Four fixes to the tool-search deferral layer, split from PR #92693 (the
availability-cache staleness fix ships separately):
1. The parallel batch planner now peels the tool_call bridge wrapper and
decides admission on the underlying tool — supports_parallel_tool_calls
works again when deferral is active. Unparseable wrappers stay
sequential barriers; bridged calls get exactly the admission the same
call gets direct. tool_search/tool_describe lookups batch concurrently.
2. _short_desc no longer truncates listing lines at 'e.g.', hostnames, or
version strings — a sentence terminator must be followed by whitespace.
3. BM25 indexes the source label (e.g. 'linear' for mcp-linear), so
service-name queries reach tools whose own name omits the service; the
dead 'mcp' prefix token is stripped.
4. Substring-fallback docstring corrected (token misses, not zero-IDF).
Salvaged from #92693 by @alt-glitch with authorship preserved.
Sites under active development but tested over the public internet
(Vercel previews, ngrok tunnels, staging domains) are public DNS, so
the local-dev never-cache rule can't catch them. web.cache_exempt_hosts
lists hosts whose pages are always fetched live: exact, "*.wildcard",
or domain-suffix matching (label-boundary aware — mysite.dev covers
preview.mysite.dev but never evilmysite.dev). Checked on both store
and lookup, so adding an exemption takes effect immediately even for
entries cached before the config change.
Repeat searches (same normalized query + provider) within a 20-minute
TTL are served from an in-process memo, and concurrent identical
queries are single-flighted so a parallel subagent fan-out pays for
one vendor request instead of N. Requested limits bucket up to
10/20/50/100 so near-identical requests share an entry; callers get
their requested count sliced from the bucket.
Repeat extracts of the same URL are served from the existing
cache/web full-text store (previously written for read_file paging
but never read back), via a small JSON sidecar index. Disk-backed, so
CLI, gateway, cron, and subagents share it. Cached extracts re-run
the normal truncate pipeline, so per-call char_limit still works.
Both caches sit after every safety gate (secret-URL, SSRF, policy,
provider resolution) and directly around the paid vendor call — hits
skip only the network request. Only successful responses cache;
rescue-served responses are never cached (one-shot rescue must stay
one-shot). Config: web.cache_enabled (default on),
web.cache_ttl_minutes (default 20, clamped 1-1440).
Idea credit: query coalescing + num-bucketing pattern observed in
Apodex FrontierAgent (Apache-2.0).
Context compression is intentionally lossy. Deployments that archive
transcript evidence to an external durable store before compaction had no
way to guarantee the archive actually happened: MemoryManager.on_pre_compress
swallows provider failures by design, so a failed archive silently degraded
into data loss.
This adds an opt-in, provider-agnostic checkpoint contract:
- memory_provider: PRE_COMPRESS_CHECKPOINT_API_VERSION = 1; providers opt in
by advertising pre_compress_checkpoint_api_version. Version 0 keeps the
historical best-effort hook semantics.
- memory_manager: supports_pre_compress_checkpoint() capability probe;
on_pre_compress(require_checkpoint=True) propagates checkpoint-provider
failures and raises when no capable provider completed the checkpoint.
- conversation_compression: new compression.checkpoint_required config key
(default false, documented in cli-config.yaml.example). When enabled,
compaction fails closed with BLOCKED_MISSING_PREREQUISITE (the
uncompressed transcript is preserved) unless a checkpoint-capable provider
confirms the durable checkpoint. Providers receive normalized direct
user/assistant evidence: tool rows, system messages, tool-call wrappers,
and prior compaction summaries are filtered host-side into one stable
contract. codex_app_server compaction is rejected under the gate because
it exposes no truthful pre-compaction transcript boundary.
- hermes_state: persistent _compressed_summary column (declarative schema
migration via _reconcile_columns) so summary provenance survives process
restarts; only the resume model history carries the marker, keeping
get_messages_as_conversation on its existing contract.
- gateway: the lossy hygiene/auto-compact paths load the memory provider
(skip_memory=False) so a required checkpoint also guards those rewrites.
The gate arms only on an explicit boolean True (bare-MagicMock agents in
existing tests have truthy auto-attributes). Default behavior is unchanged:
checkpoint_required=false preserves best-effort semantics for all existing
providers. Contract tests, including a restart round-trip of the summary
marker, in tests/agent/test_pre_compress_checkpoint_contract.py.
