Wrap the existing version label in the welcome-banner panel title
('Hermes Agent v… · upstream … · local …') with an OSC-8 terminal
hyperlink pointing at the latest git tag's GitHub release page
(https://github.com/NousResearch/hermes-agent/releases/tag/<tag>).
Clickable in modern terminals (iTerm2, WezTerm, Windows Terminal,
GNOME Terminal, Kitty, etc.); degrades to plain text on terminals
without OSC-8 support. No new line added to the banner.
New get_latest_release_tag() helper runs 'git describe --tags
--abbrev=0' in the Hermes checkout (3s timeout, per-process cache,
silent fallback for non-git/pip installs and forks without tags).
OpenRouter returns a 404 with the specific message
'No endpoints available matching your guardrail restrictions and data
policy. Configure: https://openrouter.ai/settings/privacy'
when a user's account-level privacy setting excludes the only endpoint
serving a model (e.g. DeepSeek V4 Pro, which today is hosted only by
DeepSeek's own endpoint that may log inputs).
Before this change we classified it as model_not_found, which was
misleading (the model exists) and triggered provider fallback (useless —
the same account setting applies to every OpenRouter call).
Now it classifies as a new FailoverReason.provider_policy_blocked with
retryable=False, should_fallback=False. The error body already contains
the fix URL, so the user still gets actionable guidance.
- disable ANSI dim on VTE terminals by default so dark-background reasoning and accents stay readable
- suppress local multiplexer OSC52 echo while preserving remote passthrough and add regression coverage
On ChatGPT Codex OAuth every gpt-5.x slug actually caps at 272,000 tokens,
but Hermes was resolving gpt-5.5 / gpt-5.4 to 1,050,000 (from models.dev)
because openai-codex aliases to the openai entry there. At 1.05M the
compressor never fires and requests hard-fail with 'context window
exceeded' around the real 272k boundary.
Verified live against chatgpt.com/backend-api/codex/models:
gpt-5.5, gpt-5.4, gpt-5.4-mini, gpt-5.3-codex, gpt-5.2-codex,
gpt-5.2, gpt-5.1-codex-max → context_window = 272000
Changes:
- agent/model_metadata.py:
* _fetch_codex_oauth_context_lengths() — probe the Codex /models
endpoint with the OAuth bearer token and read context_window per
slug (1h in-memory TTL).
* _resolve_codex_oauth_context_length() — prefer the live probe,
fall back to hardcoded _CODEX_OAUTH_CONTEXT_FALLBACK (all 272k).
* Wire into get_model_context_length() when provider=='openai-codex',
running BEFORE the models.dev lookup (which returns 1.05M). Result
persists via save_context_length() so subsequent lookups skip the
probe entirely.
* Fixed the now-wrong comment on the DEFAULT_CONTEXT_LENGTHS gpt-5.5
entry (400k was never right for Codex; it's the catch-all for
providers we can't probe live).
Tests (4 new in TestCodexOAuthContextLength):
- fallback table used when no token is available (no models.dev leakage)
- live probe overrides the fallback
- probe failure (non-200) falls back to hardcoded 272k
- non-codex providers (openrouter, direct openai) unaffected
Non-codex context resolution is unchanged — the Codex branch only fires
when provider=='openai-codex'.
* docs: browser CDP supervisor design (for upcoming PR)
Design doc ahead of implementation — dialog + iframe detection/interaction
via a persistent CDP supervisor. Covers backend capability matrix (verified
live 2026-04-23), architecture, lifecycle, policy, agent surface, PR split,
non-goals, and test plan.
Supersedes #12550.
No code changes in this commit.
* feat(browser): add persistent CDP supervisor for dialog + frame detection
Single persistent CDP WebSocket per Hermes task_id that subscribes to
Page/Runtime/Target events and maintains thread-safe state for pending
dialogs, frame tree, and console errors.
Supervisor lives in its own daemon thread running an asyncio loop;
external callers use sync API (snapshot(), respond_to_dialog()) that
bridges onto the loop.
Auto-attaches to OOPIF child targets via Target.setAutoAttach{flatten:true}
and enables Page+Runtime on each so iframe-origin dialogs surface through
the same supervisor.
Dialog policies: must_respond (default, 300s safety timeout),
auto_dismiss, auto_accept.
Frame tree capped at 30 entries + OOPIF depth 2 to keep snapshot
payloads bounded on ad-heavy pages.
E2E verified against real Chrome via smoke test — detects + responds
to main-frame alerts, iframe-contentWindow alerts, preserves frame
tree, graceful no-dialog error path, clean shutdown.
No agent-facing tool wiring in this commit (comes next).
* feat(browser): add browser_dialog tool wired to CDP supervisor
Agent-facing response-only tool. Schema:
action: 'accept' | 'dismiss' (required)
prompt_text: response for prompt() dialogs (optional)
dialog_id: disambiguate when multiple dialogs queued (optional)
Handler:
SUPERVISOR_REGISTRY.get(task_id).respond_to_dialog(...)
check_fn shares _browser_cdp_check with browser_cdp so both surface and
hide together. When no supervisor is attached (Camofox, default
Playwright, or no browser session started yet), tool is hidden; if
somehow invoked it returns a clear error pointing the agent to
browser_navigate / /browser connect.
Registered in _HERMES_CORE_TOOLS and the browser / hermes-acp /
hermes-api-server toolsets alongside browser_cdp.
* feat(browser): wire CDP supervisor into session lifecycle + browser_snapshot
Supervisor lifecycle:
* _get_session_info lazy-starts the supervisor after a session row is
materialized — covers every backend code path (Browserbase, cdp_url
override, /browser connect, future providers) with one hook.
* cleanup_browser(task_id) stops the supervisor for that task first
(before the backend tears down CDP).
* cleanup_all_browsers() calls SUPERVISOR_REGISTRY.stop_all().
* /browser connect eagerly starts the supervisor for task 'default'
so the first snapshot already shows pending_dialogs.
* /browser disconnect stops the supervisor.
