* fix(sse): parse Gemini 429 RetryInfo.retryDelay for model lockout (#7940)
Gemini free-tier 429 bodies carry a short explicit retry hint --
error.details[].{"@type": google.rpc.RetryInfo, retryDelay: "26s"} plus a
"Please retry in Ns." message -- but parseRetryFromErrorText only matched
'reset after'/'will reset after' text, so quotaResetHintMs came back null and
recordModelLockoutFailure fell back to getMsUntilTomorrow() for
quota_exhausted, locking the model out for ~19h instead of ~26s.
parseRetryHintFromJsonBody (retryAfterJson.ts) now walks error.details[] for
a google.rpc.RetryInfo entry and parses its retryDelay via a shared
parseDelayString helper (moved out of accountFallback.ts so
parseRetryAfterFromBody and the model-lockout path use the same grammar).
parseRetryFromErrorText also gained a 'please retry in Ns' text fallback for
bodies without a parseable details[] array. Both new paths are capped by a
dedicated MAX_SHORT_RETRY_HINT_MS (24h), independent of the existing 30-day
MAX_PROVIDER_COOLDOWN_MS, since RetryInfo/please-retry-in are short
throttling hints, not long-lived quota resets like Antigravity's 160h.
Regression test: tests/unit/bug-7940-gemini-retrydelay.test.ts (RED before
the fix: parseRetryFromErrorText returned null and the resulting lockout was
~19h; GREEN after: ~26s).
* chore(quality): register bug-7940-gemini-retrydelay test in stryker tap.testFiles
* fix(sse): preserve parallel_tool_calls for GPT-5.6 ultra/max delegation under Codex Responses Lite (#7821)
* fix(codex): drop over-broad parallel_tool_calls allowlist entry — keep #2608 stripping intact (#7821)
The static RESPONSES_API_ALLOWLIST addition made parallel_tool_calls survive for
ALL models, breaking the #2608 non-passthrough stripping guarantee for gpt-5.5.
The real #7821 fix (isCodexDelegationDependentModel gating in
enforceCodexResponsesLiteParallelToolCalls) is model/effort-scoped and does not
need the allowlist entry — native Codex traffic returns before the allowlist runs.
* fix(api): classify /api/acp/agents as loopback-only (#7948)
/api/acp/agents is spawn-capable (POST registers a client-chosen `binary`
via src/lib/acp/registry.ts; GET / POST {action:"refresh"} runs
detectInstalledAgents() -> execFileSync(probe.command, probe.args, { shell })
transitively) but was never added to LOCAL_ONLY_API_PREFIXES in
src/server/authz/routeGuard.ts, unlike every sibling spawn-capable route
(Hard Rules #15/#17). A leaked JWT via tunnel could reach the version-probe
execFileSync path.
Adds the prefix to the loopback gate plus a direct regression test
(tests/unit/route-guard-acp-agents-local-only.test.ts) and an assertion in
tests/unit/check-route-guard-membership.test.ts documenting why the existing
source-scan subcheck cannot catch this class of gap (the spawn call is
transitive via registry.ts, not in the route file itself).
* chore(quality): register route-guard-acp-agents-local-only test in stryker tap.testFiles
The nightly-compat Node 24/26 shard failures were 17/18 base-red debt already
fixed forward on the release tip (same pattern as #7025/#7140/#7675/#7742,
tracked in #6949). The one genuine still-reproducing failure was a test-harness
timing-margin bug: ratelimit-admission-control-6593's pollUntil helper used a
fixed 2000ms deadline that races Bottleneck's QUEUED->EXECUTING event-loop-tick
transition under CI/devbox contention. Widened the default deadline to 10000ms;
the admission logic itself (checkQueueAdmission / RATE_LIMIT_QUEUE_FULL / 429) is
unchanged and covered by 3 passing pure-unit tests.
createSqliteNativeError() in bin/cli/sqlite.mjs only recognized the ABI-mismatch
native error class (NODE_MODULE_VERSION/ERR_DLOPEN_FAILED). It silently passed through
the 'Could not locate the bindings file' class thrown by the bindings package when the
better-sqlite3 native addon was never built/downloaded - the exact case hit by
'npx omniroute reset-password', since npx runs a fresh ephemeral install that never
builds the addon. Users got a raw multi-line path dump instead of guidance.
