (not a numeric chat ID) crashed
all webhook/cron->Telegram home-channel delivery with 'ValueError: invalid
literal for int()'. The Telegram Bot API accepts both a numeric chat_id and
an @username string; Hermes was force-coercing every chat_id with int().
Add normalize_telegram_chat_id() (returns int for numeric values, passes
@username strings through) and apply it at the Bot API send/edit sites in
the Telegram adapter and the send_message tool. Username targets are now
recognized as explicit targets in _parse_target_ref.
Reapplies the approach from #13274 (season179), whose branch predated the
gateway/platforms/telegram.py -> plugins/platforms/telegram/adapter.py
relocation. Dupes: #13535 (Tranquil-Flow), #37572 (chewkaah).
Co-authored-by: season179 <season.saw@gmail.com>
When an MCP server requires OAuth, the interactive `hermes` TUI froze on
startup: background MCP discovery hit the OAuth flow, which on an interactive
TTY spawns a daemon thread doing a blocking `sys.stdin.readline()` (the
"paste the redirect URL" fallback in mcp_oauth._wait_for_callback). That
thread competes with the TUI's own stdin reader for the same terminal, so
keystrokes get swallowed and the TUI appears frozen (up to the 300s OAuth
timeout). Reported symptom: "MCP OAuth: authorization required / Open this URL
... the tui is freezing, not respond to typing."
Add a thread-local `suppress_interactive_oauth()` context manager in
tools/mcp_oauth.py; `_is_interactive()` returns False while it's active, so the
stdin paste-thread and prompt are never created. Background discovery
(hermes_cli/mcp_startup.py, tui_gateway/entry.py) now runs discovery inside
that context, so OAuth-requiring servers soft-skip (raise
OAuthNonInteractiveError, already handled) instead of stealing the TUI's stdin.
A real `hermes mcp login` on the main thread is unaffected (thread-local).
Salvaged from #35945 by @zapabob (authorship preserved via cherry-pick;
resolved a conflict against main's new mcp_discovery_timeout / wait_for_mcp_
discovery refactor, keeping both). Verified E2E: with suppression the paste
prompt is NOT printed and no stdin thread spawns (raises OAuthNonInteractive
soft-skip); without it the prompt shows (the freeze). Mutation-verified
(removing the suppress check in _is_interactive fails the regression test).
76 tests pass, ruff clean.
Closes#35927.
SELF-REVIEW FIX: the original #35945 used threading.local(), which does NOT
propagate to the dedicated mcp-event-loop thread where OAuth actually runs
(discover_mcp_tools dispatches the connect via run_coroutine_threadsafe), so
the suppression was a NO-OP in production (the tests passed only by stubbing
out the cross-thread dispatch). Converted to a contextvars.ContextVar, which
asyncio copies onto the scheduled coroutine — empirically verified suppression
now holds on the mcp-event-loop thread through the real _run_on_mcp_loop path.
Added a cross-thread regression test (fails on threading.local, passes on the
ContextVar) so the no-op can't regress.
The OpenAI SDK exposes client.base_url as an httpx.URL object, not str.
The isinstance(live_raw, str) guard made this branch dead code in
production. Use _normalized_runtime_url (which coerces via str()) so
the fallback actually fires.
When parent_agent.base_url still carries a stale OpenRouter URL but the
live OpenAI client already points at local Ollama, subagents were routing
API calls to OpenRouter and failing with HTTP 401. Prefer _client_kwargs
and the mounted client base_url when they disagree with the surface field.
preexec_fn=os.setsid runs Python code in the forked child before exec,
which is unsafe in multi-threaded processes (CPython docs). When the
Desktop gateway loads native libraries (onnxruntime, BLAS, provider SDKs)
with active thread pools, the fork can SIGSEGV before the child execs.
Replace all preexec_fn usage with start_new_session=True, which provides
the same setsid/process-group semantics without running Python in the
fork. This is already the pattern used throughout hermes_cli/gateway.py
and hermes_cli/_subprocess_compat.py.
Fixes#46789
The autonomous self-improvement review fork could still write to a pinned
skill — only external/bundled/hub-installed/protected-builtin skills were
guarded. The curator skips pinned skills from every auto-transition; the
review fork is the same kind of no-user-present actor and must too.
Adds a pin check to _background_review_write_guard so background-origin
edit/patch/delete/write_file/remove_file on a pinned skill are refused.
Stricter than the foreground _pinned_guard (delete-only) by design: with
no user in the loop there is no one to consent to an edit.
Fixes#25839
The dangerous-command approval layer already blocks `hermes gateway
(stop|restart)`, `pkill/killall hermes|gateway`, and `kill ... $(pgrep ...)`.
A reporter noted on #33071 that the agent can still achieve the same
effect by driving launchd directly against the gateway's service label
(`launchctl stop ai.hermes.gateway`, `launchctl kickstart -k
system/ai.hermes.gateway`, etc.) or by substituting `pidof` for `pgrep`
in the kill-expansion form.
This widens the "Gateway lifecycle protection" block in
`tools/approval.py` to cover both vectors:
- `launchctl (stop|kickstart|bootout|unload|kill|disable|remove)`
scoped to commands that target a Hermes label (`hermes`,
`ai.hermes`). Read-only inspection (`launchctl print …`,
`launchctl list`) and operations against unrelated labels remain
unflagged.
- `kill ... $(pidof …)` and the backtick form, alongside the existing
`pgrep` expansion. `pidof` is the BSD/Linux equivalent and is
equally opaque to the `(pkill|killall) … hermes` name pattern.
Intentionally left out of scope: plain `kill -TERM <numeric_pid>` with
a PID looked up out-of-band. Catching that would require runtime PID
state and would break the existing
`TestPgrepKillExpansion::test_safe_kill_pid_not_flagged` contract,
which guarantees that a plain literal-PID `kill 12345` stays safe.
Natural-language skill search returned a short, arbitrary list and never
surfaced NVIDIA (or OpenAI/Anthropic/HuggingFace) skills. Two causes:
1. The runtime index collapses every GitHub tap into source="github", so
there was no way to find or filter by provider at the CLI — the per-tap
identity only existed in the docs-site catalog.
2. HermesIndexSource.search matched only name/description/tags (not the
identifier or provider) and broke at the first `limit` hits in raw index
order, burying the most relevant skills. `search` also defaulted to
--limit 10 against an 86k-entry catalog.
Changes:
- GitHubSource stamps a per-tap provider label (extra.provider) on each
skill via github_provider_for(); source stays "github" so dedup/floor/
index-skip logic is untouched. Flows into the built index.
- HermesIndexSource.search now matches identifier + provider too, and
collect-then-ranks (exact > prefix > whole-word > substring) instead of
break-at-limit.
- --source nvidia|openai|anthropic|huggingface|voltagent|gstack|minimax
provider filters for browse/search (narrows merged results by provider).
- search --limit default 10 -> 25; table Source column shows the provider
label for github skills.
Tested: 181 unit tests pass; E2E against the live runtime index confirms
'nvidia'/'cuda' searches now surface NVIDIA-provider skills and
--source nvidia narrows to exactly the NVIDIA catalog.
On macOS, terminal(background=true) silently failed: the process returned a
session_id and exit_code=0 but the command never ran (empty stdout, no side
effects). Root cause is two interacting issues:
1. _find_shell was aliased to _find_bash, which prefers `shutil.which("bash")`
→ /bin/bash (GNU bash 3.2, still shipped on macOS) over $SHELL (/bin/zsh).
2. process_registry.spawn_local runs [shell, "-lic", "set +m; <cmd>"] with
stdin=/dev/null. bash 3.2 as a login shell sources ~/.bash_profile, which on
many macOS setups contains `exec /bin/zsh -l`; that exec replaces bash but
drops the -c argument, so the command is swallowed (exit 0, no output).
Decouple _find_shell from _find_bash: _find_shell now prefers the user's
configured $SHELL on POSIX (the shell they actually log in with), falling back
to _find_bash when $SHELL is unset/missing. _find_bash is unchanged, so callers
that genuinely need bash (e.g. the _run_bash login-shell snapshot) keep bash
semantics. zsh handles -lic correctly even with redirected stdin.
Salvaged from #42219 by @liuhao1024 (authorship preserved via cherry-pick).
On top of the original (8 unit tests covering $SHELL-set/unset/missing/empty,
Windows-ignores-$SHELL, _find_bash-unchanged), added an E2E regression test
that reproduces the real bash-3.2 login-shell swallow (exit 0 / no file) and
asserts the shell _find_shell selects actually executes a -lic background
command. Mutation-verified: reverting _find_shell to the bash alias fails the
$SHELL-preference test. Bug reproduced directly: /bin/bash 3.2 -lic with a
.bash_profile->exec-zsh creates no file; zsh -lic does.
Closes#42203. Supersedes #42290.
Rebuilds the iron-proxy egress feature cleanly onto current main. The
original feat/iron-proxy branch had diverged from main with an
unmergeable history (no usable merge-base after main history motion),
so the feature's content diff was re-applied onto a fresh main cut and
the three config/docs conflicts (commands.py status/egress, config.py
proxy vs computer_use, slash-commands.md) resolved keeping main's
content plus the egress additions.
Optional, off-by-default TLS-intercepting egress proxy for remote
terminal sandboxes. Sandboxes hold opaque proxy tokens; iron-proxy
swaps them for real provider API keys at the network boundary.