CDP URL resolution for the supervisor:
1. BROWSER_CDP_URL / browser.cdp_url override.
2. Fallback: session_info['cdp_url'] from cloud providers (Browserbase).
browser_snapshot merges supervisor state (pending_dialogs + frame_tree)
into its JSON output when a supervisor is active — the agent reads
pending_dialogs from the snapshot it already requests, then calls
browser_dialog to respond. No extra tool surface.
Config defaults:
* browser.dialog_policy: 'must_respond' (new)
* browser.dialog_timeout_s: 300 (new)
No version bump — new keys deep-merge into existing browser section.
Deadlock fix in supervisor event dispatch:
* _on_dialog_opening and _on_target_attached used to await CDP calls
while the reader was still processing an event — but only the reader
can set the response Future, so the call timed out.
* Both now fire asyncio.create_task(...) so the reader stays pumping.
* auto_dismiss/auto_accept now actually close the dialog immediately.
Tests (tests/tools/test_browser_supervisor.py, 11 tests, real Chrome):
* supervisor start/snapshot
* main-frame alert detection + dismiss
* iframe.contentWindow alert
* prompt() with prompt_text reply
* respond with no pending dialog -> clean error
* auto_dismiss clears on event
* registry idempotency
* registry stop -> snapshot reports inactive
* browser_dialog tool no-supervisor error
* browser_dialog invalid action
* browser_dialog end-to-end via tool handler
xdist-safe: chrome_cdp fixture uses a per-worker port.
Skipped when google-chrome/chromium isn't installed.
* docs(browser): document browser_dialog tool + CDP supervisor
- user-guide/features/browser.md: new browser_dialog section with
workflow, availability gate, and dialog_policy table
- reference/tools-reference.md: row for browser_dialog, tool count
bumped 53 -> 54, browser tools count 11 -> 12
- reference/toolsets-reference.md: browser_dialog added to browser
toolset row with note on pending_dialogs / frame_tree snapshot fields
Full design doc lives at
developer-guide/browser-supervisor.md (committed earlier).
* fix(browser): reconnect loop + recent_dialogs for Browserbase visibility
Found via Browserbase E2E test that revealed two production-critical issues:
1. **Supervisor WebSocket drops when other clients disconnect.** Browserbase's
CDP proxy tears down our long-lived WebSocket whenever a short-lived
client (e.g. agent-browser CLI's per-command CDP connection) disconnects.
Fixed with a reconnecting _run loop that re-attaches with exponential
backoff on drops. _page_session_id and _child_sessions are reset on each
reconnect; pending_dialogs and frames are preserved across reconnects.
2. **Browserbase auto-dismisses dialogs server-side within ~10ms.** Their
Playwright-based CDP proxy dismisses alert/confirm/prompt before our
Page.handleJavaScriptDialog call can respond. So pending_dialogs is
empty by the time the agent reads a snapshot on Browserbase.
Added a recent_dialogs ring buffer (capacity 20) that retains a
DialogRecord for every dialog that opened, with a closed_by tag:
* 'agent' — agent called browser_dialog
* 'auto_policy' — local auto_dismiss/auto_accept fired
* 'watchdog' — must_respond timeout auto-dismissed (300s default)
* 'remote' — browser/backend closed it on us (Browserbase)
Agents on Browserbase now see the dialog history with closed_by='remote'
so they at least know a dialog fired, even though they couldn't respond.
3. **Page.javascriptDialogClosed matching bug.** The event doesn't include a
'message' field (CDP spec has only 'result' and 'userInput') but our
_on_dialog_closed was matching on message. Fixed to match by session_id
+ oldest-first, with a safety assumption that only one dialog is in
flight per session (the JS thread is blocked while a dialog is up).
Docs + tests updated:
* browser.md: new availability matrix showing the three backends and
which mode (pending / recent / response) each supports
* developer-guide/browser-supervisor.md: three-field snapshot schema
with closed_by semantics
* test_browser_supervisor.py: +test_recent_dialogs_ring_buffer (12/12
passing against real Chrome)
E2E verified both backends:
* Local Chrome via /browser connect: detect + respond full workflow
(smoke_supervisor.py all 7 scenarios pass)
* Browserbase: detect via recent_dialogs with closed_by='remote'
(smoke_supervisor_browserbase_v2.py passes)
Camofox remains out of scope (REST-only, no CDP) — tracked for
upstream PR 3.
* feat(browser): XHR bridge for dialog response on Browserbase (FIXED)
Browserbase's CDP proxy auto-dismisses native JS dialogs within ~10ms, so
Page.handleJavaScriptDialog calls lose the race. Solution: bypass native
dialogs entirely.
The supervisor now injects Page.addScriptToEvaluateOnNewDocument with a
JavaScript override for window.alert/confirm/prompt. Those overrides
perform a synchronous XMLHttpRequest to a magic host
('hermes-dialog-bridge.invalid'). We intercept those XHRs via Fetch.enable
with a requestStage=Request pattern.
Flow when a page calls alert('hi'):
1. window.alert override intercepts, builds XHR GET to
http://hermes-dialog-bridge.invalid/?kind=alert&message=hi
2. Sync XHR blocks the page's JS thread (mirrors real dialog semantics)
3. Fetch.requestPaused fires on our WebSocket; supervisor surfaces
it as a pending dialog with bridge_request_id set
4. Agent reads pending_dialogs from browser_snapshot, calls browser_dialog
5. Supervisor calls Fetch.fulfillRequest with JSON body:
{accept: true|false, prompt_text: '...', dialog_id: 'd-N'}
6. The injected script parses the body, returns the appropriate value
from the override (undefined for alert, bool for confirm, string|null
for prompt)
This works identically on Browserbase AND local Chrome — no native dialog
ever fires, so Browserbase's auto-dismiss has nothing to race. Dialog
policies (must_respond / auto_dismiss / auto_accept) all still work.
Bridge is installed on every attached session (main page + OOPIF child
sessions) so iframe dialogs are captured too.
Native-dialog path kept as a fallback for backends that don't auto-dismiss
(so a page that somehow bypasses our override — e.g. iframes that load
after Fetch.enable but before the init-script runs — still gets observed
via Page.javascriptDialogOpening).