Widen the condition to also match 'Could not locate the bindings file',
MODULE_NOT_FOUND, and "Cannot find module 'better-sqlite3'", and point users at the
existing self-heal command `omniroute runtime repair`, same as the ABI-mismatch branch
already does.
Regression test: tests/unit/cli-sqlite-bindings-not-found-7868.test.ts - RED against the
reporter's exact error text before the fix, GREEN after.
#7858 — accumulateBotContent() silently returned an empty delta for any
unrecognized frame shape, and finish() only had a fallback for the
type:2 finalResultMessage case; a turn with no content in ANY known
shape closed with a bare `stop` + `[DONE]`, indistinguishable from a
genuine empty answer. finish() now emits a sanitized error (Hard Rule
#12) naming the resolved tier and the likely causes, and unrecognized
update-frame shapes are logged by argument KEY only (never content,
tokens, or cookies).
#7870 — the enterprise tier only changed buildWsUrl() query params;
buildChatInvocation() always fell back to the consumer
M365_DEFAULT_OPTION_SETS (which declares the MSA-only enable_msa_user
flag) and tone:"". resolveConnectionParams()/resolveTierOverrides() now
also resolve and surface the tier itself, threaded through wsChat() ->
sendChat() -> buildChatInvocation() via a new
resolveChatInvocationOverrides() helper, so an enterprise-tier
invocation declares the enterprise_*/bizchat_* option sets, the wider
allowedMessageTypes captured from the real enterprise HAR (Discussion
#7850), and tone:"Magic" — while individual and EDU payloads stay
byte-identical to today.
Regression tests: tests/unit/copilot-m365-web-silent-empty-7858.test.ts,
tests/unit/copilot-m365-enterprise-invocation-7870.test.ts.
Root cause: validate-release-green.mjs's ESLint gate runs
`npx eslint . --format json --suppressions-location ...` without
--pass-on-unpruned-suppressions. ESLint 9.x prints the valid JSON report
to stdout first, then (if the suppressions file has any stale entries)
appends 'There are suppressions left that do not occur anymore...' to
stderr and exits 2. The gate concatenates stdout+stderr, so
parseEslintJson() received valid JSON immediately followed by that
sentence, JSON.parse() threw, and the caller reported the generic
"could not parse eslint json" HARD failure instead of the real
(harmless) unpruned-suppressions housekeeping condition.
Fix: add --pass-on-unpruned-suppressions to the gate's eslint invocation
(unpruned suppressions are release-time housekeeping, not a contributor
defect), and harden parseEslintJson() with a bracket-depth scan so it
recovers the JSON array even when trailing non-JSON text is glued on.
Regression test: tests/unit/validate-release-green.test.ts — 'parseEslintJson
tolerates ESLint's trailing unpruned-suppressions stderr sentence (#7837)',
feeding parseEslintJson() the exact byte shape ESLint's own cli.js produces
in this scenario. Confirmed RED before the fix (parsed === null), GREEN after.
Gates run: node --import tsx/esm --test tests/unit/validate-release-green.test.ts (20/20 pass),
npm run typecheck:core (clean), eslint --suppressions-location on changed files (clean),
check-complexity.mjs + check-cognitive-complexity.mjs (OK, no regression),
check-mutation-test-coverage.mjs --strict (no drift), check-changelog-integrity.mjs (OK),
check-file-size.mjs (OK, no frozen files grown).
Closes#7837
* fix: hide internal reasoning replay placeholder
* fix: hide internal reasoning replay placeholder in OpenAI→Claude translation too
Mirror the isInternalReasoningPlaceholder() guard already applied to the
Responses-API and OpenAI-Responses reasoning paths in
openaiToClaudeResponse() (open-sse/translator/response/openai-to-claude.ts).
The internal reasoning-replay sentinel was still leaking into the Claude
"thinking" content block on this translation path.
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
* test: close both-add merge of #7912 and #7905 reasoning/tool tests
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
---------
Co-authored-by: Diego Rodrigues de Sa e Souza <diegosouza.pw@gmail.com>
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
Rebuilt clean on release/v3.8.49 (branch carried old-main drift) — applies only
the 2 real commits' delta: SSE collector now captures native functionCall parts
in non-streaming, plus the test (typed emptyCollected() as AntigravityCollectedStream).