Includes the full review-cycle hardening:
- P0/P1/P2 rounds (GodsBoy, stephenschoettler, arshkumarsingh,
annguyenNous, maxpetrusenko, sxuff findings)
- v0.39 schema realignment + Docker bridge-bind/listener-role fixes
- Docker UX/enforcement hardening
Salvaged security fixes folded in with credit:
- Three P0 gaps (version-probe env scrub, Bitwarden ImportError
fail-closed, container-reuse egress-boundary) + Docker v29.5.3
empty-label edge — kuangmi-bit (#48073)
- P1/P2 (fail-closed replace.require:true, NODE_OPTIONS CA-flag
conflict, GPG checksum verify, threat-model wording) — Bartok9 (#48076)
Co-authored-by: kuangmi-bit <kuangmi@deeparchi.com>
Co-authored-by: Bartok9 <danielrpike9@gmail.com>
The cron runtime tripwire (_scan_cron_prompt) used a 10-char invisible-unicode
set while the install-time scanner (threat_patterns.INVISIBLE_CHARS) flags 17.
The cron-local set was missing U+2062-U+2064 (invisible math operators) and
U+2066-U+2069 (directional isolates), so a directive obfuscated with one of
those codepoints (e.g. "ig<U+2063>nore all previous instructions") slipped past
the runtime cron gate while being caught at install time.
Import the canonical set so the cron tripwire and install scanner can't drift
apart again. Emoji-ZWJ protection (_zwj_has_emoji_neighbour) is unchanged.
Fixes#35075
Co-authored-by: rlaope <piyrw9754@gmail.com>
The known_c2_framework threat pattern included 'praxis' in its
alternation alongside genuine offensive-security tool brands (Cobalt
Strike, Sliver, Havoc, Mythic, Metasploit, Brainworm). Unlike those
distinctive brand names, 'praxis' is a common English word (Greek for
practice/action) and a legitimate agent name, so any context file that
mentioned an agent named Praxis matched at 'context' scope and the whole
AGENTS.md / SOUL.md was replaced with a [BLOCKED] placeholder before it
reached the system prompt.
Remove 'praxis' from the alternation and add a guard comment: every
token in this list must be a distinctive tool brand, not a common word.
Real C2 brands still fire.
Fixes#36767.
Two complementary recoveries for the recurring "delete three cache files and
re-auth by hand" ritual when an MCP server's dynamically-registered OAuth
client goes dead server-side (IdP redeploy / DB wipe / rebrand):
- Auto-heal (token-endpoint subset): HermesMCPOAuthProvider now sniffs
auth-flow responses and, on a 400/401 `invalid_client` from the discovered
token endpoint, backs up + deletes `<server>.client.json` and `.meta.json`
and clears the in-memory client so the SDK re-runs RFC 7591 dynamic client
registration on the next flow. Conservative by construction: only
dynamically-registered (non config-supplied) clients, only the token
endpoint, only on a word-boundary `invalid_client` match (so RFC 7591's
`invalid_client_metadata` does not trip it); best-effort so a miss never
breaks the live flow. Covers both code-exchange and refresh when the token
endpoint was discovered. Tokens are preserved.
- `hermes mcp reauth [<name>|--all]`: the reporter's primary symptom — the
IdP's in-browser "Redirect URI Mismatch" — produces no HTTP signal (the SDK
only sees a callback timeout), so it cannot be auto-detected. The new
command re-auths one or ALL `auth: oauth` servers, serially: one browser
flow at a time, which also fixes the startup popup storm when several
servers are stale at once. Single-server reauth is factored out of
`mcp login` and shared.
Tests: +14 (poison helper x2; token-endpoint detection x5 incl. wrong-endpoint,
success-response, pre-registered, and invalid_client_metadata negative guards;
a bridge integration test driving the real async_auth_flow generator to prove
the detection hook preserves the bidirectional asend() forwarding contract;
reauth CLI x6). Verified against the pinned mcp==1.26.0: scripts/run_tests.sh
122/122 green for the touched suites; check-windows-footguns.py and ruff clean.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Force redact_sensitive_text(force=True) on the browser_type text arg so
recognized credentials (API keys, tokens, JWTs) are masked in tool
progress, previews, callbacks, and return payloads even when the global
security.redact_secrets opt-out is set — a typed credential reaching chat
history is a security boundary, not log hygiene. Normal typed text matches
no pattern and stays fully readable for debuggability.
Tests assert the API-key-shaped secret is masked across every surface and
that normal text passes through unchanged.
* feat(kanban): typed block reasons + unblock-loop breaker
Stops the kanban blocked-task loop: a worker blocks a task, a cron
unblocks it, the worker re-blocks for the same reason, repeat forever.
block_task now takes a typed kind and a persistent block_recurrences
counter on the tasks table:
- kind=dependency routes to todo (parent-gated, auto-resumed), never
the human 'blocked' bucket a cron would keep unblocking.
- needs_input/capability/transient/untyped land in blocked; each
same-cause re-block after an unblock increments block_recurrences,
and at BLOCK_RECURRENCE_LIMIT (default 2) the task routes to triage
for a human instead of blocked.
- unblock_task no longer resets block_recurrences (the amnesia that
let the loop run unbounded); complete_task clears it on success.
Wired through the worker kanban_block tool (new kind arg) and the
hermes kanban block --kind CLI flag, both reporting where the task
actually landed. Docs + 11 new tests; 536 existing kanban tests green.
* test(kanban): make second-block notify test use a distinct block cause
test_notifier_second_blocked_delivers blocked the same task twice with
the same (untyped) reason, which now trips the new unblock-loop breaker
and routes the second block to triage instead of blocked — so only one
'blocked' notification fired. The test's actual intent is that TWO
distinct block cycles each notify; give the two cycles different kinds
(needs_input then capability) so they're genuinely separate blocks. The
same-cause loop→triage path is covered by test_kanban_block_kinds.py.
The detector folds absolute home / Hermes-home prefixes into their canonical
~/ and ~/.hermes/ forms so static patterns catch /home/alice/.bashrc the same
way they catch ~/.bashrc (abd69b81). On native Windows this fold never fired,
so terminal commands writing to shell startup files, ~/.ssh/authorized_keys,
or ~/.hermes/config.yaml / .env returned "safe" and skipped the approval
prompt — and config.yaml carries the approval policy itself.
Two compounding causes:
1. The fold ran after the backslash-escape strip (r\m -> rm), which dissolves
the backslash separators in a Windows path (C:\Users\alice\.bashrc ->
C:Usersalice...) before the fold could match. It now runs before the strip.
2. The fold only recognized POSIX absolute paths and only the home prefix,
leaving multi-segment backslash suffixes (\.ssh\authorized_keys) to be
mangled by the strip.
Consolidated into _home_prefix_fold_regex / _fold_home_prefixes: match a home
prefix with either separator, capture the rest of the path token, and
normalize its separators to / so multi-segment patterns match. The
degenerate-path guard generalizes count("/") >= 2 to "at least two components
below the root" (also rejecting a bare drive root C:\). HOME is consulted
directly because Windows' expanduser ignores it; the more specific Hermes home
is folded first, longest candidate first, so neither fold clobbers the other.
POSIX behavior unchanged; the r\m -> rm anti-obfuscation strip still runs.
Adds TestWindowsAbsolutePathFolding, which monkeypatches a Windows-style
HOME/HERMES_HOME so the behavior is also exercised on the CI runner.
* feat(moa): expose MoA presets as selectable virtual models
Reconstructed onto current main (PR #46081's base had diverged with no common
ancestor, marking the PR dirty so CI never dispatched). MoA is now a virtual
provider: each named preset is a selectable model under provider 'moa', and the
preset's aggregator is the acting model that answers and calls tools.
Reference models fan out in parallel via a bounded ThreadPoolExecutor (the same
batch pattern delegate_task uses) — all references dispatched at once, collected
when every one finishes, then handed to the aggregator. Output order is
preserved, failures and the MoA-recursion guard stay isolated per reference.
- Removed the old mixture_of_agents model tool and moa toolset.
- Added moa as a virtual provider in the provider/model inventory.
- /moa is shortcut behavior over model selection (default preset / named preset
/ one-shot prompt).
- Dashboard + Desktop manage named presets; presets appear in model pickers.
- Parallel reference fan-out in agent/moa_loop.py with regression test.
* fix(moa): thread moa_config through _run_agent to _run_agent_inner
The reconstructed gateway MoA wiring declared moa_config on _run_agent (the
profile-scoping wrapper) and used it inside _run_agent_inner, but the wrapper
never forwarded it — _run_agent_inner had no such parameter, so the runtime hit
NameError: name 'moa_config' is not defined on the compression-failure session
sync path. Add moa_config to _run_agent_inner's signature and forward it from
both wrapper call sites (multiplex and non-multiplex). Caught by
tests/gateway/test_compression_failure_session_sync.py on CI shard test(4).