E2E VERIFIED:
* Local Chrome: 13/13 pytest tests green (12 original + new
test_bridge_captures_prompt_and_returns_reply_text that asserts
window.__ret === 'AGENT-SUPPLIED-REPLY' after agent responds)
* Browserbase: smoke_bb_bridge_v2.py runs 4/4 PASS:
- alert('BB-ALERT-MSG') dismiss → page.alert_ret = undefined ✓
- prompt('BB-PROMPT-MSG', 'default-xyz') accept with 'AGENT-REPLY'
→ page.prompt_ret === 'AGENT-REPLY' ✓
- confirm('BB-CONFIRM-MSG') accept → page.confirm_ret === true ✓
- confirm('BB-CONFIRM-MSG') dismiss → page.confirm_ret === false ✓
Docs updated in browser.md and developer-guide/browser-supervisor.md —
availability matrix now shows Browserbase at full parity with local
Chrome for both detection and response.
* feat(browser): cross-origin iframe interaction via browser_cdp(frame_id=...)
Adds iframe interaction to the CDP supervisor PR (was queued as PR 2).
Design: browser_cdp gets an optional frame_id parameter. When set, the
tool looks up the frame in the supervisor's frame_tree, grabs its child
cdp_session_id (OOPIF session), and dispatches the CDP call through the
supervisor's already-connected WebSocket via run_coroutine_threadsafe.
Why not stateless: on Browserbase, each fresh browser_cdp WebSocket
must re-negotiate against a signed connectUrl. The session info carries
a specific URL that can expire while the supervisor's long-lived
connection stays valid. Routing via the supervisor sidesteps this.
Agent workflow:
1. browser_snapshot → frame_tree.children[] shows OOPIFs with is_oopif=true
2. browser_cdp(method='Runtime.evaluate', frame_id=<OOPIF frame_id>,
params={'expression': 'document.title', 'returnByValue': True})
3. Supervisor dispatches the call on the OOPIF's child session
Supervisor state fixes needed along the way:
* _on_frame_detached now skips reason='swap' (frame migrating processes)
* _on_frame_detached also skips when the frame is an OOPIF with a live
child session — Browserbase fires spurious remove events when a
same-origin iframe gets promoted to OOPIF
* _on_target_detached clears cdp_session_id but KEEPS the frame record
so the agent still sees the OOPIF in frame_tree during transient
session flaps
E2E VERIFIED on Browserbase (smoke_bb_iframe_agent_path.py):
browser_cdp(method='Runtime.evaluate',
params={'expression': 'document.title', 'returnByValue': True},
frame_id=<OOPIF>)
→ {'success': True, 'result': {'value': 'Example Domain'}}
The iframe is <iframe src='https://example.com/'> inside a top-level
data: URL page on a real Browserbase session. The agent Runtime.evaluates
INSIDE the cross-origin iframe and gets example.com's title back.
Tests (tests/tools/test_browser_supervisor.py — 16 pass total):
* test_browser_cdp_frame_id_routes_via_supervisor — injects fake OOPIF,
verifies routing via supervisor, Runtime.evaluate returns 1+1=2
* test_browser_cdp_frame_id_missing_supervisor — clean error when no
supervisor attached
* test_browser_cdp_frame_id_not_in_frame_tree — clean error on bad
frame_id
Docs (browser.md and developer-guide/browser-supervisor.md) updated with
the iframe workflow, availability matrix now shows OOPIF eval as shipped
for local Chrome + Browserbase.
* test(browser): real-OOPIF E2E verified manually + chrome_cdp uses --site-per-process
When asked 'did you test the iframe stuff' I had only done a mocked
pytest (fake injected OOPIF) plus a Browserbase E2E. Closed the
local-Chrome real-OOPIF gap by writing /tmp/dialog-iframe-test/
smoke_local_oopif.py:
* 2 http servers on different hostnames (localhost:18905 + 127.0.0.1:18906)
* Chrome with --site-per-process so the cross-origin iframe becomes a
real OOPIF in its own process
* Navigate, find OOPIF in supervisor.frame_tree, call
browser_cdp(method='Runtime.evaluate', frame_id=<OOPIF>) which routes
through the supervisor's child session
* Asserts iframe document.title === 'INNER-FRAME-XYZ' (from the
inner page, retrieved via OOPIF eval)
PASSED on 2026-04-23.
Tried to embed this as a pytest but hit an asyncio version quirk between
venv (3.11) and the system python (3.13) — Page.navigate hangs in the
pytest harness but works in standalone. Left a self-documenting skip
test that points to the smoke script + describes the verification.
chrome_cdp fixture now passes --site-per-process so future iframe tests
can rely on OOPIF behavior.
Result: 16 pass + 1 documented-skip = 17 tests in
tests/tools/test_browser_supervisor.py.
* docs(browser): add dialog_policy + dialog_timeout_s to configuration.md, fix tool count
Pre-merge docs audit revealed two gaps:
1. user-guide/configuration.md browser config example was missing the
two new dialog_* knobs. Added with a short table explaining
must_respond / auto_dismiss / auto_accept semantics and a link to
the feature page for the full workflow.
2. reference/tools-reference.md header said '54 built-in tools' — real
count on main is 54, this branch adds browser_dialog so it's 55.
Fixed the header. (browser count was already correctly bumped
11 -> 12 in the earlier docs commit.)
No code changes.
Generates a full dedicated Docusaurus page for every one of the 132 skills
(73 bundled + 59 optional) under website/docs/user-guide/skills/{bundled,optional}/<category>/.
Each page carries the skill's description, metadata (version, author, license,
dependencies, platform gating, tags, related skills cross-linked to their own
pages), and the complete SKILL.md body that Hermes loads at runtime.
Previously the two catalog pages just listed skills with a one-line blurb and
no way to see what the skill actually did — users had to go read the source
repo. Now every skill has a browsable, searchable, cross-linked reference in
the docs.
- website/scripts/generate-skill-docs.py — generator that reads skills/ and
optional-skills/, writes per-skill pages, regenerates both catalog indexes,
and rewrites the Skills section of sidebars.ts. Handles MDX escaping
(outside fenced code blocks: curly braces, unsafe HTML-ish tags) and
rewrites relative references/*.md links to point at the GitHub source.