Refs #93986
web_extract stopped using an auxiliary LLM long ago (deterministic
truncate-and-store), but browser snapshots still routed oversized
accessibility trees through the auxiliary web_extract model, keeping a
dead-looking aux slot alive across every config/picker surface.
- tools/browser_tool.py: remove _extract_relevant_content and
_get_extraction_model; oversized snapshots always truncate at line
boundaries, store the full tree to cache/web, and append a read_file
pointer (element refs beyond the cut live in the file)
- tools/browser_camofox.py: same — no LLM path
- Remove auxiliary.web_extract slot: config_defaults (removal note, same
pattern as session_search/PR #27590), cli.py defaults + env bridge,
gateway/run.py bridged keys, hermes config display, hermes model picker,
dashboard REST slots, desktop + web AUX_TASKS, i18n labels (en/zh/
zh-hant/ja/ar)
- Docs: env-vars, configuration, fallback-providers, browser + zh-Hans
mirrors (web-search zh-Hans was stale on the old LLM pipeline — synced
to truncate-and-store truth)
- Tests updated: aux bridge uses approval slot, browser tests assert the
LLM path is gone and stored files are secret-redacted
Follow-up to the #65422 salvage:
- startup_orphan_reap joins _RECOVERABLE_END_REASONS (kept distinct from
ws_orphan_reap for forensics): every recovery fence
(find_latest_gateway_session_for_peer, unarchive_recoverable_session,
promote_to_session_reset) now treats a startup-swept row as an
accidental end, so a sweep never makes a session unresumable.
- Config key moves from sessions.orphan_reaper to
dashboard.startup_orphan_sweep in DEFAULT_CONFIG, next to its siblings
ws_ping_interval / ws_ping_timeout / ws_orphan_reap_grace_s; the raw
loader in tui_gateway.server reads the new key (fail-open on missing).
cli-config.yaml.example and website/docs/user-guide/configuration.md
follow the dashboard.* documentation pattern.
- New regression test: a stranded 'active' row (ended_at NULL, no live
runtime) is swept AND still recoverable via peer-keyed lookup and fully
revivable via reopen_session afterward.
Close session rows left ended_at IS NULL when the in-process websocket
orphan timer dies with the process (#65194). Dual-clock staleness
(started_at AND newest message), desktop included, live in-memory
sessions excluded, scheduled once from both entry.main and the WS
sidecar so desktop/dashboard boots also run the sweep.
Adversarial-review fixes for the #93057 snapshot-compaction PR:
- Fail-closed detachment: only re-enable compression after
bind_session_state successfully severs the engine's parent binding.
A failed rebind keeps the historical compression_enabled=False
behavior and warns, instead of running compaction against a
compressor still bound to the parent's SessionDB (#38727 re-open).
- Warm-cache parity: defer both compression gates (turn-prologue
preflight + pre-API pressure check) until the fork's first provider
response, so the first request replays the full snapshot as the
intended cached read and compaction applies from the second request
on — matching the documented budget mental model.
- Tests: regression for the rebind-failure fail-closed path (red on
pre-fix code) and the existing threshold-crossing test reworked to a
two-request review asserting the warm first request + compacted
second request. 116 tests green across all touched suites; ruff
clean.
Detach the review fork's compressor from the parent SessionDB/session_id
and re-enable in-memory-only compaction for oversized snapshots, instead
of the historical compression_enabled=False guard that left the fork's
replayed transcript unbounded (350k-384k input tokens per request, 1.49M
total across one 8-request review). Add an aggregate input-token budget
(auxiliary.background_review.max_input_tokens, default 600k) so repeated
tool calls cannot recreate an unbounded transcript; the tool loop stops
before the provider call that would cross it.
Closes#93057
- New dashboard.ws_ping_interval / dashboard.ws_ping_timeout defaults
(20.0/20.0) in DEFAULT_CONFIG; hermes_cli/web_server.py reads them for
non-loopback binds. Loopback keeps ws_ping=None (event-loop stalls must
never kill a healthy local connection).
- New dashboard.ws_orphan_reap_grace_s (20.0): tui_gateway/server.py's
_WS_ORPHAN_REAP_GRACE_S now resolves from config via
_resolve_ws_orphan_reap_grace(); the HERMES_TUI_WS_ORPHAN_REAP_GRACE_S
env var is kept as an internal override for backward compat and wins
when set.
- tests/test_ws_keepalive_config.py: real load_config against a temp
HERMES_HOME yaml — defaults, propagation, deep-merge, env override,
invalid-value fallback.