* fix(moa): classify moa as a virtual provider in the catalog
The moa virtual provider has no PROVIDER_REGISTRY/ProviderProfile entry, so
provider_catalog() fell through to the default auth_type="api_key" with no
env vars — tripping two catalog invariants:
- test_provider_catalog: api_key providers must expose a credential env var
- test_provider_parity: every hermes-model provider must be desktop-configurable
moa already declares auth_type="virtual" in HERMES_OVERLAYS; consult that
overlay as an auth_type fallback so the catalog reports moa as virtual (no real
credential, no network endpoint). Exempt virtual providers from the desktop
parity union check the same way 'custom' is exempt — derived from the catalog,
not a hardcoded slug, so future virtual providers are covered too.
The trailing-whitespace expansion in _map_normalized_positions
unconditionally consumed whitespace after the matched region — including
the word-boundary space that separates the match from the next token.
This caused silent file corruption when the fuzzy matcher fell back to
the whitespace_normalized strategy.
Guard the expansion on the normalized match actually ending with
whitespace (i.e. the original had a run of spaces that were collapsed).
When the match ends with a non-space character, the first whitespace in
the original is a boundary and must not be consumed.
Fixes#52491
todo_tool crashed with `AttributeError: 'str' object has no attribute 'get'`
when the LLM emitted the `todos` param as a JSON-encoded string instead of an
array, or as a list containing non-dict items (observed intermittently on
Claude 4.5/4.6/4.7, and after a prior tool-call rejection where the model
"self-corrects" by wrapping the list in json.dumps).
Three additive guards, no behavior change for well-formed input:
- todo_tool(): if `todos` is a str, json.loads it; reject unparseable strings
and non-list values with a clear tool_error instead of crashing downstream.
- _validate(): non-dict items return a {id:"?", content:"(invalid item)"}
placeholder rather than calling .get() on a str/int/None.
- _dedupe_by_id(): non-dict items get a synthetic key so _validate handles them.
Salvaged from #14785 by @Tranquil-Flow (authorship preserved via cherry-pick).
Comprehensive tests: JSON-string coercion (parse / unparseable / non-list /
non-string), non-dict list items (str/None/int/mixed), and a well-formed-
unchanged regression class — both guards mutation-verified to fail without them.
Closes#14185. Supersedes #14187, #22505, #14350 (same fix, less/no test
coverage) and #16952 (bundled unrelated scope-creep).
During stdio MCP server startup, _run_stdio (an async method) called the
synchronous check_package_for_malware() inline. That makes a blocking
urllib HTTPS POST to api.osv.dev whose own timeout doesn't reliably cover a
stalled SSL handshake, so an intermittent network issue froze the entire
asyncio event loop for up to ~120s — blowing past the TUI/gateway's 15s
startup budget and showing "gateway startup timeout".
Run the check via asyncio.to_thread (off the loop) AND bound it with
asyncio.wait_for(timeout=_OSV_MALWARE_CHECK_TIMEOUT_S=12s). The malware check
is fail-open, so on timeout we log and proceed rather than blocking startup.
Salvaged from #29190 by @qdaszx (re-applied on current main — the call site
moved since the PR was opened), combining the to_thread approach also proposed
in #29192 by @ygd58. Two load-bearing tests: event-loop-not-blocked-during-
check and timeout-fails-open — both mutation-verified to fail against the old
inline blocking call.
Closes#29184.
Co-authored-by: ygd58 <buraysandro9@gmail.com>
Replace _count_real_sudo_invocations (which called
_rewrite_real_sudo_invocations and discarded the rewritten string) with
a lightweight token scan that reuses the same tokeniser but skips string
building. Remove the agent-facing tip about nested sudo in heredocs —
the cache-cleared warning is enough.
Pipe one password line per sudo invocation in compound commands so a correct
password is not rejected on the second `sudo` in `sudo a && sudo b`. Drop the
session cache when sudo returns Authentication failed, surface sudo_auth_failed
in the tool result, and add hints for interactive sessions.
Continuable cron jobs (attach_to_session / cron.mirror_delivery, default
OFF) now prefer a dedicated thread on thread-capable platforms, falling
back to origin-DM mirroring where threads don't exist.
- Thread-capable (Telegram topics, Discord/Slack threads): open a fresh
thread for the job via the shipped adapter.create_handoff_thread,
route the brief into it, and seed the thread-keyed session so the
user's in-thread reply continues with full context. This is the
'continuable cron opens its own thread' interface.
- DM-only (WhatsApp/Signal/SMS): create_handoff_thread returns None ->
fall back to mirroring into the origin DM session (existing behaviour).
Reuses existing infrastructure end-to-end — no new adapter surface, no
provider-chain signature change:
- adapter.create_handoff_thread (already implemented per-platform,
returns None on unsupported platforms = the fallback signal)
- the live SessionStore via adapter._session_store (already set on every
adapter), reached without threading a new param through the frozen
CronScheduler.start() contract
- gateway.mirror.mirror_to_session for the seed/append
- existing per-target delivery routing carries the new thread_id for free
Mirrors GatewayRunner._process_handoff's open-thread-or-fallback +
seed pattern, standalone for the cron delivery path. thread_seeded
guards against a double-mirror after seeding. Scoped to the origin
target only; fan-out/broadcast targets are never threaded or mirrored.
Config docs updated (cron.mirror_delivery) + cronjob tool
attach_to_session description reframed around continuable/thread-preferred.
Tests: +5 (thread id returned on thread platform; None on DM platform;
None without capability/loop; seed creates thread session + mirrors;
seed no-op on empty). 22/22 in TestCronDeliveryMirror; 532 cron tests
pass (4 failures pre-existing: croniter-not-installed + TZ).
Multi-participant parity with interactive send_message, which passes
HERMES_SESSION_USER_ID to gateway.mirror.mirror_to_session so the mirror
lands in the exact participant's session.
- cronjob_tools._origin_from_env now captures user_id from the session
context at job-create time (alongside platform/chat_id/thread_id).
- _maybe_mirror_cron_delivery forwards user_id to mirror_to_session.
- _deliver_result threads origin.user_id through for the origin target.
Effect: in a per-user-isolated group chat (group_sessions_per_user=True,
the default), the mirror resolves to the member who scheduled the job
instead of conservatively no-op'ing on ambiguous candidates. DMs and
shared group/thread sessions are unaffected (single candidate). Default
still OFF.
Tests: helper forwards user_id; E2E _deliver_result forwards origin
user_id. 17/17 in TestCronDeliveryMirror; 527 cron tests pass (4 failures
pre-existing: croniter-not-installed + TZ, identical on baseline).
Adds an opt-in path so a cron job's delivered output is also appended to
the TARGET chat's gateway session transcript (as an assistant turn), so a
user reply to a recurring delivery (daily brief, reminder) is answered with
the delivery in context instead of 'what is that?' amnesia.
- Reuses the shipped gateway.mirror.mirror_to_session — the same primitive
interactive send_message mirroring already uses. No messaging-toolset
change (cron still can't call send_message; this rides delivery).
- Gated: per-job attach_to_session overrides global cron.mirror_delivery
(config.yaml). Default OFF — historical isolation preserved byte-for-byte.
- Mirrors the CLEAN agent output, not the cron header/footer wrapper.
- Alternation/cache-safe: append lands at a turn boundary, never mid-loop,
never mutates the cached system prompt. Cold-start (no target session)
is a silent no-op; mirror errors never fail a successful delivery.
- Surfaced on the cronjob tool (attach_to_session) + config schema.
Driven by enterprise cron-as-control-plane use case. 10 new tests; full
cron + cronjob-tool suites pass (600).
terminal_tool() resolves a per-task cwd override that WINS over config["cwd"]:
cwd = overrides.get("cwd") or config["cwd"]
config["cwd"] is sanitized for container backends in _get_env_config() (host
prefixes /Users//home//C:\\/C:/ and relative paths are replaced with the
backend default /root). But the override was applied RAW — it was never run
through that guard. The gateway/TUI registers the host launch dir as a cwd
override for workspace tracking (tui_gateway/server.py _register_session_cwd
-> _terminal_task_cwd -> _session_cwd -> os.getcwd()), so on a container
backend a host path leaked straight to `docker run -w <host-path>`:
- Windows desktop: -w C:\Users\<user> -> container fails to start (exit 125)
- POSIX: -w /home/<user> -> same
The ACP adapter translates its override cwd (acp_adapter/session.py
_translate_acp_cwd), but the gateway path did neither translation nor
sanitization, so the override bypassed the one guard that would have caught it.
Fix: extract the host/relative-path predicate into a shared
_is_unusable_container_cwd() helper (so the existing _get_env_config()
sanitizer and the new guard can't drift), and re-apply it to the *resolved*
cwd at the override-resolution site. Valid in-container override paths
(RL/benchmark sandboxes that set cwd to /workspace, /root, ...) are absolute
non-host paths and pass through untouched.
Tests: unit-pin the predicate (Windows backslash/forwardslash, POSIX home,
macOS /Users, relative, valid container paths) AND an E2E call-site pin that
drives terminal_tool() with a host-path override registered and asserts the
cwd reaching _create_environment is sanitized. Mutation-verified: reverting
the call-site guard makes the two host-path E2E tests fail (showing the raw
host path leaking) while the valid-/workspace-override test stays green.
The gateway-side BEHAVIOUR layer that consumes the relay scale-to-zero
primitives (gateway-gateway Phase 5): the gateway decides it is idle and
drives the relay transport dormant so the platform (Fly autostop:"suspend")
can suspend the now-traffic-idle machine, which wakes on the connector's
wakeUrl poke (decisions.md Q3=C', D1-D13).