Co-authored-by: Wital <wital@example.com>
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
* fix(combo): strip boolean reasoning field for opencode-go providers
opencode-go backed providers (ollama-cloud, opencode-go, opencode,
opencode-zen) use a Go ChatCompletionRequest struct where the reasoning
field is typed as openai.Reasoning (a structured type). When a client
sends reasoning: true or reasoning: false — valid per the OpenAI API —
the Go JSON decoder rejects it with:
400: json: cannot unmarshal bool into Go struct field
ChatCompletionRequest.reasoning of type openai.Reasoning
This strips the boolean reasoning field before forwarding to these
providers, allowing the upstream to apply its own default reasoning
behavior. Object/string forms are left untouched.
Observed in production: 3 consecutive 400 errors from ollama-cloud/glm-5.2
in a 30-second window, each with the unmarshal error.
* fix(opencode): add null/primitive guard in stripBooleanReasoning
Adds defensive check for null, undefined, and non-object inputs
as suggested in review. Added unit test coverage for these edge cases.
monaco-editor 0.56.0 (bumped in #7897) ships a restrictive `exports` map
("./*.js" -> "./esm/vs/*.js", "./*" -> "./esm/vs/*.js") that rewrites every
subpath by prepending esm/vs/. The pre-0.56 deep import
`monaco-editor/esm/vs/editor/editor.api` therefore resolved to the doubled,
non-existent `esm/vs/esm/vs/editor/editor.api.js` and broke the production
Turbopack build (Module not found in MonacoEditor.tsx).
Switch to the 0.56-compatible specifier `monaco-editor/editor/editor.api.js`,
which resolves to the same file (esm/vs/editor/editor.api.js) as before.
Adds tests/unit/monaco-editor-import-path.test.ts as a regression guard:
asserts the specifier has no esm/vs/ prefix and resolves to editor.api.js
against the installed monaco 0.56.
Rebuilt clean on release/v3.8.49 (branch forked from old main, ~drift). Resolved
2 real conflicts against the current tip: providerModelsConfig.ts keeps BOTH the
tip's DashScope text-model helpers (#7882) and this PR's ProviderModelsHeaderContext
type; OAuthModal.tsx takes this PR's DEVICE_CODE_PROVIDERS set (superset of the
tip's hardcoded chain + grok-cli), dropping the now-dead qwen entry (#7866 removed
qwen OAuth).
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
#7843's test reorganization removed the 2 no-explicit-any usages that this
suppression covered, leaving a stale entry that fails lint:json --max-warnings 0
(exit 2, 'suppressions left that do not occur anymore') on the release tip for
every fresh PR run. Pruning tightens the gate — no rebaseline.
* feat(qwen): add Qwen3.8 Max Preview catalogs [Part 2/3]
Rebuilt clean on release/v3.8.49 after Part 1 (#7866) squash-merged — applies
only the Part-2 delta (Qwen Web / Qoder qwen3.8-max-preview registration +
required-thinking allowlist + Qoder client rework) onto the current tip. No
migration in this part (that was Part 1).
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
* feat(qwen): add regional Alibaba and Qwen Cloud providers [Part 3/3]
Rebuilt clean on top of Part 2 (#7874) over the current release tip — applies
only the Part-3 delta (alibaba Model Studio, Alibaba Token Plan, qwen-cloud,
qwen-cloud-token-plan with region selector). No migration in this part.
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
---------
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
Rebuilt clean on release/v3.8.49 after Part 1 (#7866) squash-merged — applies
only the Part-2 delta (Qwen Web / Qoder qwen3.8-max-preview registration +
required-thinking allowlist + Qoder client rework) onto the current tip. No
migration in this part (that was Part 1).
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
* Create WEB-COOKIE-GUIDE.md for Web Cookie providers
Added a comprehensive guide for using Web Cookie providers with OmniRoute, including setup instructions, credential formats, limitations, and troubleshooting tips.