/review takes the last 10 chat messages plus optional instructions,
spawns a full-privilege background subagent (the async delegation
rail) that investigates the referenced work (PR, code, docs), and its
complete review re-enters the spawning session as a normal
async-delegation completion the primary agent can act on.
- agent/review_engine.py: shared engine (snapshot, briefing,
auxiliary.review credential resolution, dispatch, note formatting)
- tools/delegate_tool.py: internal credentials_cfg per-call override
(never model-facing) resolved through the same credential system as
delegation.provider pins
- auxiliary.review config block (provider/model/base_url/api_key/
api_mode); provider auto + empty model = inherit the main model
- Surfaces: CLI process_command, gateway run.py dispatch +
slash_commands handler (binds the approval session key so the
completion routes back), TUI/Desktop live dispatch in
tui_gateway/server.py, CommandDef registry (+Slack /hermes-only cap)
- Docs: delegation.md section + slash-commands.md (both tables)
- Tests: 15 engine tests (sabotage-verified: credentials_cfg and
dispatch tests fail without the fix), 4 gateway handler tests
through the real async rail
Rebase onto main (post-#93102) left two bot_mode dicts in
config_defaults.py — later duplicate key silently wins in a Python
literal, so envelope_ttl_seconds would have shadowed turn_wait_seconds'
section. Single section now carries both keys.
A Bot Mode agent invoked by a handoff runs as a short-lived
`hermes -p <bot> chat -Q --query-file ...` process. When it dispatches
its reply via message_agent / bot_relay — spawned as
terminal(background=true, notify_on_complete=true) per the Bot Chat
protocol — the one-shot parent exits as soon as the turn ends. The
reply child writes to a stdout pipe owned by the dying parent and is
destroyed a few seconds later, so the handoff reply is silently lost
while the sender waits for a notification that can never come (#90879).
Fix (class-wide, not DM-specific): before the one-shot exit paths tear
down, the parent now lingers — bounded by the new
terminal.oneshot_completion_wait_seconds config (default 600s, 0
disables) — for every tracked background process spawned with
notify_on_complete=true. Plain background processes (servers, daemons,
watch-pattern monitors) carry no completion contract and are never
waited on.
- tools/process_registry.py: ProcessRegistry.wait_for_pending_completions()
— bounded, interrupt-safe wait over pending notify_on_complete
sessions; reconciles orphaned-pipe exits (#17327) each pass so a
wedged reader cannot burn the full bound; KeyboardInterrupt aborts
the linger without skipping the caller's durable teardown.
- cli.py: _finalize_single_query() lingers first, before the durable
session flush / cleanup (covers -q and -Q, i.e. the DM recipient
shape and bot_relay waiter spawns from one-shot agents).
- hermes_cli/oneshot.py: same linger before agent.close() (which
kill_all()s the task's processes) on the -z path.
- hermes_cli/config_defaults.py: terminal.oneshot_completion_wait_seconds.
Tests: tests/tools/test_oneshot_completion_linger.py — unit coverage of
the wait semantics (no-op, completion, timeout, task filter, disable,
config fallback, reconcile path), exit-path ordering contracts, and a
real-process E2E: a short-lived python parent spawns a delivery child
through the real ProcessRegistry, lingers, exits, and the delivery
completes; sabotaging the linger makes the same E2E reproduce the
destroyed-delivery symptom.
Fixes#90879
Lowest-hop voice path in both directions for desktop + remote gateway:
mic audio goes straight to the profile's STT provider and reply text is
synthesized on the desktop with the profile's TTS provider. The
desktop-gateway link carries only text (which the chat stream carries
anyway). No second key store: GET /api/audio/voice-config returns the
profile's resolved provider/model/language/key using the exact resolution
chains transcription_tools/tts_tool use, over the authenticated REST
channel. Keys live in renderer memory only.
Backend:
- tools/voice_client_config.py: single resolver; per-provider client
wire shapes (openai-multipart, xai-stt, elevenlabs-stt, openai-speech,
elevenlabs-tts). Server-host-only providers (local whisper, edge,
command/plugin) and missing credentials resolve to {mode: relay}.
xAI OAuth stays relay (bearer refreshes server-side).
- web_server.py: GET /api/audio/voice-config, profile-scoped via the
same _config_profile_scope seam as /api/audio/transcribe.
- config_defaults.py: voice.client_direct gate (default true).
Desktop:
- lib/voice-client-direct.ts: config fetch keyed by (connection,
profile) with 60s TTL, provider-direct STT + TTS calls, sentence
cutter mirroring the server pipeline's contract.
