Summary:
- Reduce active-only diagnostic liveness noise by emitting transient event-loop max delay samples as info-level telemetry.
- Keep warnings for queued or waiting work and for sustained high P99 loop delay.
- Cover the active-only path in the diagnostic stability tests and changelog.
Verification:
- pnpm format:check src/logging/diagnostic-stability.ts src/logging/diagnostic.test.ts CHANGELOG.md
- pnpm test src/logging/diagnostic.test.ts
- pnpm check:changed
- GitHub PR checks passed on head 25e674fe41.
fix(gateway): remove unnecessary type assertion in buildAgentPrompt
fix(gateway): reject unsupported forced tool_choice modes
tool_choice=required and named function tool_choice are now rejected
with invalid_request_error until hard enforcement is implemented
at the agent runtime layer. Only auto and none remain supported.
docs: update Chat Completions tool_choice contract to match rejection behavior
Only auto and none are currently accepted; required and named
function tool_choice are rejected until hard enforcement exists.
The remaining whole-file transcript scans flagged by ClawSweeper triage on
splitting on newlines. On long-running sessions where transcripts grow into
the multi-MB / 100s-of-MB range that scales peak RSS with file size and is
the practical OOM risk in the report.
Add a shared streaming helper module `src/config/sessions/transcript-stream.ts`
exposing:
- `streamSessionTranscriptLines(filePath, { signal? })`: forward async-iterator
over trimmed non-empty lines using `fs.createReadStream` + `readline` with
`crlfDelay: Infinity`. Bounded to one line of memory at a time and honours
an abort signal between lines.
- `readSessionTranscriptTailLines(filePath, { maxBytes? })`: tail-only read of
the last `maxBytes` of a file (default 4 MiB, clamped to [1 KiB, 64 MiB]),
returning trimmed non-empty lines in reverse order. Drops the leading line
of the slice when the window does not start at byte zero so callers never
see a partial-line suffix.
Migrate every flagged whole-file scan to these helpers while preserving the
malformed-line tolerance and idempotency-key return semantics callers depend
on (see `Remaining risk / open question` on the issue):
- `src/config/sessions/transcript.ts`: `readLatestAssistantTextFromSessionTranscript`,
`readTailAssistantTextFromSessionTranscript`, and the delivery-mirror dedupe
helper `findLatestEquivalentAssistantMessageId` now use the tail helper;
`transcriptHasIdempotencyKey` uses the forward stream helper.
- `src/gateway/server-methods/chat.ts`: the inline `transcriptHasIdempotencyKey`
used by chat-method append idempotency now uses the forward stream helper
and tolerates malformed lines mid-scan (matching the sibling helper in
`config/sessions/transcript.ts`).
- `src/gateway/session-compaction-checkpoints.ts`: `readTranscriptEntriesForForkAsync`
builds the fork entry array from the forward stream helper instead of one
big `fileHandle.readFile("utf-8")` call.
Fixes#54296.
* fix(doctor): warn when per-agent model omits fallbacks key and defaults chain is non-empty
`resolveAgentModelFallbacksOverride` in `src/agents/agent-scope.ts` returns
`[]` (no fallbacks) when a per-agent model is configured without an explicit
`fallbacks` key. At runtime this silently clobbers
`agents.defaults.model.fallbacks`, leaving the agent with no fallbacks.
Two config patterns hit this:
1. String form: `"model": "openai/gpt-5.5"` — user likely means "use this model,
inherit fallbacks from defaults".
2. Object without `fallbacks` key: `"model": { "primary": "openai/gpt-5.5" }` —
user likely means "just set the primary, keep defaults fallbacks".
The only explicit "no fallbacks" signal is `fallbacks: []`. This change adds
`collectImplicitFallbackClobberWarnings` / `noteImplicitFallbackClobberWarnings`
to the doctor config-analysis flow, which warns on either ambiguous shape only
when `agents.defaults.model.fallbacks` is non-empty (so there is something to
clobber).
No semantic change to how fallbacks resolve; this is a doctor-only diagnostic.
Closes#79369. Extends the scope of closed PR #79389 (string-form only) to
also cover the object-without-fallbacks case.
* fix(doctor): mirror runtime model primary normalization
* fix(doctor): strengthen fallback warning proof
* fix(doctor): tolerate malformed agent lists
* fix(doctor): type guarded agent runtime policy
* fix(ui): sync quick settings i18n baseline
---------
Co-authored-by: Peter Steinberger <steipete@gmail.com>