* Fix formatting in WEB-COOKIE-GUIDE.md
* Add Web Cookie Providers section to providers guide
Added section for Web Cookie Providers with a reference to the WEB-COOKIE-GUIDE.md.
* Enhance CLAUDE_WEB.md with user guidance
Added introductory information and guidance for new users of the Claude Web provider.
Agent clients often send message.content as [{type:\"text\",text:\"...\"}]
instead of a plain string. buildNotionMessageStep previously required a
string and silently dropped those turns, so system injects (jailbreak /
agentic conversion) and multimodal user messages never reached Notion.
Normalize string | content-parts | bare string parts via
extractNotionMessageText, and add regression coverage in the transcript
unit suite.
#7251 added four usage.* quota-visibility keys to en.json without the Vietnamese
counterparts, leaving __MISSING__ markers that fail i18n-vi-completeness on the
release tip for every fresh PR run. Translated with the locale's existing quota
vocabulary (hạn mức).
* feat(dashboard): add per-operator quota row visibility on usage tab
Adds a "hide this quota row" action (visibility_off icon button) to
each model-quota row on the provider limits card, and a "Hidden: …"
strip at the bottom of the card to restore any hidden row. The
visibility preference is keyed per-provider (settings.quotaVisibility)
and persisted via PATCH /api/settings, so it survives refresh/reload.
Distinct from the existing model-catalog isHidden/isDeleted mechanism
(collectHiddenQuotaModelIds/filterHiddenModelQuotas in
ProviderLimits/utils.tsx), which hides rows the ADMIN marked hidden in
the model catalog. This is a personal dashboard preference an operator
can toggle without touching the catalog — e.g. temporarily decluttering
a quota card with many low-signal rows.
New pure helpers (getQuotaVisibilityKey, filterQuotasByVisibility,
getHiddenQuotaRows) live in ProviderLimits/utils.tsx and are unit
tested directly; the settings key is added to DEFAULT_SETTINGS and
validated via updateSettingsSchema like the existing
providerStrategies field. New UI strings are filled in en.json and
synced to the other 42 locales as `__MISSING__` placeholders.
Co-authored-by: nguyenha935 <208228297+nguyenha935@users.noreply.github.com>
Inspired-by: https://github.com/decolua/9router/pull/2371
* chore(changelog): fragment for #7251
* refactor(usage): extract useQuotaVisibility hook (file-size budget)
ProviderLimits/index.tsx grew past its frozen 1127-line budget with the
quota-visibility wiring; extract the state + persistence + hide/show handlers
into a dedicated hook (same pattern as useCodexResetCreditRedemption).
No behavior change — the live validation evidence on the PR covers this flow
end-to-end.
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
* refactor(dashboard): shed QuotaCard complexity to baseline + type useQuotaVisibility settings fetch (#7251 CI green)
* test(dashboard): rename quota-row visibility test (path claimed by #7360)
* fix(dashboard): restore providerTierField case-collision fix dropped by merge auto-resolve
The git merge of origin/release/v3.8.49 auto-resolved providerTierFieldApi.ts
(added post-merge-base by commit 3ba3cd145e to fix a case-only filename
collision with ProviderTierField.tsx that breaks webpack on case-insensitive
filesystems) back down to its pre-fix name providerTierField.ts, silently
reverting that base-red fix and dropping its stryker tap.testFiles
registration and changelog fragment. Restore all four: the renamed helper
file, the import path in ProviderTierField.tsx, the import path in
tests/unit/provider-tier-field-helpers.test.ts, and the
combo-skip-conn-disable-plugin-block-7806.test.ts stryker registration.
---------
Co-authored-by: nguyenha935 <208228297+nguyenha935@users.noreply.github.com>
* fix: reserve chat admission before parsing
* fix: release admission on handler failure
* fix: reconcile chat admission with the #7862 parse-once route
Post-#7862 rebase: the admission-rebuilt request is json()-parsed directly over
the already-buffered bytes — the clone()+json() pair is gone, and the
parse-once regression tests now pin the post-admission contract (original
request: zero json() calls, zero clone() calls; the single materialization is
admitChatRequest()'s bounded byte reader).