- gateway/scale_to_zero.py: pure helpers — scale_to_zero_enabled (the NAS
Labs HERMES_SCALE_TO_ZERO stamp, D11/Q8=A), parse_idle_timeout_seconds
(config.yaml gateway.scale_to_zero.idle_timeout_minutes, D2),
messaging_is_relay_only_or_absent (F6/D1), should_arm (D1/D11/§3.4(1)),
is_idle (D2/D3/F7).
- gateway/run.py: _last_inbound_at clock stamped on user inbound in
_handle_message (F13); the arm-gate + idle predicate + the
_scale_to_zero_watcher dormant sequence (mark draining -> adapter
go_dormant() -> cooldown), started only when armed. Deliberately NOT the
stop path and NOT mark_resume_pending (F12/D13).
- tools/process_registry.py: has_any_active() for the bg-work guard (D3/F7).
- hermes_cli/config.py: gateway.scale_to_zero.idle_timeout_minutes default 5.
Tests: 38 pure-logic + 6 watcher (incl. bg-work regression guard proven RED).
Full relay + scale-to-zero suites: 184 passed. The 20 unrelated failures in
the broader run are PRE-EXISTING on origin/main (custom-provider/tools tests),
confirmed via a pristine baseline worktree.
The published Docker image seals the agent venv (root-owned, read-only
/opt/hermes) and sets HERMES_DISABLE_LAZY_INSTALLS=1 so a runtime install
can't mutate and brick the core. But opt-in backends (Firecrawl web search,
Exa, Feishu, ...) deliberately keep their SDKs in tools/lazy_deps.py and out
of [all] (pyproject policy 2026-05-12: one quarantined release must not break
every install). The two policies collided: the SDK isn't baked in AND can't
lazy-install, so the default Firecrawl web_search/web_extract fail out of the
box in Docker (#51136), as do Exa (#49445) and Feishu (#50205).
Fix the whole class instead of baking in one backend: when
HERMES_LAZY_INSTALL_TARGET is set, lazy installs are redirected to a writable
dir on the durable /opt/data volume via `pip/uv install --target`, and that
dir is APPENDED to the end of sys.path. Because the core venv always wins
name collisions, a package installed this way can only ADD new modules — it
can never shadow, downgrade, or break a module the core ships. The worst a
bad/incompatible backend package can do is fail to import and report itself
unavailable; the agent core stays healthy. That structural guarantee is what
made it safe to seal the venv, and it is preserved here even with installs
re-enabled.
- tools/lazy_deps.py: durable-target mode — `--target` install + core-pinned
`--constraint` file (shared deps resolve to core's versions, conflicts fail
loudly at install time), append-only sys.path activation, ABI/Python-version
stamp that wipes the store if an image rebuild bumps the interpreter, and a
reworked gate so HERMES_DISABLE_LAZY_INSTALLS=1 redirects (rather than hard-
blocks) when a target is set. security.allow_lazy_installs=false still
disables installs in every mode.
- hermes_bootstrap.py: activate the durable target on sys.path at first import
(before any backend imports its SDK) so packages installed on a previous run
are importable on this run.
- Dockerfile: set HERMES_LAZY_INSTALL_TARGET=/opt/data/lazy-packages.
- docker/stage2-hook.sh: seed + chown the dir on the data volume.
- tests: real-install E2E proving installs land in the target, import cleanly,
don't leak into the sealed venv, and that a core package is never shadowed;
ABI-stamp wipe/preserve; gate matrix; Dockerfile/stage2 contract test.
Fixes#51136
When tempfile.mkdtemp() raises OSError (e.g. disk full), the exception
propagated past the try/finally block, so _mark_install_failed() was
never called. The 24h backoff marker never engaged, causing unbounded
retry on every command -- each attempt leaked a tirith-install-* temp
directory, eventually filling /tmp completely.
Fix: wrap mkdtemp in its own try/except OSError, returning
(None, "no_space") so the caller's normal failure path (including
_mark_install_failed) executes.
Salvaged from #51831 by @liuhao1024.
Closes#51826
_strip_blocked_tools used a hardcoded set missing 'cronjob'. Children
on gateway platforms could inherit the cronjob toolset, scheduling
persistent jobs that outlive the delegation despite DELEGATE_BLOCKED_TOOLS.
Fix: derive the strip set from DELEGATE_BLOCKED_TOOLS at runtime so the
two lists can never drift. Add 'cronjob' to DELEGATE_BLOCKED_TOOLS for
documentation consistency. Two regression tests lock the invariant.
Salvaged from #43687 by @riyas22. Adapted test to current main (no
'messaging' toolset exists -- send_message is intentionally not
registered as an agent tool).
Closes#43466
The drift guard (introduced for #26045) correctly protects replace/remove
from clobbering un-roundtrippable content, but it also fires on the add
path. Since add only appends and never overwrites, the guard is
unnecessary and causes false positives when prior add() calls in the same
session shift the byte count of the on-disk file.
Add skip_drift parameter to _reload_target() and pass True from add().
Replace/remove continue to use the drift guard unchanged.
Salvaged from #42880 by @liuhao1024.
Closes#42874
After `hermes update`, a globally-installed agent-browser's npm postinstall
(fixUnixSymlink) re-points the global symlink (e.g. /opt/homebrew/bin/agent-browser)
at our local node_modules binary. The next update wipes node_modules, leaving a
dangling symlink that `which` still reports but exec fails on with exit 127 —
silently breaking every browser tool (#48521).
Root cause is trust-on-presence: shutil.which/Path.exists accept a name that
resolves but won't run. Add hermes_constants.agent_browser_runnable() (resolves
the path + runs --version) and gate all four resolution sites on it:
_find_agent_browser now skips a dead candidate and falls through to the next
working one (extended PATH -> local .bin -> npx), self-healing the dangling link.
dep_ensure/doctor/nous_subscription validate too; doctor warns on a broken link.
Closes#48521.
deliver=origin (or omitted) from a TUI or classic-CLI session produces a
job with origin=null, because those sessions never populate the
HERMES_SESSION_PLATFORM/CHAT_ID context vars that _origin_from_env reads.
The scheduler then resolves no delivery target and skips delivery — the
job runs and saves output to last_output, but nothing reaches the user
and they only find out by polling cronjob(action='list') (#51568).
This is by design (local sessions have no live-delivery channel), so the
fix surfaces it instead of silently dropping the intent:
- cronjob create now appends an informational notice to its result when
a created job resolves to zero delivery targets and the user did not
explicitly ask for deliver='local'. The check uses the scheduler's own
_resolve_delivery_targets so it accounts for origin, home channels,
'all', and explicit platform targets — no false positives.
- PLATFORM_HINTS gains a 'tui' entry (the TUI had none) and the 'cli'
hint now states that cron jobs from these sessions are local-only and
that deliver must target a gateway-connected platform to notify the
user. This stops the agent promising a delivery that never happens.
No scheduler/delivery behavior change; no new env var; cron isolation
invariant untouched.
/reload-mcp -> shutdown_mcp_servers -> _kill_orphaned_mcp_children(include_active=True)
-> _send_signal -> killpg(pgid, SIGTERM). When a tracked MCP stdio child shares
the gateway's OWN process group, killpg delivers SIGTERM to the gateway itself,
firing its SIGTERM handler -> os._exit(0): /reload-mcp crashes the gateway.
Pre-compute the gateway's own pgid (os.getpgrp(), None on Windows/restricted)
and, in _send_signal, skip killpg when pgid == own pgid, falling through to the
per-pid os.kill path so the child is still reaped without self-signaling.
Adds a regression test (folded in) that pins the guard: with a tracked pgid
equal to the gateway's own pgid, killpg is never called for that pgid and the
per-pid kill fallback is used. Mutation-checked.
Co-authored-by: kshitijk4poor <82637225+kshitijk4poor@users.noreply.github.com>
* Revert "fix(cron): scope job execution to its owning profile (#32091 follow-up) (#50993)"
This reverts commit 660e36f097.
* Revert "fix(cron): anchor cron storage at the default root home (not the active profile)"
This reverts commit a5c09fd176.
The subprocess-stdin guard (TUI gateway fd-inheritance protection) flagged
the `permissions grant` call. None of the cua-driver probes/grant read
stdin, so DEVNULL is correct; apply it to the shared `_run` helper and the
grant call.
The card was macOS-only. cua-driver also runs on Windows and Linux, so
fold `cua-driver doctor` (cross-platform binary/health probes) into a
single OS-aware `ready` signal:
- macOS: ready == both TCC grants; keeps the permission rows + grant flow.
- Windows/Linux: no TCC toggles, so ready == driver health, with a
per-OS note (SmartScreen/UIAccess on Windows; X11/XWayland on Linux).
`computer_use_status()` replaces the macOS-only `permissions_status()` and
surfaces `platform`, `ready`, `can_grant`, and the doctor `checks` (non-ok
ones render as warnings). CLI `permissions status`, the REST endpoint, and
the desktop card all key off the one payload. Grant stays macOS-only (400
elsewhere — nothing to grant).
Vision mode called a `screenshot` MCP tool that cua-driver dropped in
0.5.x (full-window PNG capture was folded into `get_window_state`). The
driver replied "Unknown tool: screenshot", so `images` came back empty,
`png_b64` stayed None, and capture returned a 0x0 result with no image
on every call. `som`/`ax` were unaffected because they already use
`get_window_state`, which masked the regression.
