Adds docs/reference/REMOVED_PROVIDERS.md (policy + register: puter #10210,
the keyless provider removed in #12440), links it from the docs index and
AGENTS.md's provider checklist, and adds a regression test that fails if any
registered id, alias or domain shows up again in the provider catalogs, the
executor map or the registry/executor sources.
Co-authored-by: Markus Hartung <diegosouzapw@users.noreply.github.com>
* feat(db): a2a task history module over migration-002 tables
* feat(a2a): persist task lifecycle to a2a_tasks with 30d retention purge
* feat(api): a2a task history listing + historical detail fallback
* feat(dashboard): orchestration History tab (Airflow-grid) over persisted runs (2.2)
* fix(dashboard): assert history preset window + loading and time axis in the grid
* chore(dashboard): history i18n + changelog
* fix(dashboard): explicit history purge cascade + keep live drawer off the History tab
* docs: document OMNIROUTE_A2A_HISTORY_RETENTION_DAYS
* refactor(dashboard): split HistoryTab helpers under the complexity ratchet
---------
Co-authored-by: Markus Hartung <diegosouzapw@users.noreply.github.com>
* chore(quality): register video-bridge memory suppression test in stryker tap.testFiles
* fix(ci): point the node_modules cache key at wreqJsNative after the tls-client removal
---------
Co-authored-by: Markus Hartung <diegosouzapw@users.noreply.github.com>
Migrates the Claude, Grok, LMArena, Notion and Perplexity web-cookie transports from the tls-client-node/Koffi sidecar to the exactly pinned wreq-js 3.2.0 runtime, keeping the per-provider browser/OS profiles, making request cookies ephemeral, bounding and generation-protecting the shared native transport pool, removing the legacy downloader and repair path, and carrying the native binding and license evidence through the npm, standalone, Electron, Docker and Bun packaging surfaces.
This is the consolidation of the two competing migrations, and the consolidation was decided by evidence rather than by preference. #11753's six suites were installed over this implementation and run as an independent specification: 31 of 36 passed. All five failures are artefacts of #11753 being the older design, not coverage gaps —
- two hardcode the 3.0.0 pin in their assertions (this branch pins 3.2.0, which is what the release tip already resolves; #11753's 3.0.0 would have conflicted);
- one reads open-sse/services/chatgptTlsClient.ts, deleted when #11754 retired ChatGPT Web, so the test is stale against the current tip;
- two import WREQ_JS_NATIVE_BINARY_NAMES / resolveWreqJsNativeBinaryName, which this branch redesigned into WREQ_JS_NATIVE_BINDINGS / resolveWreqJsNativeBinding plus WREQ_JS_VERSION — a rename from modelling natives as file names to modelling them as package bindings, verified as an API difference rather than a lost capability (the linux-x64-gnu .node is present and serviceable).
This branch is also the strict superset by scope: 7 files exclusive to it, including the wreq-js Rust license inventory and notices, .trivyignore, open-sse/utils/tlsClient.ts and assembleStandalone.mjs. #11753 had one exclusive file, its changelog fragment. Nothing needed porting, so #11753 is superseded rather than merged, and the changelog entry credits both.
Reconciled on merge: clean against the tip. The new migration suite (tests/unit/tls-client-wreq-migration.test.ts, 1374 lines, 31 cases) is frozen at its exact LOC with the rationale — it shares one native-transport harness, so splitting it mid-merge would duplicate that harness for no coverage gain. Verified that no existing cap moves.
Verified: 182/182 across the eight TLS, native-manifest, postinstall, standalone-bundle, pack-artifact and provider-validation suites, typecheck:core clean, check:cycles OK, check-changelog-integrity OK, check-file-size OK, and every changed TypeScript file parses.
Restores ChatGPT Web on a clean-room browser transport, merged on the operator's explicit decision.
Worth stating precisely, because this touches a provenance decision: the PR does not revert #11754. It narrows RETIRED_COMMON_CHATGPT_WEB_PROVIDER_IDS to the single GPL-derived alias cgpt-web and registers chatgpt-web as a separate clean-room id. The old implementation stays retired and blocked; the retirement machinery, its error code and its 410 contract are untouched. All four retirement suites agree with that distinction and pass unchanged.
The 44 protected agent-instruction surfaces this PR touches (AGENTS.md, llm.txt and its 42 mirrors, README) were verified rather than trusted: masking digits and comparing the removed and added line sets gives 264 lines on each side, identical — every change is a provider-count substitution, with no sentence added, removed or reworded.
Reconciled on merge: clean against the tip, with the two chat chokepoints this PR grows (src/sse/handlers/chat.ts +40, open-sse/handlers/chatCore.ts +30) recorded in the file-size baseline under an annotation. Verified that exactly those two caps move and nothing else, so the #12411 ratchet holds. The rebaseline is carried on this branch rather than left in a validation worktree — the propagation mistake that put the 2026-09-02 merge waves base-red in #12434.