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
---------
Co-authored-by: Ravi Tharuma <RaviTharuma@users.noreply.github.com>
Co-authored-by: Diego Rodrigues de Sa e Souza <diegosouza.pw@gmail.com>
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
* feat(sse): add live gRPC-web quota fetcher for grok-cli (#6844)
* fix(sse): send gRPC-web request frame + decode real GetGrokCreditsConfig schema
Live validation against grok.com (real bearer token, tier-4 account) proved
the #6844 grok-cli quota fetcher was a silent no-op:
- The POST to GetGrokCreditsConfig had no body. gRPC-web requires a request
frame even for a no-argument RPC; without one the upstream returns
`grpc-status: 13 "Missing request message."` with a 0-byte response.
Fixed by sending the empty gRPC-web frame (flag 0x00 + 4-byte length 0).
- The decoder's field mapping (top-level field 1 = double percent, field 2 =
string resetAt) was reverse-engineered from a third-party doc and never
matched the real response. The real shape is: top-level field 1 is a
NESTED message whose subfield 1 is a fixed32 float usage ratio (0..1) and
subfield 5 is a Timestamp{seconds,nanos} reset time. grokCliQuotaFrame.ts
now decodes that nested shape; grokCliQuotaFetcher.ts's buildQuota()
rescales the decoder's 0-100 percentUsed back to the 0-1 fraction the rest
of the quota pipeline expects (quotaPreflight.ts::remainingPercentFrom).
- The response's 2nd gRPC-web frame (trailer, flag 0x80) is now explicitly
walked-and-skipped instead of relying on incidental length-bounding.
Test fixtures in both files now encode the real captured wire structure
(nested message, fixed32 ratio, Timestamp reset, trailer frame) instead of
the old synthetic fixed64-double buffers, and the stale "Cloudflare
non-blocking is an assumption" comment is corrected to reflect that it is
now live-validated.
Reject repair and Remove CA when no sudo password is supplied or cached,
instead of spawning sudo -S with an empty string. Add a password modal to
AgentBridgeMaintenanceCard and regression tests for the 400 gate.
Address PR review feedback: reject whitespace-only sudoPassword values,
avoid caching unusable passwords, extract closePasswordModal helper, and
replace the route test that invoked real sudo with pure gate assertions.
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
* fix(notion-web): add browser fingerprint headers to reduce Cloudflare challenges
Adds sec-ch-ua, sec-fetch-*, cache-control, pragma, and priority headers
that real Chromium browsers send. Without these, Cloudflare may challenge
or block requests that look like non-browser clients.
Applied to:
- buildNotionExecuteHeaders (inference requests)
- buildNotionBrowserHeaders (workspace discovery)
- buildNotionModelsDiscoveryHeaders (model discovery)
Headers match the real browser capture from Chrome 149 on Linux.
Addresses gemini-code-assist review:
- Fixed platform mismatch: sec-ch-ua-platform now matches USER_AGENT (Windows)
- Deduplicated headers via shared BROWSER_HEADERS constant in notionWebModels.ts
- Both executor and model discovery use the same constant
* fix(notion-web): align Chrome version to 149 and add browser header tests
Addresses maintainer review feedback on #7864:
- Align User-Agent and NOTION_USER_AGENT to Chrome/149 (was 145 and 150)
matching sec-ch-ua already declaring v="149"
- Add test assertions that browser fingerprint headers (sec-ch-ua,
sec-fetch-mode, cache-control, pragma) are sent on both executor
and models-discovery requests
* refactor(providers): extract notion-web fallback catalog to its own module
notionWebModels.ts crossed the 800-line new-file cap (875) once the browser
header tests landed; move the NOTION_WEB_FALLBACK_MODELS catalog + its type to
notionWebFallbackModels.ts (pure data, re-exported for existing consumers).
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
---------
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
Rebuilt clean on release/v3.8.49 (branch forked from an old main and carried
~68 files of base drift). Reconciled with #7237 on the tip: supportsVision
keeps the authoritative getResolvedModelCapabilities() resolution (the PR's
isVisionModelId heuristic predates that fix); the PR's real change lands —
OAuth 'claude' now counts as a direct Anthropic transport for OmniGlyph, and
claude-fable-5 joins the vision model ids. Plumbing test now pins the
'|| provider === "claude"' invariant instead of exact formatting.