- Dictation (use-prompt-actions + session-tile) tries client-direct
first; null -> existing relay unchanged; provider rejections surface.
- voice-playback.ts: client-direct speech session as the top rung of
startSpeechStream/playSpeechText; WS relay + POST fallback unchanged
below it. Barge-in via the same stopVoicePlayback sequence bump.
Validation: 13/13 backend E2E (real temp HERMES_HOME + real resolution),
live FastAPI TestClient E2E (direct + gate-flip), 15/15 client tests
(wire shapes, scope-keyed caching, rejection surfacing, sentence cutter),
sibling suites 72/72 + 36/36, tsc + eslint + ruff clean.
Docs: voice-mode.md client-direct section ships in this PR.
Downscope of the salvaged #89134 per review: the 3->8 default raise was
symptom tolerance for the false-positive class the off-loop heartbeat +
two-witness probe fixes at the root — fleet-wide it would only delay
genuine-wedge recovery ~2.7x. The three tuning knobs keep independent
operator value and stay:
- default max_strikes back to 3 everywhere (constant, dataclass,
from_dict fallback, floor clamp, tests)
- gateway/config.py + gateway/run.py now reference the
shutdown_watchdog DEFAULT_* constants instead of duplicating literals
in three places (drift hazard)
- knobs registered in hermes_cli/config_defaults.py alongside the
sibling gateway.loop_watchdog bool
Adapts the in-place branch update from PR #89507 (@willfrombr) onto the
switch-by-default behavior: the deterministic switch path remains the
default so non-interactive updates (desktop, gateway, cron) never dead-end
on a merge conflict, and deliberate custom-branch users opt in with
updates.parked_branch_strategy: update_in_place. --switch-branch overrides
the in-place strategy for one run (deep feature branches that must not
accumulate update merge commits). Docs + config comments + tests cover
all three routes.
Co-authored-by: Willian Santos <285090322+willfrombr@users.noreply.github.com>
A clean checkout parked on a feature branch now always switches to the
update target. Unmerged commits are safe on the branch (git checkout
never discards committed work) and get a loud 'kept' notice naming the
branch, count, and the checkout command to resume the work. Previously
the update hard-skipped with exit 1 — a dead end for the desktop update
button, gateway /update, and cron, which have no way to resolve a skip.
Dirty trees (uncommitted changes) still skip loudly, and the
updates.auto_switch_parked_branch: false opt-out still pins the branch.
The feature flag was only documented in cli-config.yaml.example; every other
browser.* key is declared in DEFAULT_CONFIG so config tooling (dashboard
editor, hermes config get) can see it. Defaults unchanged: enabled=False,
developer_mode=False. Surfaced during review of PR #85351.
With the gateway owning the suspend, the idle predicate covers every
work source (agent turns, cron jobs, API-server runs, background work,
fail-awake on unreadable sources) and the relay drains + flips before
the freeze, so a long timeout no longer buys safety — real work always
blocks the suspend and resume is sub-second. 5 idle minutes just bills
idle RAM. Per-instance override stays config.yaml
gateway.scale_to_zero.idle_timeout_minutes (D2).
New behavior-contract test: invalid config values degrade to the module
default (whatever it is), never zero/negative — asserts the RELATION,
not the literal, per the no-change-detector-tests rule.
Builds on @fattchris resolve_turn_limit salvage (#67696): flips the default
from a numeric cap to unlimited across all construction paths (CLI, agent_init,
run_agent subagents), adds inf/infinity/null to the unlimited spellings, and
sets DEFAULT_CONFIG agent.max_turns to null. The turn cap caused more problems
than it solved (silent mid-task truncation).
When the chosen/keyed backend fails a web_search or web_extract call
(bad key, upstream outage, 5xx, raised exception), that single call
retries on the keyless free-tier ring instead of erroring. The next
call attempts the chosen backend again — no sticky failover, no state.
Resolves the keyed half of #78984/#32159 (keyless half landed in the
ring PR).