- website/docs/reference/skills-catalog.md — regenerated; each row links to
the new dedicated page.
- website/docs/reference/optional-skills-catalog.md — same.
- website/sidebars.ts — Skills section now has Bundled / Optional subtrees
with one nested category per skill folder.
- .github/workflows/{docs-site-checks,deploy-site}.yml — run the generator
before docusaurus build so CI stays in sync with the source SKILL.md files.
Build verified locally with `npx docusaurus build`. Only remaining warnings
are pre-existing broken link/anchor issues in unrelated pages.
* feat(config): make tool output truncation limits configurable
Port from anomalyco/opencode#23770: expose a new `tool_output` config
section so users can tune the hardcoded truncation caps that apply to
terminal output and read_file pagination.
Three knobs under `tool_output`:
- max_bytes (default 50_000) — terminal stdout/stderr cap
- max_lines (default 2000) — read_file pagination cap
- max_line_length (default 2000) — per-line cap in line-numbered view
All three keep their existing hardcoded values as defaults, so behaviour
is unchanged when the section is absent. Power users on big-context
models can raise them; small-context local models can lower them.
Implementation:
- New `tools/tool_output_limits.py` reads the section with defensive
fallback (missing/invalid values → defaults, never raises).
- `tools/terminal_tool.py` MAX_OUTPUT_CHARS now comes from
get_max_bytes().
- `tools/file_operations.py` normalize_read_pagination() and
_add_line_numbers() now pull the limits at call time.
- `hermes_cli/config.py` DEFAULT_CONFIG gains the `tool_output` section
so `hermes setup` writes defaults into fresh configs.
- Docs page `user-guide/configuration.md` gains a "Tool Output
Truncation Limits" section with large-context and small-context
example configs.
Tests (18 new in tests/tools/test_tool_output_limits.py):
- Default resolution with missing / malformed / non-dict config.
- Full and partial user overrides.
- Coercion of bad values (None, negative, wrong type, str int).
- Shortcut accessors delegate correctly.
- DEFAULT_CONFIG exposes the section with the right defaults.
- Integration: normalize_read_pagination clamps to the configured
max_lines.
* feat(skills): add design-md skill for Google's DESIGN.md spec
Built-in skill under skills/creative/ that teaches the agent to author,
lint, diff, and export DESIGN.md files — Google's open-source
(Apache-2.0) format for describing a visual identity to coding agents.
Covers:
- YAML front matter + markdown body anatomy
- Full token schema (colors, typography, rounded, spacing, components)
- Canonical section order + duplicate-heading rejection
- Component property whitelist + variants-as-siblings pattern
- CLI workflow via 'npx @google/design.md' (lint/diff/export/spec)
- Lint rule reference including WCAG contrast checks
- Common YAML pitfalls (quoted hex, negative dimensions, dotted refs)
- Starter template at templates/starter.md
Package verified live on npm (@google/design.md@0.1.1).
MCP stdio servers' stderr was being dumped directly onto the user's
terminal during hermes launch. Servers like FastMCP-based ones print a
large ASCII banner at startup; slack-mcp-server emits JSON logs; etc.
With prompt_toolkit / Rich rendering the TUI concurrently, these
unsolicited writes corrupt the terminal state — hanging the session
~80% of the time for one user with Google Ads Tools + slack-mcp
configured, forcing Ctrl+C and restart loops.
Root cause: `stdio_client(server_params)` in tools/mcp_tool.py was
called without `errlog=`, and the SDK's default is `sys.stderr` —
i.e. the real parent-process stderr, which is the TTY.
Fix: open a shared, append-mode log at $HERMES_HOME/logs/mcp-stderr.log
(created once per process, line-buffered, real fd required by asyncio's
subprocess machinery) and pass it as `errlog` to every stdio_client.
Each server's spawn writes a timestamped header so the shared log stays
readable when multiple servers are running. Falls back to /dev/null if
the log file cannot be opened.
Verified by E2E spawning a subprocess with the log fd as its stderr:
banner lines land in the log file, nothing reaches the calling TTY.
FloatingOverlays (SessionPicker, ModelPicker, SkillsHub, pager,
completions) was nested inside the !isBlocked guard in ComposerPane.
When any overlay opened, isBlocked became true, which removed the
entire composer box from the tree — including the overlay that was
trying to render. This made /resume with no args appear to do nothing
(the input line vanished and no picker appeared).
Since 99d859ce (feat: refactor by splitting up app and doing proper
state), isBlocked gated only the text input lines so that
approval/clarify prompts and pickers rendered above a hidden composer.
The regression happened in 408fc893 (fix(tui): tighten composer — status
sits directly above input, overlays anchor to input) when
FloatingOverlays was moved into the input row for anchoring but
accidentally kept inside the !isBlocked guard.
so here, we render FloatingOverlays outside the !isBlocked guard inside
the same position:relative Box, so overlays
stay visible even when text input is hidden. Only the actual input
buffer lines and TextInput are gated now.
Fixes: /resume, /history, /logs, /model, /skills, and completion
dropdowns when blocked overlays are active.
Two fixes on top of the fuzzy-@ branch:
(1) Rebase artefact: re-apply only the fuzzy additions on top of
fresh `tui_gateway/server.py`. The earlier commit was cut from a
base 58 commits behind main and clobbered ~170 lines of
voice.toggle / voice.record handlers and the gateway crash hooks
(`_panic_hook`, `_thread_panic_hook`). Reset server.py to
origin/main and re-add only:
- `_FUZZY_*` constants + `_list_repo_files` + `_fuzzy_basename_rank`
- the new fuzzy branch in the `complete.path` handler
(2) Path scoping (Copilot review): `git ls-files` returns repo-root-
relative paths, but completions need to resolve under the gateway's
cwd. When hermes is launched from a subdirectory, the previous
code surfaced `@file:apps/web/src/foo.tsx` even though the agent
would resolve that relative to `apps/web/` and miss. Fix:
- `git -C root rev-parse --show-toplevel` to get repo top
- `git -C top ls-files …` for the listing
- `os.path.relpath(top + p, root)` per result, dropping anything
starting with `../` so the picker stays scoped to cwd-and-below
(matches Cmd-P workspace semantics)
`apps/web/src/foo.tsx` ends up as `@file:src/foo.tsx` from inside
`apps/web/`, and sibling subtrees + parent-of-cwd files don't leak.