Co-authored-by: quanturbo <168349709+quanturbo@users.noreply.github.com>
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
* fix(vision-bridge): reroute auto/ prefix to vision model when images present
Rebuilt clean on release/v3.8.49 (the original branch forked from an old main
and dragged ~70 unrelated files of base drift). Reconciled with the newer
VibeProxy credential guards on the tip: the reroute now also fires for auto/
models, the keep-credentialed-model skip does not apply to auto (keeping auto
would land on a text-only candidate), and the reroute-target credential guard
is preserved.
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
* test(guardrails): compact image_url literals to fit the 800-line test cap
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
---------
Co-authored-by: herjarsa <204746071+herjarsa@users.noreply.github.com>
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
* fix(usage): harden account identity reconciliation
* fix(db): renumber codex strong-identity migration 128 -> 129
release/v3.8.49 tip took slot 128 via #7839 (auto_candidate_overrides) after
this branch forked; renumber the new migration and its test references.
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
---------
Co-authored-by: Diego Rodrigues de Sa e Souza <diegosouza.pw@gmail.com>
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
* fix(sse): convert OpenAI media parts to Claude blocks in the CC bridge
OpenAI-format clients (OpenCode/Kilo/Cline) reach the Claude-Code-compatible
bridge untranslated: chatCore skips the OpenAI->Claude translator when
sourceFormat is OPENAI, so image_url / AI-SDK image / file parts either went
upstream in OpenAI shape (silently ignored) or were dropped by the text-only
extraction, and media-only user turns were removed by hasValidContent().
- claudeCodeCompatible: convertOpenAiMediaBlock() converts image_url
(base64 + remote), AI-SDK string image and file parts (pdf->document,
image mime->image) to Claude blocks in both bridge paths; Claude-native
blocks pass through unchanged and the text-only wire image is preserved.
- claudeHelper: hasValidContent() now counts image/document blocks so
media-only user turns are not silently deleted.
Reported-by: beingshafin
Refs #7777
* refactor(sse): extract CC media-block conversion to ccOpenAiMediaBlocks.ts
claudeCodeCompatible.ts is frozen at 1202 lines by check:file-size; the #7777
helpers pushed it to 1291. Move them to a dedicated module, no behavior change.
* test(quality): register cc-bridge-openai-image-7777 test in stryker tap.testFiles
* feat(providers): add OpenRouter speech-to-text (audio transcription) provider
Adds OpenRouter as a speech-to-text provider for POST /v1/audio/transcriptions. Transcription requests route to OpenRouter's dedicated STT endpoint (https://openrouter.ai/api/v1/audio/transcriptions), converting the multipart audio upload into OpenRouter's JSON input_audio { data, format } shape and forwarding optional language, temperature, and response_format fields.
OpenRouter STT reuses your existing OpenRouter connection, so no separate credential is required. Eleven transcription models are available (Deepgram Nova-3, Microsoft MAI-Transcribe 1.5, NVIDIA Parakeet, Mistral Voxtral, Qwen3 ASR, Google Chirp 3, and the OpenAI Whisper / GPT-4o transcribe family).
The media-providers STT playground card now narrows its model picker to transcription-only models, so the OpenRouter chat catalog is not shown for speech-to-text, and the provider page shows an existing-connection note for OpenRouter STT.
* fix(providers): address OpenRouter STT review feedback
- Extract the OpenRouter transcription handler into
open-sse/handlers/openrouterTranscription.ts to keep audioTranscription.ts
under the file-size cap.
- Qualify the STT card's submitted model id with the connection's provider
prefix so OpenRouter models route to OpenRouter rather than the model's own
vendor (the transcription route resolves the provider from the leading
segment of the model id).
- Coerce temperature to a number and forward timestamp_granularities on the
JSON input_audio payload; match the base MIME type when resolving the audio
format so codec parameters (e.g. audio/webm;codecs=opus) do not fall back to wav.
- Split the OpenRouter cases into tests/unit/audio-transcription-openrouter.test.ts
and add coverage for temperature, timestamp granularities, MIME codec params,
and qualified-id routing.