Verified: 504/504 across the PR's 44 test files plus all four chatgpt-web retirement suites, check:provider-consistency OK (272 REGISTRY entries, 355 canonical providers), check-file-size OK, and every changed TypeScript file parses.
Thanks @backryun — separating the clean-room id from the retired alias, instead of reopening the old one, is what made this reviewable.
Reduced on merge rather than closed, because the useful half is not subsumed.
The production change is: #12423 landed first and reached the same end state for scripts/build/pack-artifact-policy.ts — one volatileEnvPath.mjs entry, keeping the #11437 comment that explains why it is REQUIRED (bin/omniroute.mjs calls describeVolatileEnvWarning on every CLI boot, and bin/cli/ is only an allowlist prefix, so its absence would otherwise be silent). This PR's base carried three occurrences and reduced them to one; the tip is already there, so that file takes the tip's side.
What survives is the guard test, which does not exist on the tip: it asserts the four artifact path policy arrays contain no duplicate entries, so the class of defect cannot come back quietly. Verified by proof rather than assumption — re-introducing the duplicate makes it fail, removing it makes it pass again.
Verified: 18/18 in pack-artifact-policy after the reduction.
Thanks — the duplicate was real and the guard is the part worth keeping.
P1 of #12150: transcript text no longer reaches call logs or durable memory in the clear.
Reconciled on merge — the two persisted-requestBody assertions were failing, and the failure signature was misleading enough to be worth recording. They reported "expected: true, actual: false", which reads like the redaction not applying. It was not: pollForCallLog waited at most 120 tries x 20ms = 2.4s for the asynchronous SQLite write, then returned null, so assert.ok(row) failed before any redaction assertion ran. The observed durations were 5253ms and 4467ms against that 2.4s ceiling — a starved runner, not a leak. The control test failing alongside the positive one was the tell: a real redaction defect would break one direction, not both.
Replaced the fixed try count with a 30s wall-clock deadline: far past any healthy write, still bounded, and a fast machine still returns on the first pass. A privacy test should not depend on how busy the box is.
Verified: 4/4 three consecutive times under synthetic load, and 3/3 unloaded beforehand. Note the synthetic load reached ~7, below the ~38 where the original failure appeared, but the budget is now 12.5x larger and deadline-based rather than count-based.
Applies the repository Prettier style to tests/unit/grok-web.test.ts. No production code, no assertion changes.
The AST-identical claim was verified independently rather than taken on trust: minifying both sides through esbuild produces byte-identical output.
The reformat expands the file from 2436 to 2713 lines, past its 2437 cap, so the baseline carries a new annotated entry at 2985 — the real LOC plus ~10% headroom, per the operator's instruction, so routine additions to this suite do not re-trip the gate on formatting alone. It is recorded as a deliberate exception to the down-only ratchet #12411 re-tightened; no other entry moves.
Verified: 68/68 in the reformatted suite, check-file-size OK.
Closes the two remaining unresolved provider-asset records without weakening the provenance boundary: public/providers/nimble-search.svg is removed because no sufficient redistribution evidence is recorded, and Nimble Search renders with the internal generic icon before any CDN tier while staying fully functional; Opper is registered in the local SVG resolver with its existing bytes proven against opper-ai/provider-omniroute at immutable commit 9aacef7d6a, recording the MIT repository license from that same commit while keeping trademarkClearance null.
Validated in a combined worktree with the batch's ready set boarded onto the current tip: parse sweep clean on every changed TypeScript file, typecheck:core clean, check:dashboard-typecheck OK (207 pre-existing errors, all within baseline), check:cycles OK, check-file-size OK, 203/205 focused node tests and 94/94 vitest — the two failures belong to #12427, which is held back.
The current release dependencies already resolve the ChatGPT Web vendor diagnostics the allowance covered, so the stale open-sse typecheck allowance is removed and any future diagnostic becomes a blocking regression again. No vendor source, package manifest, checker script or runtime behaviour changes.
Validated in a combined worktree with the batch's ready set boarded onto the current tip: parse sweep clean on every changed TypeScript file, typecheck:core clean, check:dashboard-typecheck OK (207 pre-existing errors, all within baseline), check:cycles OK, check-file-size OK, 203/205 focused node tests and 94/94 vitest — the two failures belong to #12427, which is held back.
Closes the remaining noImplicitAny gap in global system-prompt injection without changing valid OpenAI or Claude request behaviour: injectSystemPrompt gets a caller-preserving generic type, unknown bodies/message entries/content are narrowed before access, malformed entries are skipped instead of throwing, and request/message/content immutability is preserved.