New test `test_fuzzy_paths_relative_to_cwd_inside_subdir` builds a
3-package mono-repo, runs from `apps/web/`, and verifies completion
paths are subtree-relative + outside-of-cwd files don't appear.
Copilot review threads addressed: #3134675504 (path scoping),
#3134675532 (`voice.toggle` regression), #3134675541 (`voice.record`
regression — both were stale-base artefacts, not behavioural changes).
Rebase-artefact cleanup on this branch:
- Restore `voice.status` and `voice.transcript` cases in
createGatewayEventHandler plus the `voice` / `submission` /
`composer.setInput` ctx destructuring. They were added to main in
the 58-commit gap that this branch was originally cut behind;
dropping them was unintentional.
- Rebase the test ctx shape to match main (voice.* fakes,
submission.submitRef, composer.setInput) and apply the same
segment-anchor test rewrites on top.
- Drop the `#14XXX` placeholder from the tool.complete comment;
replace with a plain-English rationale.
- Rewrite the broken mid-word "pushInlineDiff- Segment" in
turnController's dedupe comment to refer to
pushInlineDiffSegment and `kind: 'diff'` plainly.
- Collapse the filter predicate in recordMessageComplete from a
4-line if/return into one boolean expression — same semantics,
reads left-to-right as a single predicate.
Copilot review threads resolved: #3134668789, #3134668805,
#3134668822.
Visual polish on top of the segment-anchor change: diff blocks were
butting up against the narration around them. Tag diff-only segments
with `kind: 'diff'` (extended on Msg) and give them `marginTop={1}` +
`marginBottom={1}` in MessageLine, matching the spacing we already
use for user messages. Also swaps the regex-based `diffSegmentBody`
check for an explicit `kind === 'diff'` guard so the dedupe path is
clearer.
Revisits #13729. That PR buffered each `tool.complete`'s inline_diff
and merged them into the final assistant message body as a fenced
```diff block. The merge-at-end placement reads as "the agent wrote
this after the summary", even when the edit fired mid-turn — which
is both misleading and (per blitz feedback) feels like noise tacked
onto the end of every task.
Segment-anchored placement instead:
- On tool.complete with inline_diff, `pushInlineDiffSegment` calls
`flushStreamingSegment` first (so any in-progress narration lands
as its own segment), then pushes the ```diff block as its own
segment into segmentMessages. The diff is now anchored BETWEEN the
narration that preceded the edit and whatever the agent streams
afterwards, which is where the edit actually happened.
- `recordMessageComplete` no longer merges buffered diffs. The only
remaining dedupe is "drop diff-only segments whose body the final
assistant text narrates verbatim (or whose diff fence the final
text already contains)" — same tradeoff as before, kept so an
agent that narrates its own diff doesn't render two stacked copies.
- Drops `pendingInlineDiffs` and `queueInlineDiff` — buffer + end-
merge machinery is gone; segmentMessages is now the only source
of truth.
Side benefit: Ctrl+C interrupt (`interruptTurn`) iterates
segmentMessages, so diff segments are now preserved in the
transcript when the user cancels after an edit. Previously the
pending buffer was silently dropped on interrupt.
Reported by Teknium during blitz usage: "no diffs are ever at the
end because it didn't make this file edit after the final message".
Typing `@appChrome` in the composer should surface
`ui-tui/src/components/appChrome.tsx` without requiring the user to
first type the full directory path — matches the Cmd-P behaviour
users expect from modern editors.
The gateway's `complete.path` handler was doing a plain
`os.listdir(".")` + `startswith` prefix match, so basenames only
resolved inside the current working directory. This reworks it to:
- enumerate repo files via `git ls-files -z --cached --others
--exclude-standard` (fast, honours `.gitignore`); fall back to a
bounded `os.walk` that skips common vendor / build dirs when the
working dir isn't a git repo. Results cached per-root with a 5s
TTL so rapid keystrokes don't respawn git processes.
- rank basenames with a 5-tier scorer: exact → prefix → camelCase
/ word-boundary → substring → subsequence. Shorter basenames win
ties; shorter rel paths break basename-length ties.
- only take the fuzzy branch when the query is bare (no `/`), is a
context reference (`@...`), and isn't `@folder:` — path-ish
queries and folder tags fall through to the existing
directory-listing path so explicit navigation intent is
preserved.
Completion rows now carry `display = basename`,
`meta = directory`, so the picker renders
`appChrome.tsx ui-tui/src/components` on one row (basename bold,
directory dim) — the meta column was previously "dir" / "" and is
a more useful signal for fuzzy hits.
Reported by Ben Barclay during the TUI v2 blitz test.
Adds a per-ink-text measurement cache keyed by width|widthMode to avoid
re-squashing and re-wrapping the same text when yoga calls measureFunc
multiple times per frame with different widths during flex layout re-pass.
When a tool schema declares `type: array` or `type: object` and the model
emits the value as a JSON string (common with complex oneOf discriminated
unions), the MCP server rejects it with -32602 "expected array, received
string". Extend `_coerce_value` to attempt `json.loads` for these types
and replace the string with the parsed value before dispatch.
Root cause confirmed via live testing: `add_reminders.reminders` uses a
oneOf discriminated union (relative/absolute/location) that triggers model
output drift. Sending a real array passes validation; sending a string
reproduces the exact error.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The voice.toggle handler was persisting display.voice_enabled /
display.voice_tts to config.yaml, so a TUI session that ever turned
voice on would re-open with it already on (and the mic badge lit) on
every subsequent launch. cli.py treats voice strictly as runtime
state: _voice_mode = False at __init__, only /voice on flips it, and
nothing writes it back to disk.
Drop the _write_config_key calls in voice.toggle on/off/tts and the
config.yaml fallback in _voice_mode_enabled / _voice_tts_enabled.
State is now env-var-only (HERMES_VOICE / HERMES_VOICE_TTS), scoped to
the live gateway subprocess — the next launch starts clean.