Route vision by capability:
- driver advertises `screenshot` (older builds) -> use it (no AX walk)
- otherwise -> call `get_window_state` but discard the AX tree/elements,
returning only the PNG so vision stays free of element noise
- capabilities not yet discovered -> try `screenshot`, fall back to
`get_window_state` on an empty image, so the path self-heals
Add `_image_from_tool_result` to pull the PNG from either an MCP image
content-part or `structuredContent.screenshot_png_b64`, and use it on
the som path too so the image won't silently drop on driver builds that
deliver it via structuredContent instead of a content part.
Verified live (vision: 1568x954, 0 elements; som: image + 527 elements)
and with unit coverage of all four routing cases.
Computer Use already worked through the desktop backend (the cua-driver
toolset enables + installs via Settings -> Skills & Tools), but there was
no in-app way to see or grant the two macOS permissions it needs, so "give
a model my Mac" was tribal knowledge.
The grants attach to cua-driver's OWN TCC identity (com.trycua.driver /
the installed CuaDriver.app), not Hermes -- so no app entitlement is
involved. cua-driver 0.5+ exposes `permissions status/grant`, which we wrap:
- tools/computer_use/permissions.py: thin client over the two subcommands
- hermes computer-use permissions {status,grant}: CLI parity
- GET /api/tools/computer-use/status, POST .../permissions/grant: desktop REST
- ComputerUsePanel: live Accessibility + Screen Recording state with a
Grant button (dialog attributed to CuaDriver), shown in the expanded
Computer Use toolset row. Binary install stays in the existing provider
post-setup runner.
Follow-ups: i18n the card copy; a "Stop driver" control (cua-driver stop)
for the runaway-`serve` case.
The #32091 fix moved every profile's cron jobs into one shared root store,
but never wired the execution-scoping half it recommended: a job still ran
under whichever profile's ticker picked it up, not its owning profile. So a
job created under `hermes -p donna` could execute with the root profile's
.env / config.yaml / credentials.
- jobs.py: create_job auto-captures the active profile (explicit profile=
override available) and stores it on the job; resolve_profile_home() maps a
profile name to its HERMES_HOME; legacy jobs backfill to 'default'.
- scheduler.py: run_job applies the job's profile via a scoped HERMES_HOME
override (env var + in-process ContextVar) before any .env/config/script
load, restored in finally. tick() routes profile-mismatched jobs to the
single-worker sequential pool so the env mutation can't race.
- cronjob tool threads profile through (NOT exposed in the model schema, to
avoid cross-profile privilege escalation); hermes cron add gains --profile.
E2E verified against a temp HERMES_HOME with a real profile dir: a root-profile
ticker runs a profile='donna' job with HERMES_HOME=donna during execution and
restores the ticker env afterward.
File tools (read_file, write_file, patch, list_directory, etc.) used
os.path.expanduser() which reads the gateway process HOME env var.
In Docker/systemd/s6 deployments where the gateway HOME differs from
interactive sessions, tilde expanded to the wrong directory.
Add _expand_tilde() helper that delegates to get_subprocess_home() when
available, falling back to os.path.expanduser(). Replace all 9
expanduser() call sites in file_tools.py with _expand_tilde().
Follow-up to the /memory approve fresh-store fix. Both the CLI fallback and
the messaging-gateway handler built a bare MemoryStore() with the hardcoded
default char limits (2200/1375), ignoring the user's configured
memory.memory_char_limit / user_char_limit. A live agent honors those
overrides (agent/agent_init.py), so an approval applied without a live agent
could accept a write the user's lower cap would reject, or vice versa.
Extract a shared tools.memory_tool.load_on_disk_store() factory that reads
the configured limits (falling back to defaults if config can't load) and
wire both the CLI and gateway handlers to it, closing the gap on both
surfaces and de-duplicating the construction block.
* chore: re-trigger CI (workflows did not dispatch on prior head)
* fix(image/video gen): make schema delivery instruction platform-neutral
The image_generate and video_generate tool schema descriptions hardcoded
a gateway-only delivery instruction ('display it with markdown
 and the gateway will deliver it'). That schema
is sent on every platform, so on CLI it directly contradicted the CLI
platform hint ('Do NOT emit MEDIA:/path tags ... state its absolute path
in plain text'), and on messaging platforms it was also wrong about the
mechanism (local file paths are delivered via MEDIA: tags, not markdown
image syntax — markdown ![]() only works for URLs).
The per-platform file-delivery convention is already owned correctly by
the platform hints in prompt_builder.py. The tool schema now just
describes the result shape (URL or absolute path in the image/video field)
and defers 'how to deliver' to the active platform's guidance.
Provider/model injection already works via _build_dynamic_image_schema()
(the 'Active backend: <provider> · model: <model>' line); no change there.
capture(app='screen'|'desktop') now resolves to the OS shell/desktop
window (Windows Progman/WorkerW desktop or Shell_TrayWnd taskbar, macOS
Finder/Dock) so 'show me my screen' and 'click the taskbar' work.
Previously capture() only matched application windows, and the schema
advertised 'or the whole screen' without any code path delivering it.
cua-driver is window-oriented (no virtual-desktop or per-monitor MCP
tool), so a single image still cannot span multiple monitors — the
schema now states this and the no-desktop-window path returns a clear
message instead of silently grabbing the frontmost app.
cua-driver 0.6.x removed the standalone screenshot MCP tool, so
capture(mode='vision') hit 'Unknown tool: screenshot' and returned a
0x0 image with no PNG while som/ax (which use get_window_state) still
worked. Route vision through get_window_state(capture_mode='vision').
Salvaged from PR #50771; same fix submitted earlier as #39262 by
@Tranquil-Flow.
* feat(computer_use): disable cua-driver telemetry by default, add opt-in
cua-driver ships anonymous PostHog usage telemetry ENABLED by default
upstream (fires cua_driver_install / cua_driver_doctor events to
eu.i.posthog.com). Hermes now disables it for our users unless they
explicitly opt in.
- New config key `computer_use.cua_telemetry` (default false) in
DEFAULT_CONFIG.
- `cua_backend.cua_driver_child_env()` injects
`CUA_DRIVER_RS_TELEMETRY_ENABLED=0` into the child env when telemetry is
disabled (the default); leaves the var untouched on opt-in so the driver
uses its own default. Reads config fail-safe — any error defaults to
telemetry off.
- Routed every cua-driver spawn site through the policy: MCP backend
(StdioServerParameters env), `cua_driver_update_check`, doctor's
health_report Popen, the install.sh/install.ps1 runner, and the
`--version` / status probes.
- Docs: new Telemetry subsection in computer-use.md (EN).
- Tests: tests/computer_use/test_cua_telemetry.py — default disables,
explicit-false disables, opt-in leaves var untouched, config-failure
fails safe, inherited-enabled is overridden off.
Verified live on Linux against the real cua-driver-rs 0.6.0 binary: with
the var=0 the driver reports "telemetry: disabled via
CUA_DRIVER_RS_TELEMETRY_ENABLED" and sends no event; with it unset it logs
"sending event: cua_driver_doctor". 213 computer_use + install tests green.
* fix(dashboard): fold computer_use config category into agent tab
The new computer_use.cua_telemetry key created a single-field dashboard
config category, tripping test_no_single_field_categories (web_server's
invariant that categories with <2 fields must be merged to avoid tab
sprawl). Add computer_use -> agent to _CATEGORY_MERGE, matching the
existing onboarding/telegram single-field folds.
* chore: re-trigger CI (workflows did not dispatch on prior head)
* fix(delegation): emit high-concurrency cost warning once per process
_get_max_concurrent_children() runs on every get_definitions() schema
rebuild (via _build_top_level_description / _build_tasks_param_description),
not just on actual delegate_task calls. With max_concurrent_children>10 the
cost advisory fired on every turn / agent spawn across every session, spamming
the log even when delegate_task was never used. Gate it behind a module-level
_HIGH_CONCURRENCY_WARNED flag so it warns at most once per process.
The runtime gate (check_computer_use_requirements) and the hermes tools
platform_gate both enable linux alongside darwin/win32, but several
docstrings/comments still described Linux as "alpha, gated off until it
flips upstream" — contradicting the code that ships it. Bring the prose in
line with the gate that's actually live:
- tool.py / cua_backend.py module docstrings: Linux is enabled (X11 today,
Wayland via XWayland), not gated off.
- toolsets.py description and hermes tools display name: (macOS/Windows) ->
(macOS/Windows/Linux).
No behavior change — the gate already allowed all three platforms.
Make the computer_use toolset platform-agnostic by driving cua-driver on
macOS, Windows, and Linux. Consumes the 8 cua-driver decoupling surfaces
(capability discovery, structuredContent AX tree, opaque element_token,
click button enum, explicit mimeType, machine-readable manifest,
structured list_windows, structured health_report), each degrading
gracefully on older drivers.
Adds `hermes computer-use doctor` (drives cua-driver health_report with a
per-OS check matrix and an exit 0/1/2 ok/degraded/blocked contract), full
typed wrappers for the previously-uncovered cua-driver tools plus a generic
call_tool escape hatch, per-session agent-cursor lifecycle, platform-aware
system-prompt guidance (host-deterministic, cache-safe), and honors
HERMES_CUA_DRIVER_CMD end-to-end.