- tools/web_tools.py: _rescue_eligible (keyed ring vendors + non-ring
backends eligible; keyless-mode calls excluded — they already walked
the ring), _rescue_search/_rescue_extract (search annotates
rescued_from + backend_error naming the original failure and the
retry-next-call semantics; extract rescues only whole-batch failures,
partial failures pass through untouched; rescue failure preserves the
ORIGINAL backend error with the rescue note appended)
- both dispatchers wrap the provider call: failure-results AND raised
exceptions rescue; ineligible paths re-raise unchanged
- web.keyless_rescue config key (default true; implicitly off when
keyless_fallback is off); docs updated
Live E2E: keyed Tavily with an invalid key 401'd and the call was
served by the real ring with the rescue annotation; a second call
re-attempted Tavily first (statelessness proven); whole-batch extract
rescue returned real page content. 13 new tests; 67 green across the
keyless suites.
Fresh installs with zero web credentials now rotate web_search/
web_extract across FIVE vendors' public free tiers — Exa, Parallel,
Tavily, Firecrawl, Keenable — instead of a 2-vendor 50/50 split, with
next-in-line ring failover on rate limits (multi-hop until a vendor
serves or the ring is exhausted; served_by marks the actual vendor).
- plugins/web/keenable/: new bundled provider (search via /v1/search,
fetch via /v1/fetch; keyed Bearer or keyless with the mandatory
X-Keenable-Title app header). Credit: integration proposed by
Ilya Gusev (Keenable) in #49758; Free/Paid picker rows included.
- keyless_mcp: tavily/firecrawl/keenable keyless search+extract
wrappers, _KEYLESS_RING + per-process round-robin cursor (seeded by
the random session id, advances per unpinned request), pinned-vendor
entry (pin = start there; rotation off), paid-pinned vendors excluded
from the ring entirely.
- Tavily/Firecrawl providers route keyless traffic through the ring;
both are now default-on ring members (no longer selection-gated).
- web_tools/registry: keenable in backend sets, auto-detect, availability
probes; _keyless_preference() delegates to the ring cursor.
- KEENABLE_API_KEY in OPTIONAL_ENV_VARS; docs updated (ring semantics).
Live E2E: all 10 vendorXcapability paths (5 search + 5 extract) served
real results keyless; rotation cycled all five vendors over 5 dispatch
calls; double-throttle failover walked exa->parallel->tavily.
- Updated the Tavily API key description to clarify that it is optional and keyless access is supported.
- Modified the Tavily plugin and provider to handle requests with or without an API key, using Bearer authentication when the key is provided.
- Enhanced documentation to reflect the new keyless functionality and updated environment variable descriptions.
- Added tests to ensure correct behavior for both keyed and keyless requests.
Composio eval traces showed Hermes wasting turns re-issuing identical tool
calls (same tool, same args, same result — 3x/4x in one run) and ending
turns by announcing an action it never took. Two conservative, config-gated
guards (agent.stall_guards, default true):
- Identical-call loop breaker: ToolCallGuardrailController.observe_identical_call
tracks the consecutive streak of (tool, canonical args, result-hash); on
the 3rd identical call a compact one-line notice is appended to that tool
RESULT at construction time (cache-safe — tool results are append-only).
Never blocks the call. Pollers (process, *_get_result, *_poll) are exempt
via STALL_GUARD_REPEATABLE_TOOLS. Streak resets on any different call,
changed result, or new turn. Observed on the raw result before the
tool-loop warning suffix so its changing count can't defeat matching.
- Said-continue-but-stopped recovery: trailing_continue_intent() detects a
short reply ENDING on an announced next action ('Let me now…', 'I will
now…', 'Next, I…'); the conversation loop feeds it into the EXISTING
intent-ack continuation path (same interim-assistant + user-nudge
mechanism, same codex_ack_continuations cap of 2), preserving message
alternation — no parallel recovery machinery.
Config: agent.stall_guards in DEFAULT_CONFIG; docs in configuration.md;
unit tests for streak/allowlist/reset/gate and detector pos/neg cases.
Un-fences OPENAI_MODEL_EXECUTION_GUIDANCE from the gpt/codex/grok substring
check and gives it its own injection gate, independent of
tool_use_enforcement, controlled by config.yaml `agent.execution_guidance`
(auto/true/false/list — same semantics as tool_use_enforcement). The "auto"
list (EXECUTION_GUIDANCE_MODELS) now also covers deepseek, kimi, qwen, glm,
minimax, mimo, and mistral.
Composio agentic-eval traces showed Hermes+DeepSeek/Kimi failing where
competitors passed: financial math done in prose, no read-back after
external writes, malformed identifiers "repaired", completeness claimed
despite count mismatches. The discipline block existed but those models
never received it.