Crash-log stack trace (tui_gateway_crash.log) from the user's session
pinned the regression: SIGPIPE arrived while main thread was blocked on
for-raw-in-sys.stdin — i.e., a background thread (debug print to stderr,
most likely from HERMES_VOICE_DEBUG=1) wrote to a pipe whose buffer the
TUI hadn't drained yet, and SIG_DFL promptly killed the process.
Two fixes that together restore CLI parity:
- entry.py: SIGPIPE → SIG_IGN instead of the _log_signal handler that
then exited. With SIG_IGN, Python raises BrokenPipeError on the
offending write, which write_json already handles with a clean exit
via _log_exit. SIGTERM / SIGHUP still route through _log_signal so
real termination signals remain diagnosable.
- hermes_cli/voice.py:_debug: wrap the stderr print in a BrokenPipeError
/ OSError try/except. This runs from daemon threads (silence callback,
TTS playback, beep), so a broken stderr must not escape and ride up
into the main event loop.
Verified by spawning the gateway subprocess locally:
voice.toggle status → 200 OK, process stays alive, clean exit on
stdin close logs "reason=stdin EOF" instead of a silent reap.
SIG_DFL for SIGPIPE means the kernel reaps the gateway subprocess the
instant a background thread (TTS playback, silence callback, voice
status emitter) writes to a stdout the TUI stopped reading — before
the Python interpreter can run excepthook, threading.excepthook,
atexit, or the entry.py post-loop _log_exit.
Replace the three SIG_DFL / SIG_IGN bindings with a _log_signal
handler that:
- records which signal (SIGPIPE / SIGTERM / SIGHUP) fired and when;
- dumps the main-thread stack at signal delivery AND every live
thread's stack via sys._current_frames — the background-thread
write that provoked SIGPIPE is almost always visible here;
- writes everything to ~/.hermes/logs/tui_gateway_crash.log and prints
a [gateway-signal] breadcrumb to stderr so the TUI Activity surfaces
it as well.
SIGINT stays ignored (TUI handles Ctrl+C for the user).
Gateway exits weren't reaching the panic hook because entry.py calls
sys.exit(0) on broken stdout — clean termination, no exception. That
left "gateway exited" in the TUI with zero forensic trail when pipe
breaks happened mid-turn.
Entry.py now tags each exit path — startup-write failure, parse-error-
response write failure, per-method response write failure, stdin EOF —
with a one-line entry in ~/.hermes/logs/tui_gateway_crash.log and a
gateway.stderr breadcrumb. Includes the JSON-RPC method name on the
dispatch path, which is the only way to tell "died right after handling
voice.toggle on" from "died emitting the second message.complete".
When the gateway subprocess raises an unhandled exception during a
voice-mode turn, nothing survives: stdout is the JSON-RPC pipe, stderr
flushes but the process is already exiting, and no log file catches
Python's default traceback print. The user is left with an
undiagnosable "gateway exited" banner.
Install:
- sys.excepthook → write full traceback to tui_gateway_crash.log +
echo the first line to stderr (which the TUI pumps into
Activity as a gateway.stderr event). Chains to the default hook so
the process still terminates.
- threading.excepthook → same, tagged with the thread name so it's
clear when the crash came from a daemon thread (beep playback, TTS,
silence callback, etc.).
- Turn-dispatcher except block now also appends a traceback to the
crash log before emitting the user-visible error event — str(e)
alone was too terse to identify where in the voice pipeline the
failure happened.
Zero behavioural change on the happy path; purely forensics.
TTS feedback loop (hermes_cli/voice.py)
The VAD loop kept the microphone live while speak_text played the
agent's reply over the speakers, so the reply itself was picked up,
transcribed, and submitted — the agent then replied to its own echo
("Ha, looks like we're in a loop").
Ported cli.py:_voice_tts_done synchronisation:
- _tts_playing: threading.Event (initially set = "not playing").
- speak_text cancels the active recorder before opening the speakers,
clears _tts_playing, and on exit waits 300 ms before re-starting the
recorder — long enough for the OS audio device to settle so afplay
and sounddevice don't race for it.
- _continuous_on_silence now waits on _tts_playing (up to 60 s) before
re-arming the mic with another 300 ms gap, mirroring
cli.py:10619-10621. If the user flips voice off during the wait the
loop exits cleanly instead of fighting for the device.
Without both halves the loop races: if the silence callback fires
before TTS starts it re-arms immediately; if TTS is already playing
the pause-and-resume path catches it.
Red REC badge (ui-tui appChrome + useMainApp)
Classic CLI (cli.py:_get_voice_status_fragments) renders "● REC" in
red and "◉ STT" in amber. TUI was showing a dim "REC" with no dot,
making it hard to spot at a glance. voiceLabel now emits the same
glyphs and appChrome colours them via t.color.error / t.color.warn,
falling back to dim for the idle label.
Three issues surfaced during end-to-end testing of the CLI-parity voice
loop and are fixed together because they all blocked "speak → agent
responds → TTS reads it back" from working at all:
1. Wrong result key (hermes_cli/voice.py)
transcribe_recording() returns {"success": bool, "transcript": str},
matching cli.py:_voice_stop_and_transcribe. The wrapper was reading
result.get("text"), which is None, so every successful Groq / local
STT response was thrown away and the 3-strikes halt fired after
three silent-looking cycles. Fixed by reading "transcript" and also
honouring "success" like the CLI does. Updated the loop simulation
tests to return the correct shape.
2. TTS speak-back was missing (tui_gateway/server.py + hermes_cli/voice.py)
The TUI had a voice.toggle "tts" subcommand but nothing downstream
actually read the flag — agent replies never spoke. Mirrored
cli.py:8747-8754's dispatch: on message.complete with status ==
"complete", if _voice_tts_enabled() is true, spawn a daemon thread
running speak_text(response). Rewrote speak_text as a full port of
cli.py:_voice_speak_response — same markdown-strip regex pipeline
(code blocks, links, bold/italic, inline code, headers, list bullets,
horizontal rules, excessive newlines), same 4000-char cap, same
explicit mp3 output path, same MP3-over-OGG playback choice (afplay
misbehaves on OGG), same cleanup of both extensions. Keeps TUI TTS
audible output byte-for-byte identical to the classic CLI.