* fix(compression): skip CCR on tool outputs to preserve agent loop
When OmniRoute is used as a chat-completion PROVIDER (not as an MCP server),
the upstream LLM cannot call `omniroute_ccr_retrieve` to expand CCR markers
on demand. Replacing tool outputs with `[CCR retrieve hash=… chars=…]`
placeholders therefore makes the LLM stall — it sees an opaque marker
where the actual tool result should be and has no way to recover the
verbatim content.
Scope:
- OpenAI format: `{ role: "tool", tool_call_id, content }`
- Anthropic format: `{ role: "user", content: [{ type: "tool_result", … }] }`
Fix: extend `processMessages` in the CCR engine to skip both shapes
verbatim. The engine still applies to plain user / assistant text blocks,
which is where compression yields token savings AND the LLM can reason
about the marker.
Tests:
- New `tests/unit/compression/ccr-skip-tool-outputs.test.ts` covers both
formats (4 cases) plus a regression guard that plain user text is still
compressed.
- All 57 pre-existing CCR tests still pass.
Reported-by: herjarsa
Refs: AGENTS.md agent feedback — agent loop stalled on bash/read/grep
outputs after CCR collapsed them to markers.
* fix(compression): guard CCR against null / non-object parts
Apply gemini-code-assist review feedback on PR #7869:
- Use optional chaining when reading `part["type"]` so malformed
client payloads (null entries, non-object entries in the parts
array) cannot throw `TypeError: Cannot read properties of null`.
- The skip rule (this branch) and the existing text-part compression
path (the next branch) both get the guard, since both dereference
`part["type"]` directly.
Test:
- New defensive case: a user message whose content array contains a
`null` entry alongside a `tool_result` must not crash the engine.
Refs: gemini-code-assist review on #7869 (PRR_kwDORPf6ys8AAAABGj5G7Q)
---------
Co-authored-by: herjarsa <herjarsa@users.noreply.github.com>
* fix: avoid cmd.exe spawn on Windows by using os.hostname() before execSync fallback
On Windows, execSync() wraps the command in cmd.exe /d /s /c,
spawning a new cmd.exe process. getMachineIdRaw() called
execSync("hostname") as Strategy 4 before trying os.hostname()
as Strategy 5 -- meaning every dashboard API call spawned an
unnecessary cmd.exe process.
This commit:
- Moves os.hostname() to Strategy 4 (no child process, native binding)
- Keeps execSync("hostname") as Strategy 5 fallback
- Adds module-level caching so getMachineIdRaw() only runs once per
process lifetime since the machine ID never changes at runtime
- Caches all strategy results at the first successful return
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent)
Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>
* fix: avoid cmd.exe spawn on Windows by using os.hostname() before execSync fallback
Prioritize os.hostname() (sync, no subprocess) over execSync hostname fallback. Cache the result so subsequent calls never spawn. Export resetMachineIdCache() for test isolation.
Tests: 10 tests covering cache behavior, strategy fallback order, and consistent machine ID hashing. The 2 tests that mock os.hostname() now also stub fs.readFileSync for /etc/machine-id to throw, so Strategy 3 (Linux machine-id file) does not preempt Strategy 4 on real Linux runners.
---------
Co-authored-by: tientien17 <tientien17@users.noreply.github.com>
Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>
* fix(cli): use rundll32 instead of cmd.exe for Windows browser fallback in dashboard command
Extract resolveOpenCommand(platform, url) as an exported pure function so tests import the actual production code instead of duplicating logic.
The openFallback function in bin/cli/commands/dashboard.mjs used cmd /c start to open the dashboard URL on Windows, spawning an unnecessary cmd.exe process. Replaces with rundll32 url.dll,FileProtocolHandler which opens the URL directly through the Windows shell handler API without any shell wrapper.
Tests: 5 tests importing the actual resolveOpenCommand function, covering all platform branches (darwin, win32, linux) and URL pass-through. Changelog fragment included.
* chore: rename changelog fragment 7842->7844 to match actual PR number
---------
Co-authored-by: tientien17 <tientien17@users.noreply.github.com>
#7838 added six providers.tierOverride* keys to en.json without the Vietnamese
counterparts; i18n-vi-completeness (key parity + both ICU checks) fails on the
release tip for every fresh PR run. Translated using the locale's existing tier
vocabulary and inserted at the mirrored position.