The block is extended with compact clauses distilled from that analysis:
- external-write read-back (tool-call success is not task success; internal
file edits already confirmed by the tool are not re-verified)
- count reconciliation (declared totals/has_more are hard assertions)
- literal preservation (never normalize identifiers that fail a stated
format; lookup success does not validate a malformed token)
- retry-differently (empty/partial/suspiciously narrow results get a
broader retry before concluding)
- completion gated on verification (done = every named acceptance
criterion verified, never a plausible subset)
The todo tool description now encourages enumeration-as-checklist for
"all N items" tasks and gates completed status on verified work, never
intent.
Guidance is chosen once at session start keyed on model name, so the
system prompt stays byte-stable for the life of a conversation.
Supersedes/absorbs prior contributor proposals: #20588, #35087, #41874
(MiMo), #53847 (GLM tool-calls-as-text stall).
Co-authored-by: Mat-London <56627804+Mat-London@users.noreply.github.com>
Co-authored-by: intelac <8803887+intelac@users.noreply.github.com>
Co-authored-by: 6ylqq <51219463+6ylqq@users.noreply.github.com>
Co-authored-by: tauros1983 <267660491+tauros1983@users.noreply.github.com>
Both _resolve_openai_audio_client_config resolvers now switch on the
stored provider string: 'nous' (or legacy use_gateway: true) => managed
openai-audio gateway only, erroring by selection name when unentitled —
the STT twin previously never read the stored gateway intent at all, so
a direct OPENAI_API_KEY silently overrode the Nous Subscription pick;
stored vendor => direct credentials only with a selection-naming error
on missing keys (no silent managed fallback); never-configured keeps the
legacy ladder. DEFAULT_CONFIG stops seeding stt.provider: local, and the
seeded value on existing configs is treated as no-selection so autodetect
keeps working for that installed base.
Exa and Parallel now each render as two picker rows in hermes tools —
'Free (keyless)' and 'Paid (API key)'. Selection persists to
web.provider_tier.<name>:
- free: always the anonymous public endpoint, even with a key set
- paid: always the keyed SDK path; missing key errors instead of
silently downgrading to the free tier (is_keyless_available also
returns False so the auto-fallback walk can't route there)
- unset: auto (key present -> paid, else keyless)
Mechanism: get_setup_schema() gains a 'variants' list the picker
flattens into sibling rows sharing one web_backend; selection writes
the tier via both _write_provider_config sites; active-row detection
matches the tier (auto mirrors use_keyless). Routing goes through a
single use_keyless() chokepoint shared by search+extract in both
providers.
Live E2E: tier=free with a fake key present searched keyless OK (a
keyed call would have 401'd); tier=paid without key errored naming
PARALLEL_API_KEY; picker rows verified for both vendors x both tiers.
With zero web credentials configured, web_search/web_extract previously
resolved to the nonfunctional firecrawl sentinel and errored. Now the
backend resolution walks a strictly-last keyless tier: Parallel's and
Exa's public anonymous MCP endpoints (the same free tiers opencode ships
as its default search path).
- plugins/web/keyless_mcp.py: minimal JSON-RPC tools/call client for
mcp.exa.ai + search.parallel.ai (SSE + plain JSON parsing, typed
errors, per-process random session id, no user identifiers)
- WebSearchProvider.is_keyless_available(): separate weaker tier that
never leaks into is_available(), so keyed setups are never pre-empted
- Exa/Parallel providers: route to keyless endpoints when their key is
absent; keyed SDK path unchanged
- registry + _get_backend(): keyless walk (parallel -> exa) strictly
after every keyed/importable candidate; check_web_api_key() lights
the tools up on zero-credential installs
- web.keyless_fallback config key (default true) to disable the tier
- docs: web-search.md + configuration.md
E2E-verified against both live endpoints from an isolated HERMES_HOME
(search + extract via the real dispatchers, disable-flag negative path).
Add model.reasoning_echo (default false) and per-fallback-entry
reasoning_echo to preserve assistant reasoning_content when
replaying history to custom providers and OpenAI-compatible gateways
that proxy thinking-mode models (Kimi K3, GLM-5.2, DeepSeek, etc.)
but are not matched by the built-in host-based _REASONING_ECHO_RULES.
The flag is per-active-provider, not a global toggle:
- Primary: read from model.reasoning_echo at init and switch_model
- Fallback: set by try_activate_fallback from the fallback entry
- Restore: restore_primary_runtime copies the switch_model snapshot
Unlike PR #76019 global agent.reasoning_echo toggle, the
per-provider flag travels with the active provider — falling back to
a strict provider (Mistral, Groq, Cerebras) correctly strips
reasoning_content even when the primary had the flag enabled,
because the flag is False for the strict fallback.