3. Auto-submit swallowed on non-empty composer (createGatewayEventHandler.ts)
The voice.transcript handler branched on prev input via a setInput
updater and fired submitRef.current inside the updater when prev was
empty. React strict mode double-invokes state updaters, which would
queue the submit twice; and when the composer had any content the
transcript was merely appended — the agent never saw it. CLI
_pending_input.put(transcript) unconditionally feeds the transcript
as the next turn, so match that: always clear the composer and
setTimeout(() => submitRef.current(text), 0) outside any updater.
Side effect can't run twice this way, and a half-typed draft on the
rare occasion is a fair trade vs. silently dropping the turn.
Also added peak_rms to the rec.stop debug line so "recording too quiet"
is diagnosable at a glance when HERMES_VOICE_DEBUG=1.
The TUI had drifted from the CLI's voice model in two ways:
- /voice on was lighting up the microphone immediately and Ctrl+B was
interpreted as a mode toggle. The CLI separates the two: /voice on
just flips the umbrella bit, recording only starts once the user
presses Ctrl+B, which also sets _voice_continuous so the VAD loop
auto-restarts until the user presses Ctrl+B again or three silent
cycles pass.
- /voice tts was missing entirely, so users couldn't turn agent reply
speech on/off from inside the TUI.
This commit brings the TUI to parity.
Python
- hermes_cli/voice.py: continuous-mode API (start_continuous,
stop_continuous, is_continuous_active) layered on the existing PTT
wrappers. The silence callback transcribes, fires on_transcript,
tracks consecutive no-speech cycles, and auto-restarts — mirroring
cli.py:_voice_stop_and_transcribe + _restart_recording.
- tui_gateway/server.py:
- voice.toggle now supports on / off / tts / status. The umbrella
bit lives in HERMES_VOICE + display.voice_enabled; tts lives in
HERMES_VOICE_TTS + display.voice_tts. /voice off also tears down
any active continuous loop so a toggle-off really releases the
microphone.
- voice.record start/stop now drives start_continuous/stop_continuous.
start is refused with a clear error when the mode is off, matching
cli.py:handle_voice_record's early return on `not _voice_mode`.
- New voice.transcript / voice.status events emit through
_voice_emit (remembers the sid that last enabled the mode so
events land in the right session).
TypeScript
- gatewayTypes.ts: voice.status + voice.transcript event
discriminants; VoiceToggleResponse gains tts; VoiceRecordResponse
gains status for the new "started/stopped" responses.
- interfaces.ts: GatewayEventHandlerContext gains composer.setInput +
submission.submitRef + voice.{setRecording, setProcessing,
setVoiceEnabled}; InputHandlerContext.voice gains enabled +
setVoiceEnabled for the mode-aware Ctrl+B handler.
- createGatewayEventHandler.ts: voice.status drives REC/STT badges;
voice.transcript auto-submits when the composer is empty (CLI
_pending_input.put parity) and appends when a draft is in flight.
no_speech_limit flips voice off + sys line.
- useInputHandlers.ts: Ctrl+B now calls voice.record (start/stop),
not voice.toggle, and nudges the user with a sys line when the
mode is off instead of silently flipping it on.
- useMainApp.ts: wires the new event-handler context fields.
- slash/commands/session.ts: /voice handles on / off / tts / status
with CLI-matching output ("voice: mode on · tts off").
Backward compat preserved for voice.record (was always PTT shape;
gateway still honours start/stop with mode-gating added).
tui_gateway/server.py:3486/3491/3509 imports start_recording,
stop_and_transcribe, and speak_text from hermes_cli.voice, but the
module never existed (not in git history — never shipped, never
deleted). Every voice.record / voice.tts RPC call hit the ImportError
branch and the TUI surfaced it as "voice module not available — install
audio dependencies" even on boxes with sounddevice / faster-whisper /
numpy installed.
Adds a thin wrapper on top of tools.voice_mode (recording +
transcription) and tools.tts_tool (text-to-speech):
- start_recording() — idempotent; stores the active AudioRecorder in a
module-global guarded by a Lock so repeat Ctrl+B presses don't fight
over the mic.
- stop_and_transcribe() — returns None for no-op / no-speech /
Whisper-hallucination cases so the TUI's existing "no speech detected"
path keeps working unchanged.
- speak_text(text) — lazily imports tts_tool (optional provider SDKs
stay unloaded until the first /voice tts call), parses the tool's
JSON result, and plays the audio via play_audio_file.
Paired with the Ctrl+B keybinding fix in the prior commit, the TUI
voice pipeline now works end-to-end for the first time.
When the user runs /voice and then presses Ctrl+B in the TUI, three
handlers collaborate to consume the chord and none of them dispatch
voice.record:
- isAction() is platform-aware — on macOS it requires Cmd (meta/super),
so Ctrl+B fails the match in useInputHandlers and never triggers
voiceStart/voiceStop.
- TextInput's Ctrl+B pass-through list doesn't include 'b', so the
keystroke falls through to the wordMod backward-word branch on Linux
and to the printable-char insertion branch on macOS — the latter is
exactly what timmie reported ("enters a b into the tui").
- /voice emits "voice: on" with no hint, so the user has no way to
know Ctrl+B is the recording toggle.
Introduces isVoiceToggleKey(key, ch) in lib/platform.ts that matches
raw Ctrl+B on every platform (mirrors tips.py and config.yaml's
voice.record_key default) and additionally accepts Cmd+B on macOS so
existing muscle memory keeps working. Wires it into useInputHandlers,
adds Ctrl+B to TextInput's pass-through list so the global handler
actually receives the chord, and appends "press Ctrl+B to record" to
the /voice on message.
Empirically verified with hermes --tui: Ctrl+B no longer leaks 'b'
into the composer and now dispatches the voice.record RPC (the
downstream ImportError for hermes_cli.voice is a separate upstream
bug — follow-up patch).