Validated in a combined worktree with the batch's ready set boarded onto the current tip: parse sweep clean on every changed TypeScript file, typecheck:core clean, check:dashboard-typecheck OK (207 pre-existing errors, all within baseline), check:cycles OK, check-file-size OK, 203/205 focused node tests and 94/94 vitest — the two failures belong to #12427, which is held back.
The API-route and Open-SSE typecheck gates treated every top-level baseline value as a diagnostic map, so policy metadata such as _relax_velocity_2026_08_30 was iterated character by character and reported as fabricated numeric TypeScript improvements. Both gates now share one fail-closed baseline boundary: underscore-prefixed top-level keys are reserved for metadata and skipped, and real file entries must be plain objects.
Worth a follow-up: check-dashboard-typecheck still prints the same fabricated entries, so it looks like a third gate with the same defect that this PR's scope does not cover.
Validated in a combined worktree with the batch's ready set boarded onto the current tip: parse sweep clean on every changed TypeScript file, typecheck:core clean, check:dashboard-typecheck OK (207 pre-existing errors, all within baseline), check:cycles OK, check-file-size OK, 203/205 focused node tests and 94/94 vitest — the two failures belong to #12427, which is held back.
The rerank-provider listing route dynamically imported @/lib/localDb — the barrel Hard Rule #2 forbids — and the stale path meant local rerank-capable provider nodes never appeared in GET /api/memory/rerank-providers. Now imports the specific @/lib/db/readCache module, with a route-level regression test through the public GET handler.
Validated in a combined worktree with the batch's ready set boarded onto the current tip: parse sweep clean on every changed TypeScript file, typecheck:core clean, check:dashboard-typecheck OK (207 pre-existing errors, all within baseline), check:cycles OK, check-file-size OK, 203/205 focused node tests and 94/94 vitest — the two failures belong to #12427, which is held back.
The service operator asked in writing (2026-08-30) that their service be
removed from OmniRoute entirely: executor, registry entry, no-auth catalog
entry and alias, icon mapping, env var, docs rows, dedicated tests and
snapshots, and every passing mention in comments, fixtures and CHANGELOG
entries. Provider count drops from 355 to 354 on every canonical surface.
Co-authored-by: Markus Hartung <diegosouzapw@users.noreply.github.com>
* chore(lint): adopt eslint-plugin-react-hooks 7.1.1
The #12146 migration (284 react-hooks compiler-rule violations resolved in 8
batches) completed on 2026-09-01, unblocking the 7.1.1 adoption the pin test
was holding back. Exact pin kept in both devDependencies and overrides; the
pin test moves to 7.1.1 (the dependabot-level ignore from #12329 stays — a
lint plugin coupled to the compiler rules always bumps via its own reviewed
PR, never riding a group).
* chore(lint): lockfile for the react-hooks 7.1.1 adoption
Generated with a bare 'npm install --package-lock-only' (naming the package
on the CLI rewrites the devDependency with a caret, which npm 11 then rejects
against the exact override). Validated on the .113 with a fresh npm ci +
cold NODE_OPTIONS=8G lint:json --max-warnings 0 → exit 0 (zero new
violations from the 7.1.1 rule set) and the re-pinned version test green.
Drains the two remaining Fast Quality Gates reds the #11513 (UC) merge left
on the tip:
- error-helper: ucTts.ts and uc/ws.ts built error payloads from raw
err.message (Hard Rule #12) — now wrapped in sanitizeErrorMessage(),
behavior otherwise identical (uc suites 51/51).
- model-lifecycle: the UC catalog registers the vendor-retired gpt-5.2-codex
(bare id; only the prefixed openai/gpt-5.2-codex was allowlisted). Added to
allowedRetiredInCatalog per its policy — forwarding globally would rewrite
the just-approved provider's model. Tracking: Refs #12436.
file-size, the third red of this window, was already drained by #12434.
check-file-size was red on release/v3.8.51 with nine violations — seven source files and two test files that the 2026-09-02 merge waves grew at existing chokepoints (#12359-#12404, #11461, #11513, #12423).
The growth itself was reviewed: each file was measured and justified while validating those batches. What went wrong is the propagation — the rebaseline was computed in the throwaway combined validation worktree, and the PRs were then merged individually through their own branches, so the code landed and the caps did not. A shared-file edit made only in the validation tree reaches nothing.
This records the caps against the merged state, each entry attributed to the PR that grew it, under one _rebaseline annotation. Verified mechanically: 9 caps recorded, 0 raised beyond the file's real merged LOC, 0 unrelated entries moved — the ratchet #12411 re-tightened is intact.
Verified: check-file-size OK (135 frozen source entries across 4515 files; 39 frozen test entries across 5365), prettier clean.