Complements PR #27361 (dynamic detection) which fires after the first
API response; this PR covers turn-1 and history-replay-on-fresh-session
where dynamic detection has not fired yet.
Closes#76018
Refs: #27297, #27361, #76019
Signed-off-by: Yingliang Zhang <zhangyingliang@outlook.com>
Live incident 2026-08-17: the source checkout was parked on a stale feature
branch (claude-code-inspired/local-terminal-memory-limit, days behind main),
left there by earlier tooling. 'hermes update' autostashed, refreshed lazy
backends, synced skills, and printed '✓ Code updated!' / '✓ Update complete!'
while the checkout stayed on the stale branch with none of main's new code.
Two sessions burned time on 'the fix is missing' confusion.
- Parked-branch guard: auto-switch back to the update target ONLY when the
parked branch is clean and fully merged (git cherry origin/<target> shows
nothing unmerged); the checkout then STAYS on the target instead of being
re-parked. Otherwise: loud CODE UPDATE SKIPPED block naming the branch,
behind-count, and resolution commands; exit 1; branch untouched.
- The up-to-date (commit_count == 0) path no longer switches back to a
fully-merged parked branch either.
- Post-pull gate additionally refuses to print '✓ Code updated!' when HEAD
ends up attached to a non-target branch.
- Summary lines now carry the actual branch + HEAD short-sha:
'✓ Update complete! [main @ 30fcf9580]' — drift visible at a glance.
- New config toggle updates.auto_switch_parked_branch (default true).
- Real-git-fixture regression tests (init/clone/branch, no subprocess
mocks): clean+merged auto-switch, dirty skip, unmerged skip, cherry-picked
equivalence, config opt-out, unverifiable ref, on-main fast path,
up-to-date no-repark, summary branch/sha assertions.
Follow-up on the salvaged commits from PRs #87965 and #87967
(@AiwendilInTheWoods):
- Promote the media-send timeout to the standard resolution pattern:
HERMES_CRON_MEDIA_SEND_TIMEOUT env var, then
cron.media_send_timeout_seconds in config.yaml, then 300s default
(mirrors script_timeout_seconds; .env stays secrets-only).
- Register the config key in DEFAULT_CONFIG and document both surfaces
(environment-variables reference + cron user guide).
- Fold the empty-str() exception fallback into the error string recorded
in delivery_errors (post-#88631 the reason reaches the run status, not
just the log line).
- Tests: timeout resolution precedence + TimeoutError reason fallback.
Review feedback from NVIDIA (Nir Paz), minus the LLM items (declined
on the thread: cost-by-default + prompt-injection surface; static-only
also keeps the timeout moot at ~1.5s vs the 120s ceiling):
- Incomplete-validator findings are now PRESERVED as partial evidence;
only the validator's pass/fail verdict is excluded from the advisory
verdict. A report with findings from an incomplete check no longer
reads as clean.
- Clean-report wording is now "no findings from completed checks"
whenever any validator was incomplete.
- Pinned both scanner binaries to known releases in code comments,
config guidance, and docs: SkillEvaluator v0.1.0, SkillSpector v2.9.5.
- Tests: 29 (was 28) — partial-evidence preservation flips the old
discard-pinning test, plus the completed-checks wording case.
Adds an optional, advisory second-opinion scan to the skills hub install
path using NVIDIA SkillEvaluator's deterministic, keyless Tier 1 checks
(PII, unicode smuggling, script lint).
- tools/skillevaluator_scan.py: subprocess adapter — runs the scanner
over the quarantined bundle, parses the JSON report, classifies
secrets-class findings (private keys, tokens, credentialed connection
strings) apart from advisory PII findings. Every failure mode
(binary missing, timeout, crash, bad JSON) degrades to a no-op.
- hermes_cli/skills_hub.py: prints the advisory panel after the built-in
guard's policy decision and before the install confirmation. Findings
are shown with file:line; secrets-class findings render red with a
loud warning. Warn-and-continue by design — the built-in skills guard
remains the only enforcement layer, because the upstream PII scanner
has known false-positive classes (git@github.com, docs example
emails, op:// references).
- hermes_cli/web_routers/skills.py: the dashboard Browse-hub scan
endpoint returns the same advisory data in a new `tier1` field.
- config: skills.tier1_advisory (default true; no-op without the
optional scanner binary on PATH).