The 300s default was too tight for high-reasoning models on non-trivial
delegated tasks — e.g. gpt-5.5 xhigh reviewing 12 files would burn >5min
on reasoning tokens before issuing its first tool call, tripping the
hard wall-clock timeout with 0 api_calls logged.
- tools/delegate_tool.py: DEFAULT_CHILD_TIMEOUT 300 -> 600
- hermes_cli/config.py: surface delegation.child_timeout_seconds in
DEFAULT_CONFIG so it's discoverable (previously the key was read by
_get_child_timeout() but absent from the default config schema)
Users can still override via config.yaml delegation.child_timeout_seconds
or DELEGATION_CHILD_TIMEOUT_SECONDS env var (floor 30s, no ceiling).
Fixes a broader class of 'tools.function.parameters is not a valid
moonshot flavored json schema' errors on Nous / OpenRouter aggregators
routing to moonshotai/kimi-k2.6 with MCP tools loaded.
## Moonshot sanitizer (agent/moonshot_schema.py, new)
Model-name-routed (not base-URL-routed) so Nous / OpenRouter users are
covered alongside api.moonshot.ai. Applied in
ChatCompletionsTransport.build_kwargs when is_moonshot_model(model).
Two repairs:
1. Fill missing 'type' on every property / items / anyOf-child schema
node (structural walk — only schema-position dicts are touched, not
container maps like properties/$defs).
2. Strip 'type' at anyOf parents; Moonshot rejects it.
## MCP normalizer hardened (tools/mcp_tool.py)
Draft-07 $ref rewrite from PR #14802 now also does:
- coerce missing / null 'type' on object-shaped nodes (salvages #4897)
- prune 'required' arrays to names that exist in 'properties'
(salvages #4651; Gemini 400s on dangling required)
- apply recursively, not just top-level
These repairs are provider-agnostic so the same MCP schema is valid on
OpenAI, Anthropic, Gemini, and Moonshot in one pass.
## Crash fix: safe getattr for Tool.inputSchema
_convert_mcp_schema now uses getattr(t, 'inputSchema', None) so MCP
servers whose Tool objects omit the attribute entirely no longer abort
registration (salvages #3882).
## Validation
- tests/agent/test_moonshot_schema.py: 27 new tests (model detection,
missing-type fill, anyOf-parent strip, non-mutation, real-world MCP
shape)
- tests/tools/test_mcp_tool.py: 7 new tests (missing / null type,
required pruning, nested repair, safe getattr)
- tests/agent/transports/test_chat_completions.py: 2 new integration
tests (Moonshot route sanitizes, non-Moonshot route doesn't)
- Targeted suite: 49 passed
- E2E via execute_code with a realistic MCP tool carrying all three
Moonshot rejection modes + dangling required + draft-07 refs:
sanitizer produces a schema valid on Moonshot and Gemini
Cron now resolves its toolset from the same per-platform config the
gateway uses — `_get_platform_tools(cfg, 'cron')` — instead of blindly
loading every default toolset. Existing cron jobs without a per-job
override automatically lose `moa`, `homeassistant`, and `rl` (the
`_DEFAULT_OFF_TOOLSETS` set), which stops the "surprise $4.63
mixture_of_agents run" class of bug (Norbert, Discord).
Precedence inside `run_job`:
1. per-job `enabled_toolsets` (PR #14767 / #6130) — wins if set
2. `_get_platform_tools(cfg, 'cron')` — new, the blanket gate
3. `None` fallback (legacy) — only on resolver exception
Changes:
- hermes_cli/platforms.py: register 'cron' with default_toolset
'hermes-cron'
- toolsets.py: add 'hermes-cron' toolset (mirrors 'hermes-cli';
`_get_platform_tools` then filters via `_DEFAULT_OFF_TOOLSETS`)
- cron/scheduler.py: add `_resolve_cron_enabled_toolsets(job, cfg)`,
call it at the `AIAgent(...)` kwargs site
- tests/cron/test_scheduler.py: replace the 'None when not set' test
(outdated contract) with an invariant ('moa not in default cron
toolset') + new per-job-wins precedence test
- tests/hermes_cli/test_tools_config.py: mark 'cron' as non-messaging
in the gateway-toolset-coverage test
Themes and plugins can now pull off arbitrary dashboard reskins (cockpit
HUD, retro terminal, etc.) without touching core code.
Themes gain four new fields:
- layoutVariant: standard | cockpit | tiled — shell layout selector
- assets: {bg, hero, logo, crest, sidebar, header, custom: {...}} —
artwork URLs exposed as --theme-asset-* CSS vars
- customCSS: raw CSS injected as a scoped <style> tag on theme apply
(32 KiB cap, cleaned up on theme switch)
- componentStyles: per-component CSS-var overrides (clipPath,
borderImage, background, boxShadow, ...) for card/header/sidebar/
backdrop/tab/progress/badge/footer/page
Plugin manifests gain three new fields:
- tab.override: replaces a built-in route instead of adding a tab
- tab.hidden: register component + slots without adding a nav entry
- slots: declares shell slots the plugin populates
10 named shell slots: backdrop, header-left/right/banner, sidebar,
pre-main, post-main, footer-left/right, overlay. Plugins register via
window.__HERMES_PLUGINS__.registerSlot(name, slot, Component). A
<PluginSlot> React helper is exported on the plugin SDK.
Ships a full demo at plugins/strike-freedom-cockpit/ — theme YAML +
slot-only plugin that reproduces a Gundam cockpit dashboard: MS-STATUS
sidebar with live telemetry, COMPASS crest in header, notched card
corners via componentStyles, scanline overlay via customCSS, gold/cyan
palette, Orbitron typography.
Validation:
- 15 new tests in test_web_server.py covering every extended field
- tests/hermes_cli/: 2615 passed (3 pre-existing unrelated failures)
- tsc -b --noEmit: clean
- vite build: 418 kB bundle, ~2 kB delta for slots/theme extensions
Co-authored-by: Teknium <p@nousresearch.com>
- AUTHOR_MAP entry for 130918800+devorun for #6636 attribution
- test_moa_defaults: was a change-detector tied to the exact frontier
model list — flips red every OpenRouter churn. Rewritten as an
invariant (non-empty, valid vendor/model slugs).