The #11461 × #11513 merge ate the closing '],' + '},' of the maxai entry in
webSessionCredentials.ts — 11 syntax errors (TS1005/1137/1128) on the tip,
which also masked one real TS2322 the MaxAI block introduced in the models
route (providerSpecificData is unknown on the connection; cast to the exact
shape resolveMaxaiCredential already takes, zero runtime change).
API Route Typecheck gate: OK — 289 pre-existing, all baselined. typecheck:core: 0.
Three regressions inherited by every PR rebased onto release/v3.8.51, caught and documented with the exact failing output.
The one that mattered most: src/shared/providers/webSessionCredentials.ts did not parse. The UC merge (#11513) inserted the uc: entry inside maxai.storageKeys and lost the array's closing ], plus the entry's }, leaving `ERROR: Expected "]" but found ":"` at line 351. That module is imported by the provider API routes, bulk-web-session, autoCombo's virtualFactory, keepaliveThreshold and dashboard components, so the break was live on the tip and flooded unrelated catalog tests with transform failures. That was my conflict resolution, not the contributor's code — thank you for catching it and for tracing it to the root commit rather than patching around the symptom.
Also fixed: the duplicate bin/cli/utils/volatileEnvPath.mjs entry in PACK_ARTIFACT_REQUIRED_PATHS (findMissingArtifactPaths reported it twice), and UC image models made prefix-addressable without letting them claim historical bare model ids belonging to other providers.
Reconciled on merge: #12394 landed the busy_timeout/probe work first, so src/lib/db/core.ts takes the tip's side. probeUtils.ts is the union of both rather than either side — this PR's message regex is wider (SQLite also reports "database table is locked", "database schema is locked" and "database is busy"), while #12394 added the driver code/errcode path that keeps a transient lock from being classified as corruption and renaming the database away. Taking either alone would have dropped the other half; this PR's own ENOENT test is what surfaced it.
Verified: 76/76 across uc-image, probe-9541-repro, web-session-contract, pack-artifact-policy, bulk-web-session-import and exclusive-connection-leases, and every changed .ts file parses.
Thanks @backryun.
GET /v1/providers/gemini-business/models returned nothing because gemini-business had no RegistryEntry: the listing route resolves the provider through getRegistryEntry and filters the unified catalog by owned_by, and open-sse/config/providers/index.ts only registered gemini and gemini-web.
Adds a registry entry mirroring gemini_webProvider — id gemini-business, alias gembiz, cookie auth — with the twelve ids from the executor's MODEL_CATEGORY_MAP. Each model is declared toolCalling: false, supportsReasoning: false, the same live-behaviour contract applied to gemini-web in #9356: the executor returns plain text, hard-wires the thinking mode and parses no tool calls.
Reconciled on merge: the only conflict was the reserved-prefix count assertion, which the tip had moved. Took the tip's text and measured the real value with this PR applied — 406 to 408, the gemini-business id plus its gembiz alias — rather than carrying the branch's number.
Validated in a combined worktree with all 25 PRs of this batch boarded together (typecheck:core clean, 443/443 node-runner plus 14/14 vitest, all static gates green), and re-verified standalone on the current tip after the other 24 landed: 33/33 across provider-node-reserved-prefix, gemini-business-model-registry-12107 and web-cookie-validation-fallback, with check:provider-consistency OK at 272 REGISTRY entries and 355 canonical providers.
Thanks @pacocartones.
The gamification anomalies page had hard-coded English for its loading state, Status column header and Suspicious badge, and was the only standalone non-redirect dashboard page without a sidebar entry. Both are fixed: the strings come from the common catalog, and the page joins the Gamification sidebar group as a hideable section item shown only by the "all" preset, like its siblings. The loading and empty states also become role="status" aria-live="polite" live regions with aria-busy, matching profile/page.tsx and health/page.tsx. Three new keys in en.json, propagated to the other 42 locales with the __MISSING__ sentinel.
Validated in a combined worktree with all 25 PRs of this batch boarded together: typecheck:core clean, 443/443 node-runner tests plus 14/14 vitest across every test file the batch touches, and check-changelog-integrity, check:cycles (418 files), check:provider-consistency (272 REGISTRY entries, 355 canonical providers), check:docs-counts, check:docs-sync (42 locales) and check-file-size all green.
Thanks @pacocartones.
The API key permissions modal silently dropped allowedCombos entries its Combo picker cannot render — routing-rule names such as rt-*, which matchesComboAccessRule() already honours. Stored entries rendered as zero selected, and clicking All then Restrict then Save persisted allowedCombos: [], which is deny-all for combo requests.
Those entries now survive the All toggle, are listed read-only under the combo list so the header count and the list agree, and are saved back verbatim. The UI does not learn routing-rule semantics (option 1 from the issue). The Allowed Combos section moves out of the frozen ApiManagerPageClient.tsx into its own component following the UsageLimitSettings pattern.