Replaces the macOS-only skills/apple/macos-computer-use skill with a
cross-platform skills/computer-use skill, and refreshes the EN + zh-Hans
docs.
Supersedes #44221 (Windows-enablement salvage of #30660).
Co-authored-by: Teknium <127238744+teknium1@users.noreply.github.com>
* feat(goals): add /goal wait <pid> barrier to park the loop on a background process
The /goal loop re-pokes the agent every turn via the post-turn judge. When a
goal is gated on a long-running background process (CI poller, build, test
matrix, deploy) that produces nothing to judge yet, this spins the agent into
'is it done?' busy-work and burns the turn budget.
/goal wait <pid> [reason] parks the loop: while the PID is alive, the judge is
skipped, no turn is consumed, no continuation fires, and /goal status shows a
parked indicator. The barrier auto-clears the moment the process exits (the
agent's notify_on_complete watcher is the natural wake signal), then the next
turn resumes normal judging. /goal unwait clears it manually; pause/resume/clear
drop it; a dead/stale PID can never wedge the loop.
Wired across CLI, gateway, and the mid-run command guard for parity. Barrier
persists in SessionDB.state_meta (survives /resume); GoalState gains
backward-compatible waiting_on_pid/waiting_reason/waiting_since fields. 12 new
tests; docs updated.
* fix(goals): use gateway.status._pid_exists for liveness, not os.kill(pid,0)
The Windows-footguns CI guard flagged os.kill(pid, 0) in _pid_alive — on
Windows that's not a no-op, it routes to CTRL_C_EVENT and hard-kills the
target's console process group (bpo-14484). Delegate to the canonical
footgun-safe gateway.status._pid_exists (psutil + ctypes/POSIX fallback)
instead, with a direct-psutil last resort.
* feat(goals): judge-driven auto-wait — the loop parks itself, no manual /goal wait
Makes the wait barrier automatic. Every turn the judge is shown the agent's
live background processes (pid, command, uptime, output tail from the
process_registry) alongside the goal + response, and can return a new 'wait'
verdict instead of continue:
{"verdict":"wait","wait_on_pid":N} → park until that process exits
{"verdict":"wait","wait_for_seconds":N} → park until the deadline passes
evaluate_after_turn acts on the directive (sets the barrier, parks the loop)
so the agent isn't re-poked into busy-work while CI/builds/deploys run. Adds a
time-based waiting_until barrier alongside the pid barrier; both auto-clear and
can never wedge the loop. Drivers (CLI, gateway, tui_gateway) feed the live
registry in via gather_background_processes(). Manual /goal wait stays as an
override. Judge verdict contract widened to (verdict, reason, parse_failed,
wait_directive); legacy {"done":bool} shape still accepted.
* test(goals): update kanban _fake_judge to the 4-tuple judge contract
CI test(3) caught it: test_kanban_goal_mode's _fake_judge still returned the
3-tuple (verdict, reason, parse_failed), but the kanban loop now unpacks the
4-tuple (+ wait_directive). Update the fake to return None for the directive
and accept the background_processes kwarg.
* feat(goals): trigger-based wait — park on a process's own signal, not just exit
Addresses two gaps in the judge-driven wait: (1) the judge could only express
'wait until PID exits' or 'wait N seconds', so a long-lived watcher/server that
fires a trigger MID-RUN (and may never exit) couldn't be waited on; (2) the
process's own watch_patterns/notify_on_complete trigger was invisible to the judge.
Adds a session-based barrier (waiting_on_session) that releases on the process's
OWN trigger via process_registry.is_session_waiting(): the session exits, OR (if
started with watch_patterns) its pattern matches — even while the process keeps
running. list_sessions() now surfaces session_id + watch_patterns/watch_hit/
notify_on_complete so the judge sees the trigger and is told to prefer
wait_on_session for trigger processes. Judge verdict gains a {wait_on_session}
directive (preferred over pid). Backward-compatible GoalState field; pid + time
barriers unchanged.
Tests: TestSessionTriggerBarrier (release on mid-run pattern match while alive,
release on exit, unknown-session, full park→trigger→resume, parse, validation,
backcompat load). 105 goal-surface + 85 process_registry tests green.
Plugins shelling out to bare `hermes` via the terminal tool hit
`command not found` (exit 127) when the gateway was launched without the
hermes install dir on PATH (systemd, service managers, cron, desktop
launchers) — even though `hermes` works in the user's own interactive
terminal, which sources the shell rc that exports that dir.
The terminal tool's subshell PATH was the agent process PATH plus a
static set of system dirs (_SANE_PATH); it never included wherever the
hermes console-script actually lives (~/.local/bin, the venv bin/Scripts,
pipx, nix). Resolve that dir once (which/argv0/sys.executable) and
prepend-if-missing it so bare `hermes` resolves regardless of launch
method.
Live testing against a real SIGTERM-ignoring process TREE (parent + children,
the agent-browser daemon + renderer shape) revealed psutil.wait_procs's
gone/alive partition mis-handles a parent/child tree: it reaps via
Process.wait() and could mark targets gone/alive inconsistently across the
tree, leaving survivors un-killed (flaky — sometimes the parent lived,
sometimes a child). Replace it with: sleep out the grace window, then
directly re-probe every captured target (_proc_alive, treating zombies as
dead) and SIGKILL any that's still running. Add a multi-child-tree regression
test. 6/6 escalation tests green across repeated runs; the real-tree E2E now
kills the full tree 6/6 runs.
A daemon that ignores or stalls in its SIGTERM handler currently survives the
process-registry reap and leaks until reboot (observed as agent-browser
daemons accumulating to EMFILE on long-running gateways). _terminate_host_pid
now snapshots the tree, SIGTERMs it, waits a bounded grace window
(terminal.daemon_term_grace_seconds, default 2.0s, 0 disables), then SIGKILLs
any survivor. The recycled-PID identity guard still gates the whole path, so
escalation never reaches a stranger; Windows is unchanged (taskkill /F is
already a hard kill).
Config lives in config.yaml (terminal.daemon_term_grace_seconds), NOT an env
var, per the .env-secrets-only policy.
Implements the SIGKILL-escalation idea from @tkwong's #15008, reworked onto the
current _terminate_host_pid tree-kill path (the original predated it) and
config-gated instead of env-var-gated.
Co-authored-by: Benjamin Wong <tkwong@inspiresynergy.com>
The background-process registry signalled host PIDs (recovery adoption,
detached-session kill, tree-kill) using a number captured at spawn, guarded
only by a bare liveness check. Once a session's process exits and is reaped the
kernel recycles that PID onto an unrelated process, so an alive-but-different
PID passed the check and got tree-killed.
Observed in the wild: a recycled background-session PID landed on Firefox's
session leader; a later kill/refresh walked its process tree and SIGTERMed
every tab — Firefox "closing" at irregular intervals with no crash/coredump.
This is the same PID/PGID-recycling class fixed for the MCP orphan reaper in
7bd1f8a2d, but the process_registry subsystem was never guarded — so the bug
persisted.
Fix: record each host process's kernel start time (/proc/<pid>/stat field 22)
at spawn, persist it in the checkpoint, and re-validate it before every signal
via `_host_pid_is_ours`. A PID whose start time no longer matches — or that is
gone — is never signalled:
- recover_from_checkpoint: a recycled PID is not adopted as a session.
- _refresh_detached_session: a recycled detached PID is marked exited.
- kill_process / _terminate_host_pid: refuse to tree-kill a stranger.
Legacy checkpoints and platforms without /proc (no baseline) degrade to the
prior best-effort liveness behaviour, so nothing else changes.
Adds TestPidReuseGuard: real-process tests proving a mismatched start time
refuses termination while a matching one still kills, plus recovery/refresh
recycling paths. 74 registry + 22 MCP-stability tests green.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The kanban-worker and kanban-orchestrator bundled skills existed only to
be force-loaded into dispatcher-spawned workers, gated by
environments:[kanban] so they wouldn't leak into normal CLI listings.
That gating was fragile (the leak that #50443 patched) and the
--skills auto-load was already best-effort — most workers ran without it
because the bundled skill isn't present in profile-scoped skills dirs.
Remove the skills entirely and promote their load-bearing content
(workspace kinds, deliverable artifacts, created-card integrity, profile
discovery) into KANBAN_GUIDANCE, which is already injected into every
kanban worker's system prompt. Net result: every worker reliably gets
the guidance, nothing can leak into a CLI/blank-slate session, and the
gating machinery is gone.
- agent/prompt_builder.py: promote the 4 load-bearing rules into KANBAN_GUIDANCE
- hermes_cli/kanban_db.py: drop --skills kanban-worker auto-injection + _kanban_worker_skill_available probe
- hermes_cli/kanban_swarm.py: drop skills=[kanban-orchestrator] on the root card
- hermes_cli/kanban.py: drop kanban-init skill seeding; fix help text
- delete skills/devops/kanban-{worker,orchestrator}
- docs: delete the two skill pages (EN+zh), fix sidebars/catalog/kanban.md/kanban-worker-lanes.md and the video-orchestrator + codex-lane references
- tests: update spawn-argv expectations; re-bound the guidance-size guard
Supersedes the skill-leak half of #50443 (credit @helix4u for flagging the area).