- docs: user-guide/features/skills.md section with install command and
config toggle.
Scanner install (optional):
uv tool install --python 3.13 \
"skillevaluator @ git+https://github.com/NVIDIA/SkillEvaluator.git"
E2E-validated against the real scanner binary: clean bundled skill (no
findings, "no findings" line), seeded dirty skill (email + credentialed
connection string -> yellow/red panel, install continues), config
disable via real config.yaml (silence). Real scan cost: ~0.2s per skill.
Sessions started inside a git checkout now source skills from
<root>/.hermes/skills/ and <root>/.agents/skills/ (the cross-tool
convention shared with other agent harnesses) as the highest-precedence
skill tier: project > local > external_dirs.
Loading is trust-gated per repo (skills.trusted_project_dirs, managed by
'hermes skills trust'/'untrust') because skills are executable procedure
documents — auto-sourcing them from any cloned repo is a prompt-injection
vector. Untrusted repos with skills get a one-line banner notice instead.
- agent/skill_utils.py: find_project_root, get_project_skills_dirs,
get_untrusted_project_skills_root, get_scan_ordered_skills_dirs;
project dirs join the curator read-only ownership boundary
- agent/prompt_builder.py: project tier scanned first, entries tagged
[project], same-named local entries shadowed; cache key extended
- tools/skills_tool.py: skills_list scans project dirs first (first-wins);
skill_view resolves cross-tier collisions in favor of the project tier
(same-tier ambiguity still refuses); security warning recognizes the tier
- agent/skill_commands.py + hermes_cli/commands.py: /skill-name slash
commands and gateway slash menus include project skills
- tools/credential_files.py: project dirs mounted into remote backends
- cli.py: banner notice (loaded count / trust hint)
- hermes_cli/main.py + subcommands/skills.py: hermes skills trust/untrust
- config: skills.project_discovery (default on), skills.trusted_project_dirs
- docs: Project-Local Skills section in skills.md
- tests: tests/agent/test_project_skills.py (18 cases)
Session cwd is fixed at agent build time, so the resolved tier is stable
for the conversation and the system prompt stays byte-stable (cache-safe).
Follow-ups on top of the salvaged CommandCode provider plugin (PR #32909):
- hermes_cli/config_defaults.py: COMMANDCODE_API_KEY setup-wizard entry
- hermes_cli/doctor.py: add key to the doctor env-var scan list
(health check comes free via the pluggable-profile loop)
- hermes_cli/dump.py: include commandcode in debug-dump api_keys
- docs: provider table row, fallback-provider table + supported lists
- tests: doctor dedicated-skip test now uses exact-name checks so
Bearer-authed Anthropic-COMPATIBLE gateways (CommandCode (Anthropic))
are allowed in the generic loop while native anthropic stays skipped
E2E verified with real imports: profile registration, aliases,
PROVIDER_REGISTRY auto-extension, bearer-auth host match
(positive + negative), live /models fetch (55 models).
Addresses three gaps found in review of the memory-guard PR:
P1a — standalone daemon was the one uncapped entry point. run_daemon()
now resolves kanban.max_in_progress every tick (explicit config wins,
else the memory-derived default) exactly like the gateway dispatcher and
`hermes kanban dispatch`. New shared parser configured_max_in_progress()
so all three entry points agree on what "explicitly configured" means.
P1b — max_in_progress was enforced per board while the gateway ticks
every active board, multiplying the host budget by the number of boards
(2 boards x cap 2 = 4 workers on a host sized for 2). The cap is now
host-level: _dispatch_once_locked() adds count_running_tasks_other_boards()
to the running count before deriving the tick's spawn budget. Enforced in
the shared locked path, so gateway, CLI, and daemon all inherit it.
max_spawn deliberately keeps its historical per-board semantics.
Fails open per board so one corrupt board can't brick dispatch on the rest.
P2 — the ready loop consumed the entire shared spawn budget before the
review loop ran, so a sustained ready backlog starved autonomous reviews
indefinitely. When spawnable review work exists (assigned + real profile,
mirroring the review loop's own gate) and the tick has budget, one slot
is held back from the ready lane. Reservation is per-tick and
self-releasing; the review lane still spends from the shared budget —
it gains fairness, not extra capacity.
11 new tests in tests/hermes_cli/test_kanban_host_cap.py. Existing
kanban suites: 278 passed (15 failures pre-existing, identical on clean
main baseline). ruff clean.
alt-glitch