Validated in a combined worktree with all 25 PRs of this batch boarded together: typecheck:core clean, 443/443 node-runner tests plus 14/14 vitest across every test file the batch touches, and check-changelog-integrity, check:cycles (418 files), check:provider-consistency (272 REGISTRY entries, 355 canonical providers), check:docs-counts, check:docs-sync (42 locales) and check-file-size all green.
Thanks @pacocartones.
GET /v1/models with MODELS_CATALOG_PREFIX_MODE=canonical dropped every chat row of a provider whose registry alias is undefined (antigravity) or equal to its own id (agy, most built-ins). Each emission loop pushes alias/model only when includeAlias, and canonicalProviderId/model only when the ids differ — for a self-aliased provider both are the same string, so neither fired. #11918 fixed the class for custom nodes but not built-ins, and not the static loop. The alias row is now treated as the canonical row whenever the ids coincide, across the static, synced, custom and alias-backed loops; the canonical branch's !== alias guard is untouched, so dual and alias output cannot double up. Docs that described the omission as intended are corrected.
Validated in a combined worktree with all 25 PRs of this batch boarded together: typecheck:core clean, 443/443 node-runner tests plus 14/14 vitest across every test file the batch touches, and check-changelog-integrity, check:cycles (418 files), check:provider-consistency (272 REGISTRY entries, 355 canonical providers), check:docs-counts, check:docs-sync (42 locales) and check-file-size all green.
Thanks @pacocartones.
The response de-obfuscation stripped the whole U+200B..U+200D range, so Persian/Kurdish half-spaces (U+200C), Arabic/Indic shaping and emoji ZWJ sequences (U+200D) were deleted from every assistant response — text, reasoning and tool-call arguments, streaming and non-streaming, every provider: ارائهدهنده came back as ارائهدهنده.
The request side only ever inserts a U+200D between two ASCII word characters, so the new stripObfuscationZeroWidth() removes a joiner only there, or at a string edge next to one so a word split across streaming deltas is still cleaned; U+200B and U+FEFF keep their unconditional removal. All seven copies of the old regex now go through the helper.
Validated in a combined worktree with all 25 PRs of this batch boarded together: typecheck:core clean, 443/443 node-runner tests plus 14/14 vitest across every test file the batch touches, and check-changelog-integrity, check:cycles (418 files), check:provider-consistency (272 REGISTRY entries, 355 canonical providers), check:docs-counts, check:docs-sync (42 locales) and check-file-size all green.
Thanks @pacocartones.
CircuitBreaker.execute() treated every resolved promise as a success, but handleChatCore() reports most upstream failures by resolving with { success: false, status: 5xx }. On the chat path that spurious _onSuccess() decayed failureCount right before the call site's _onFailure() for the same attempt, so a provider answering 503s indefinitely stayed CLOSED at failureCount: 1 and kept receiving traffic — the breaker was structurally unable to open. Combo dispatches hit the same cancellation through the shared per-provider breaker.
execute() now takes an optional per-call classifyResult; without it the resolved-means-success contract every throw-based caller relies on is unchanged. executeChatWithBreaker() passes ignore and the chat path accounts for the outcome exactly once where the request context lives, so a combo success is no longer counted twice.
Validated in a combined worktree with all 25 PRs of this batch boarded together: typecheck:core clean, 443/443 node-runner tests plus 14/14 vitest across every test file the batch touches, and check-changelog-integrity, check:cycles (418 files), check:provider-consistency (272 REGISTRY entries, 355 canonical providers), check:docs-counts, check:docs-sync (42 locales) and check-file-size all green.
Thanks @pacocartones.
The retryable chat_admission_busy 503 advertised a fixed Retry-After of 1s or 2s while the heavyweight lease it waits on is held for the entire SSE lifetime. Clients that honour the header — Codex CLI, agent fan-out — re-sent the same ~1 MiB /v1/responses body every second into a gate that could not have cleared, producing the queue_timeout retry storm that persisted even after OMNIROUTE_CHAT_MAX_HEAVY_IN_FLIGHT was raised.
ChatAdmissionController now tracks each live heavy lease's acquisition time and derives the hint from observed occupancy: the larger of the queue window the waiter already exhausted and the age of the youngest live lease, rounded up and capped at 60s. Both builders floor it at the historical 1s / 2s, so an idle gate answers exactly as before. Using the youngest rather than the oldest lease avoids a pessimistic hint when several slots are in flight.
Validated in a combined worktree with all 25 PRs of this batch boarded together: typecheck:core clean, 443/443 node-runner tests plus 14/14 vitest across every test file the batch touches, and check-changelog-integrity, check:cycles (418 files), check:provider-consistency (272 REGISTRY entries, 355 canonical providers), check:docs-counts, check:docs-sync (42 locales) and check-file-size all green.
Thanks @pacocartones.