The browser orphan reaper reads a daemon PID from a `.pid` file in a
world-writable, predictably-named temp dir (`/tmp/agent-browser-h_*`) it
does not write itself, then tree-kills that PID via `_terminate_host_pid`
after only a liveness check. A same-user actor could plant a fake socket
dir whose `.pid` points at an arbitrary victim process, and OS PID reuse
after the real daemon exits could land the recorded PID on an unrelated
process — either way an arbitrary same-user process (and its whole tree)
gets SIGTERMed. Local DoS.
Add `_verify_reapable_browser_daemon()`, gated before the kill: via psutil
(a hard dep, fine cross-platform for the same-user processes the reaper can
signal) require both (1) identity — `agent-browser` in the process
name/cmdline — and (2) binding — the live process references *this* session's
socket dir in its cmdline or `AGENT_BROWSER_SOCKET_DIR`. The binding check is
the real spoof defense: a planted/recycled PID won't embed our exact socket
path. Fail-closed on any ambiguity (unreadable cmdline, no match), leaving the
process and its socket dir untouched for a later sweep.
Builds on @sgaofen's fix in #14394 (cmdline identity check); rewritten to use
psutil instead of `/proc`+`ps` (cross-platform, Windows-covered) and to add
the session-socket-dir binding check for recycled-PID / spoof resistance.
Co-authored-by: sgaofen <135070653+sgaofen@users.noreply.github.com>
ipaddress.ip_address() raises ValueError on IPv6 addresses with scope
IDs (e.g. 'fe80::1%eth0'). Both is_always_blocked_url() and is_safe_url()
silently skipped these via `except ValueError: continue`.
If ALL resolved addresses for a hostname carry scope IDs, every address
is skipped and the URL passes all safety checks — a potential SSRF
bypass vector against link-local or metadata endpoints.
Fix:
- Strip the scope ID (%eth0) before parsing in both functions
- is_safe_url(): fail closed (return False) with a warning log if still
unparseable after stripping
- is_always_blocked_url(): use continue (not return False) to preserve
multi-address scanning, with a warning log
Affected: tools/url_safety.py — is_always_blocked_url(), is_safe_url()
A dangerous-command gateway approval blocks the agent's execution thread
inside _await_gateway_decision() on threading.Event.wait() until the user
responds or the 5-minute approval timeout fires. The poll loop never checked
is_interrupted(), so /stop (which flags the agent's execution thread via
AIAgent.interrupt()) was silently ignored — the session stayed wedged until
timeout, even though /stop reported the session unlocked.
Check is_interrupted() at the top of the poll loop. The wait runs on the
agent's execution thread, the exact thread interrupt() flags, so the check
sees the signal and resolves the pending approval as deny — the agent loop
receives a normal denial and unwinds cleanly. Covers /stop, /new, and the
gateway inactivity-timeout interrupt through the single shared wait loop used
by both the terminal and execute_code guards.
The read_file device guard now walks symlink hops before the file operation
layer, but that hop walk still interpreted relative paths against the Python
process cwd. In sessions where TERMINAL_CWD points at the task workspace, a
relative workspace symlink to a blocked alias such as /dev/../dev/stdin could
therefore miss the intermediate device target before later task-cwd resolution.
Anchor relative device checks to the task base before symlink-hop inspection so
the pre-I/O guard sees the same workspace path that read_file would otherwise
read. Absolute device paths and the existing final realpath fallback remain
unchanged.
Refs #10141
Refs #29158
* fix(api-server): stop silently promising async delivery on stateless HTTP path
terminal(notify_on_complete=True / watch_patterns) and delegate_task(background=True)
silently no-op'd on the API server / WebUI path (#10760): the watcher / detached
child registered, but every API-server route (OpenAI-spec /v1/chat/completions
and /v1/responses, plus the proprietary /v1/runs SSE stream) tears down its
channel when the turn ends, and APIServerAdapter.send() is a no-op stub. A
completion that fires after the response closed had nowhere to go — from the
agent side, indistinguishable from a hang.
There is no spec-compliant surface to wake the agent later on a stateless HTTP
client, so make the no-op honest instead of silent:
- Add a per-adapter capability flag supports_async_delivery (default True;
APIServerAdapter = False), propagated into a HERMES_SESSION_ASYNC_DELIVERY
contextvar via async_delivery_supported(). Toggle on the adapter, not a
hardcoded platform string — a future stateless adapter is correct-by-default.
- terminal: when delivery is unsupported, skip watcher registration, force
notify_on_complete off, and return a notify_unsupported note telling the
agent to process(action='poll').
- delegate_task: when delivery is unsupported, fall back to SYNCHRONOUS
execution (work runs and returns in the same response) with a note, instead
of handing out a handle that never resolves.
CLI (in-process completion_queue) and the real gateway platforms are unchanged.
Fixes#10760
* refactor(api-server): route session binding through a single no-delivery chokepoint
Add APIServerAdapter._bind_api_server_session() and route both agent-entry
paths (_run_agent for /v1/chat/completions + /v1/responses, and the /v1/runs
_run_sync path) through it. The helper hardwires platform="api_server" and
async_delivery=False with no async_delivery parameter to pass, so a future
route added to the API server physically cannot reintroduce the silent
no-op (#10760) by forgetting to mark the channel as non-delivering.
The binding stays request-scoped (cleared per turn), so a session resumed
later on a delivering interface (CLI / gateway platform) re-binds fresh and
is NOT blocked — the no-delivery decision tracks the interface handling the
current turn, never the session.
Follow-up to @de1tydev's poll-read-only fix. Removing the
_completion_consumed.add() from poll() fixes the gateway/tui watcher
suppression (#10156) but reintroduces the CLI duplicate that #8228 fixed:
a notify_on_complete process always enqueues a completion event, and the
CLI idle/post-turn drain would re-inject it as a [SYSTEM: ...] message
even though the agent already saw the exit inline in its poll result.
Add a separate _poll_observed set that poll() populates on an observed
exit. drain_notifications() (CLI only) skips poll-observed sessions; the
gateway/tui watchers keep checking only is_completion_consumed, so a
read-only poll never suppresses their autonomous delivery turn.
- _poll_observed pruned alongside _completion_consumed in _prune_if_needed
- 4 tests: CLI drain dedup after poll, gateway gate untouched, running
poll doesn't mark observed, wait/log still skip CLI drain
Security-hardening fix for the read_file device guard, not a new sandbox
boundary. The guard already rejects direct device paths and upstream now
has a resolved-path pass for workspace symlinks to blocked devices, but
its concrete-path helper still compared the expanded path before
normalization. That leaves residual alias cases where the dangerous path
is visible before final terminal-specific resolution, for example:
1. /dev/../dev/zero and /dev/./urandom should match the blocked-device
list as concrete paths, not only after final realpath;
2. /dev/stdin-style aliases can disappear once realpath follows them
to /proc/self/fd/0 and then to a tty path;
3. a user symlink to /dev/../dev/stdin exposes the dangerous
intermediate target before final resolution, but not necessarily
after it.
Normalize expanded paths before matching and inspect each symlink hop
before falling back to realpath. This preserves the existing /proc fd and
/proc pseudo-file guards while enforcing the intended security invariant:
model-supplied read paths must not reach blocking or infinite device
streams through spelling, normalization, or symlink-hop tricks.
Classification: security hardening / residual bypass fix for the
read_file device blocklist. This is defensive code at the file-tool
boundary, but it fixes a concrete denial-of-service class tracked as
security in #10141 and #29158.
Tests:
- normalized /dev/../dev/zero and /dev/./urandom aliases
- symlink to /dev/../dev/stdin blocked before realpath
- existing symlink-to-device and regular-symlink guards still pass
Fixes#10141Fixes#29158
When terminal.backend is docker/modal/daytona/ssh/singularity, the
terminal runs in a sandboxed container with network isolation, but the
browser still runs on the host. The SSRF guard was skipped because
_is_local_backend() only checked browser.cloud_provider, not the
terminal backend.
Now _is_local_backend() also checks TERMINAL_ENV — when the terminal
is containerized, the browser is treated as non-local and SSRF
protection is enabled.
Fixes#38690
The tool-result persistence budget was a fixed 100K chars/result and 200K
chars/turn regardless of the active model. On a small-context model (e.g. a
65K-token local model switched into mid-session) a single large tool result
(reporter: a 279K-char search result) or a full 200K-char turn (~50K tokens)
could by itself approach or exceed the window, forcing an oversized request
that the provider rejects as "Prompt too long".
- budget_config.budget_for_context_window() scales per-result/per-turn char
caps to a fraction of the model window, clamped to the historical 100K/200K
defaults (large models unchanged) and floored so small models stay usable.
- resolve_threshold() now caps the per-tool registry value at default_result_size
so tools that register a fixed 100K cap (web/terminal/x_search) don't re-inflate
a scaled-down budget. No-op for the default budget (both 100K).
- tool_executor wires the agent's live context_length (recomputed on model
switch) into all four persist/turn-budget call sites.
read_file stays inf-pinned (no persist loop). Verified E2E: a 279K-char result
against a 65K model collapses to a ~1.6K preview; a 200K model is byte-identical
to today.
A server that doesn't implement the optional 'ping' utility answers a
keepalive ping with JSON-RPC method-not-found. _is_method_not_found_error
latches that condition so the probe falls back to list_tools instead of
reconnect-looping.