The Leaderboard rendered apiKeyId.slice(0, 8)… under a column translated as "name". The route now enriches each entry with the key's display name — route-local, so the shared getTopN helper and the federation leaderboard stay id-only — and the page renders name ?? shortId with the full id in a title attribute. The lookup selects only id and name from api_keys, chunked at 200 ids, with unknown ids and blank names omitted; no key material leaves the DB layer.
Validated in a combined worktree with all 25 PRs of this batch boarded together: typecheck:core clean, 443/443 node-runner tests plus 14/14 vitest across every test file the batch touches, and check-changelog-integrity, check:cycles (418 files), check:provider-consistency (272 REGISTRY entries, 355 canonical providers), check:docs-counts, check:docs-sync (42 locales) and check-file-size all green.
Thanks @pacocartones.
The Profile page already rendered a streak card but fed it a hard-coded useState(0) with a "streak data comes from future API" note — while streaks.ts tracked per-key streaks all along and the MCP gamification_profile tool already returned them. GET /api/gamification/level now returns streak: { current, longest } next to level: the key's own streak with apiKeyId, the operator-wide maximum otherwise, matching the aggregate mode getAggregateXp uses (#3484). No new route, no OpenAPI change, no new i18n keys; a missing or zero streak keeps the card hidden exactly as before.
Validated in a combined worktree with all 25 PRs of this batch boarded together: typecheck:core clean, 443/443 node-runner tests plus 14/14 vitest across every test file the batch touches, and check-changelog-integrity, check:cycles (418 files), check:provider-consistency (272 REGISTRY entries, 355 canonical providers), check:docs-counts, check:docs-sync (42 locales) and check-file-size all green.
Thanks @pacocartones.
A thinking content part arriving with no signature — typical after a cross-provider hop where reasoning_content was converted into a thinking block — was stamped with DEFAULT_THINKING_CLAUDE_SIGNATURE. prepareClaudeRequest treats any non-empty signature on the latest assistant turn as genuine and preserves it verbatim, so the fabricated one reached Anthropic and the replay failed with "Invalid signature". A missing signature is now treated the same as an empty one, aligned with the stricter check claudeHelper.ts already used: the block is dropped rather than fabricated. Real signatures are still preserved verbatim and redacted_thinking is unchanged.
Validated in a combined worktree with all 25 PRs of this batch boarded together: typecheck:core clean, 443/443 node-runner tests plus 14/14 vitest across every test file the batch touches, and check-changelog-integrity, check:cycles (418 files), check:provider-consistency (272 REGISTRY entries, 355 canonical providers), check:docs-counts, check:docs-sync (42 locales) and check-file-size all green.
Thanks @pacocartones.
A proxy assigned to an opencode / opencode-go connection is pinned by the chat handler as the ambient proxy context before the executor runs. OpencodeExecutor only reads per-account proxies from providerSpecificData.accountProxies, so an API-key connection with none took the single-account fast path — which wrapped the dispatch in runWithDirectFetchContext(), and that direct sentinel makes patchedFetch bypass the ambient context and hit native fetch. The assigned proxy was discarded and the request egressed from the host IP, giving `403 This model is not available in your country` on geoblocked hosts. The fast path now applies the direct pin only when no ambient context exists.
Validated in a combined worktree with all 25 PRs of this batch boarded together: typecheck:core clean, 443/443 node-runner tests plus 14/14 vitest across every test file the batch touches, and check-changelog-integrity, check:cycles (418 files), check:provider-consistency (272 REGISTRY entries, 355 canonical providers), check:docs-counts, check:docs-sync (42 locales) and check-file-size all green.
Thanks @pacocartones.
POST /v1/images/generations through a combo returned a bare array instead of the OpenAI {created, data} payload: executeImageCombo() unwrapped one level too many, and the n used for cost calculation read the same double-nested shape, so it was always 0. The combo path now returns the handler payload unchanged, matching the direct-model path.
Second half: Codex image results emitted a data: URI in url whenever response_format was not b64_json, but OpenAI returns b64_json for the gpt-image-* family — clients that omit the field, Codex CLI's built-in image_gen among them, could decode neither shape. Codex now defaults to b64_json; an explicit response_format: "url" keeps its previous behaviour. Both land together because fixing one leaves Codex CLI failing at the other.
Validated in a combined worktree with all 25 PRs of this batch boarded together: typecheck:core clean, 443/443 node-runner tests plus 14/14 vitest across every test file the batch touches, and check-changelog-integrity, check:cycles (418 files), check:provider-consistency (272 REGISTRY entries, 355 canonical providers), check:docs-counts, check:docs-sync (42 locales) and check-file-size all green.
Thanks @pacocartones.
When a built-in provider's id or alias reserves the prefix of an existing OpenAI/Anthropic-compatible node — v3.8.50 added openference with alias of, shadowing nodes created earlier with prefix of — the runtime error `No active credentials for provider: openference` gave the operator nothing to act on. It now explains that the prefix routed to the built-in, names the shadowed node, and logs an AUTH warning. Precedence is unchanged and the lookup runs only on the credential-failure path when no connection was tried, so the hot routing path is byte-identical.
Validated in a combined worktree with all 25 PRs of this batch boarded together: typecheck:core clean, 443/443 node-runner tests plus 14/14 vitest across every test file the batch touches, and check-changelog-integrity, check:cycles (418 files), check:provider-consistency (272 REGISTRY entries, 355 canonical providers), check:docs-counts, check:docs-sync (42 locales) and check-file-size all green.
Thanks @pacocartones.
getDbInstance() ran PRAGMA journal_mode = WAL as the connection's first statement, before PRAGMA busy_timeout, and openSqliteDatabase() passes no driver-level timeout. A process opening the database while another closed its WAL connection — checkpoint plus WAL delete hold an EXCLUSIVE lock for a few hundred microseconds — therefore died with `database is locked` instead of waiting. That is the flake behind exclusive-connection-leases.test.ts on release/v3.8.51 runs 33525300898 and 33493797519 and on unrelated PR runs.
The second half is worse than the flake: isTransientProbeError matched /SQLITE_BUSY/ against error.message, but both drivers report the plain text `database is locked` and put the code in .code / .errcode. A transient lock during the corruption probe therefore took the corrupt-database path and renamed the file to storage.sqlite.probe-failed-… with "Manual recovery required". The probe now recognises the drivers' real BUSY/PROTOCOL/IOERR signals.
Validated in a combined worktree with all 25 PRs of this batch boarded together: typecheck:core clean, 443/443 node-runner tests plus 14/14 vitest across every test file the batch touches, and check-changelog-integrity, check:cycles (418 files), check:provider-consistency (272 REGISTRY entries, 355 canonical providers), check:docs-counts, check:docs-sync (42 locales) and check-file-size all green.
Thanks @pacocartones.
validateScoreChange() — the documented 1000 XP/min per-API-key limit plus the velocity anomaly check — was exported but never called, so the award path applied every XP delta unconditionally. It now runs before addXp; a rejected award is logged at warn level and skipped, and the fire-and-forget path never throws.
The second finding is the one that made the first invisible: getRecentXp's window query was inert. created_at is stored by the table default as YYYY-MM-DD HH:MM:SS and was compared lexically against a JS ISO string, so same-day rows never matched and the limit could not have tripped even if it had been wired. The window start is now computed in SQLite, matching the style computeZScore already used.
Validated in a combined worktree with all 25 PRs of this batch boarded together: typecheck:core clean, 443/443 node-runner tests plus 14/14 vitest across every test file the batch touches, and check-changelog-integrity, check:cycles (418 files), check:provider-consistency (272 REGISTRY entries, 355 canonical providers), check:docs-counts, check:docs-sync (42 locales) and check-file-size all green.
Thanks @pacocartones.
Four documentation claims contradicted the code: .env.example called OMNIROUTE_USE_TURBOPACK dev-only and said the production build still uses webpack (it reads the same flag and defaults to Turbopack); the README's Bun section said `bun run build` auto-detects Bun and switches to Webpack (only `bun run dev` does — the production bundler is decided by the flag alone); TROUBLESHOOTING.md gave OMNIROUTE_CHAT_MAX_HEAVY_IN_FLIGHT a default of 1 when unset means no request-count cap; and it quoted the pre-#12223 wording of the structural 503 chat_admission_busy message. The Retry-After bullet in the same section is deliberately untouched because #12395 rewrites it.
Validated in a combined worktree with all 25 PRs of this batch boarded together: typecheck:core clean, 443/443 node-runner tests plus 14/14 vitest across every test file the batch touches, and check-changelog-integrity, check:cycles (418 files), check:provider-consistency (272 REGISTRY entries, 355 canonical providers), check:docs-counts, check:docs-sync (42 locales) and check-file-size all green.
Thanks @pacocartones.
Four files embedded the U+0000 separator of a memo/group key as a raw NUL byte rather than the \\0 escape the codebase uses for the same idiom elsewhere. The runtime value is identical, but the raw byte trips the binary heuristics of git, GitHub and ripgrep: git diff --numstat reported `- -`, the introducing PRs rendered three of the files as "Binary file not shown", and rg silently skipped them in recursive mode. Rewritten as escapes, with a guard test keeping raw NUL bytes out of src/, open-sse/ and tests/.
Validated in a combined worktree with all 25 PRs of this batch boarded together: typecheck:core clean, 443/443 node-runner tests plus 14/14 vitest across every test file the batch touches, and check-changelog-integrity, check:cycles (418 files), check:provider-consistency (272 REGISTRY entries, 355 canonical providers), check:docs-counts, check:docs-sync (42 locales) and check-file-size all green.