The substring fallback only matched 'method not found' / '-32601' /
'not found: ping'. Servers that surface method-not-found as the common
'Unknown method: <name>' phrasing without a structural -32601 code (e.g.
agentmemory's MCP server) slipped through, so the fallback never latched
and the keepalive reconnect-looped every cycle.
Add 'unknown method' to the substring fallback so the ping->list_tools
keepalive fallback latches for these servers too.
Fixes#50028.
`cronjob(action='run')` (and `hermes cron run`) only set `next_run_at = now`
and returned success, relying on the scheduler ticker to actually execute the
job on its next tick. When no gateway/ticker is running — a CLI-only setup, or
the Windows case in #41037 — the job never executed: `run` reported success,
but `last_run_at` stayed null forever, no output, no delivery.
A manual `run` should actually run. `_execute_job_now` now:
- **claims the job via `claim_job_for_fire`** — the same at-most-once CAS the
scheduler/external-provider fire path uses. This both advances `next_run_at`
for recurring jobs and blocks a concurrently-running gateway ticker from
double-firing the same job; if the claim is lost, the run is skipped (the
tool reports `execution_skipped`). This closes the double-fire race that a
bare `advance_next_run` left open (a tick whose `get_due_jobs` already
captured the job between trigger and advance would still fire it).
- **delegates firing to `run_one_job`** — the single shared
execute→save→deliver→mark body the ticker and external providers use — so
failure delivery, `[SILENT]` handling, and live-adapter delivery stay
identical across paths and can't drift. (The original salvage re-implemented
this sequence inline and had already dropped failure delivery + `[SILENT]`.)
The tool response carries `executed`, `execution_success`, and either
`execution_error` or `execution_skipped`. The `hermes cron run` CLI message no
longer claims "It will run on the next scheduler tick" — it reports the actual
"Ran now: succeeded/failed" outcome (or the skip).
Salvaged from #41130 by @kyssta-exe (authorship preserved); reworked to reuse
`claim_job_for_fire` + `run_one_job` per review rather than re-implementing the
fire sequence inline. Adds tests for the claim-then-fire path, claim-lost skip,
failure reporting, and exception capture.
Fixes#41037
Co-authored-by: kyssta-exe <kyssta-exe@users.noreply.github.com>
Single-op replace/remove failed with a dead-end 'old_text is required'
error when a structured-output client omitted the optional old_text field
(it can't be schema-required without a top-level if/then combinator that
OpenAI's Codex backend 400s on). The model couldn't recover.
Now a missing old_text returns the current entry inventory plus a retry
instruction (mirroring the batch path's _batch_error), so the model can
reissue the call with old_text set. Also sharpens the old_text schema
description to state it's required for replace/remove.
Fixes#49466, #43412.
Clarify that session_search is secondary context and direct source identifiers must be inspected first when accessible. Add regression coverage for the tool description.
* feat(delegation): single-task delegate_task always runs in the background
The model no longer decides whether a subagent runs in the background — a
single-task delegate_task from the top-level agent is now always dispatched
async, so the parent turn returns immediately and the subagent's result
re-enters the conversation when it finishes.
- run_agent._dispatch_delegate_task (the live model path) forces
background=True for top-level single-task calls; the schema-level
`background` param is ignored.
- A batch (tasks with >1 item) stays synchronous (fan-out can't go async).
- A delegation from an orchestrator subagent (depth > 0) stays synchronous —
it needs its workers' results within its own turn.
- The function-level default is unchanged, so direct Python callers/tests keep
the historical synchronous behavior.
- On async-pool capacity rejection, single-task now falls through to a
synchronous run instead of erroring (the child stays attached for interrupt
propagation; detach happens only on a successful dispatch).
- Schema `background` param marked deprecated/ignored; tool description
updated to state the always-background single-task rule.
* feat(delegation): all delegate_task fan-out runs in the background
Extend the always-background behavior to the full fan-out. A batch is now
dispatched as N independent async subagents (one handle each), instead of
running synchronously. Single task and batch both return immediately; each
subagent's result re-enters the conversation as its own message when it
finishes.
- delegate_task: when background is set, loop over ALL built children and
dispatch each via dispatch_async_delegation; return a combined handle block
(count + per-task delegation_ids). Children the async pool rejects (at
capacity) run synchronously inline and are reported alongside the dispatched
handles, so nothing is silently dropped.
- run_agent._dispatch_delegate_task + registry handler: force background for
any top-level model delegation (single OR batch); orchestrator subagents
(depth > 0) still run synchronously since they need workers' results within
their own turn.
- Removed the v1 'batch async not supported' rejection.
- Tool description updated: BOTH MODES RUN IN THE BACKGROUND.
- Tests updated to assert batch fan-out dispatches each task async (verified
E2E: 3-task batch -> 3 independent completion-queue events).
* fix(delegation): background fan-out joins and returns one consolidated block
Correct the fan-out semantics: a backgrounded batch is dispatched as ONE
async unit (one handle, one async-pool slot), not N independent dispatches.
The unit runs all children in parallel, waits on every one, and emits a
SINGLE completion event carrying the consolidated per-task results. The chat
is never blocked; when all subagents finish, their full summaries re-enter
the conversation together as one message.
- async_delegation.dispatch_async_delegation_batch + _finalize_batch: a batch
occupies one slot; its runner returns the combined {results:[...]} dict and
one event with the full results list is pushed to the completion queue.
- delegate_tool: extract the sync execution+aggregation into
_execute_and_aggregate(); background dispatches it via the batch unit and
returns one handle; on pool-capacity rejection it runs the batch inline.
- process_registry._format_async_delegation: render a consolidated multi-task
block (TASK i/N + per-task summary) when the event carries is_batch/results.
- Tests updated; E2E verified: 3-task batch -> immediate return -> one combined
completion block with all three summaries.
Salvage of PR #41284 onto current main. Relocates the last 9 inline messaging
adapters (+ satellites: telegram_network, feishu_comment/_rules/meeting_invite,
wecom_crypto, wecom_callback) from gateway/platforms/ into self-contained
bundled plugins under plugins/platforms/<x>/, discovered via the platform
registry. Strips the per-platform core touchpoints from gateway/run.py,
gateway/config.py, hermes_cli/gateway.py, hermes_cli/setup.py, and
tools/send_message_tool.py.
Carries forward the migration fixes (explicit enabled:false honored,
get_connected_platforms forces discovery, plugin is_connected via
gateway.get_env_value, logs --component gateway matches plugins.platforms.*,
matrix hidden on Windows).
Additionally ports config keys main added since the PR base: the matrix
plugin's _apply_yaml_config now also covers allowed_users,
ignore_user_patterns, process_notices, and session_scope (the inline
gateway/config.py matrix block gained these in the 1340 commits the PR sat
open; they would otherwise have been silently dropped on deletion).
Route Signal send paths through shared markdown formatting helpers and render markdown bullets consistently as Unicode bullets. Add coverage for Signal formatting and send_message integration.
delegate_task has never exposed a per-call model parameter (removed
intentionally in fb0f579b1). The tool description gave no hint about how
subagent model is actually controlled, so users kept expecting a model
arg and filing it as a dropped/ignored param (e.g. #49332, #23467).
Add one bullet to the dynamically-built tool description stating that
children inherit the parent model + fallback chain, and that pinning all
subagents to a specific model is done via delegation.provider /
delegation.model in config.yaml. No behavior change.
_build_fal_payload and _build_fal_edit_payload assemble the request and then
filter it down to the model's supports / edit_supports whitelist. That filter
also covers prompt (and image_urls for edits), which every FAL endpoint
requires. Today all model configs happen to list those keys, but a single
config that omits one would silently produce a request with no prompt or no
source images — a broken generation with no error.
Always keep the mandatory keys regardless of the whitelist so a missing
whitelist entry can only drop optional knobs, never the prompt or the images.
* docs: clean up three stale comments from the #32848 audit
- tools/memory_tool.py:20 — 'read' action was intentionally removed
but the docstring still listed it. Now matches the schema.
- tools/fuzzy_match.py:9 — unicode_normalized was added but the
chain-count docstring still said '8-strategy'. Now says '9'.
- run_agent.py:1485 — 'See #<TBD>.' placeholder was never filled in.
Replaced with a backfill note.
Fixes#32848 (parts 3, 4, and 12)
* docs(memory): also remove stray memory(action=read) references in lines 144 and 201
The original #32848 audit fix (in 6fd661d6) only addressed line 20
(the action list in the module docstring), but the action was
referenced in two other places:
- tools/memory_tool.py:144 — in a class docstring, claimed
'memory(action=read)' was a way to SEE poisoned entries
- tools/memory_tool.py:201 — in a user-facing warning message,
told the user to 'use memory(action=read) to inspect'
Since the schema on line 683 only allows add/replace/remove, both
references were misleading: the first claimed a way to inspect
poisoned entries that doesn't exist, the second would error out
when the user followed the warning.
This commit removes both references:
- Line 144: '...keep the original text so the user can still SEE
poisoned entries by inspecting the source files directly, and
remove them — silently dropping them would hide the attack
from the user.'
- Line 201: '...use memory(action=remove) to delete the
original. (drop the read-action reference)'
Followup to the previous commit on this branch.
---------
Co-authored-by: KeyArgo <keyargo@argobox.com>