Thanks @pacocartones.
Under Bun the server child is spawned with --preload <path>/open-sse/utils/setupPolyfill.ts, and all three spawn sites built that path next to the server bundle — but the polyfill only ships at the package root and nothing copies it into dist/. Every `bun install -g omniroute` start died with `error: preload not found`. The preload now resolves from the supervisor module's own location and is shared by the two serve.mjs spawns, with the child argv moved into a pure buildServerSpawnArgs() so both branches are directly assertable (same seam as #8131). Node users are unaffected.
Validated in a combined worktree with all 25 PRs of this batch boarded together: typecheck:core clean, 443/443 node-runner tests plus 14/14 vitest across every test file the batch touches, and check-changelog-integrity, check:cycles (418 files), check:provider-consistency (272 REGISTRY entries, 355 canonical providers), check:docs-counts, check:docs-sync (42 locales) and check-file-size all green.
Thanks @pacocartones.
getBestVisionModel() validates a configured fixedModel with hasUsableCredentialsForModel() before short-circuiting (#8430). auto / auto/* ids are virtual combos with no provider row, so that check always reported a confirmed false and the combo was silently discarded in favour of global auto-selection — it never got the chance to rotate its members. This mirrors the exemption the reroute guard in visionBridge.ts already carries; concrete fixedModel ids keep the #8430 fall-through unchanged.
Validated in a combined worktree with all 25 PRs of this batch boarded together: typecheck:core clean, 443/443 node-runner tests plus 14/14 vitest across every test file the batch touches, and check-changelog-integrity, check:cycles (418 files), check:provider-consistency (272 REGISTRY entries, 355 canonical providers), check:docs-counts, check:docs-sync (42 locales) and check-file-size all green.
Thanks @pacocartones.
The opt-in Codex quota auto-ping pinned gpt-5.1-codex-mini. OpenAI shut that model down on 2026-07-23 and the repo's own lifecycle registry already rejects it on the request path, but the scheduler never consulted that gate — every window slide sent a dead id, hit the 15-minute failure cooldown, and retried the same id forever. The ping model now resolves per tick from the provider catalog through isModelSelectable(), the same gate chatCore uses, with the registry import kept lazy because this module sits on the instrumentation boot path (#12074). When nothing is selectable the provider is paused before any throttle slot, usage read or executor call, with one warning per state change.
Validated in a combined worktree with all 25 PRs of this batch boarded together: typecheck:core clean, 443/443 node-runner tests plus 14/14 vitest across every test file the batch touches, and check-changelog-integrity, check:cycles (418 files), check:provider-consistency (272 REGISTRY entries, 355 canonical providers), check:docs-counts, check:docs-sync (42 locales) and check-file-size all green.
Thanks @pacocartones.
The #10265 rewrite of command-code-executor.test.ts (b6412c6fe) deleted the two regression tests #10986 added for reasoning-only Command Code output, while the production fallback in createJsonResponse / createStreamResponse survived — leaving it unguarded. Both are restored, now routed through the /alpha/generate fallback that is the only way to reach the CLI translator since #10265, via a shared goPlanFallbackFetch() helper.
Validated in a combined worktree with all 25 PRs of this batch boarded together: typecheck:core clean, 443/443 node-runner tests plus 14/14 vitest across every test file the batch touches, and check-changelog-integrity, check:cycles (418 files), check:provider-consistency (272 REGISTRY entries, 355 canonical providers), check:docs-counts, check:docs-sync (42 locales) and check-file-size all green.
Thanks @pacocartones.
The least-used strategy ranked candidates by lastUsedAt alone, so after a 429 excluded the active account the replacement could be one that was merely oldest while still carrying its own backoff — it served a single request before the next one settled on a healthy account, the one-request detour with two cache misses reported on Codex. least-used now applies the backoffLevel tie-break the round-robin fallback branch already had, ahead of the existing never-used / oldest / priority order.
Validated in a combined worktree with all 25 PRs of this batch boarded together: typecheck:core clean, 443/443 node-runner tests plus 14/14 vitest across every test file the batch touches, and check-changelog-integrity, check:cycles (418 files), check:provider-consistency (272 REGISTRY entries, 355 canonical providers), check:docs-counts, check:docs-sync (42 locales) and check-file-size all green.
Thanks @pacocartones.
When every combo target is excluded because the request's max_tokens exceeds each target's known output limit, the terminal 400 now says so — requested max_tokens against the pool's highest known ceiling — instead of the unrelated "supports structured output for this request". Diagnostics (unmet, excluded[].reason, terminalReason) are unchanged; only the message for the output_tokens primary reason moves.
Validated in a combined worktree with all 25 PRs of this batch boarded together: typecheck:core clean, 443/443 node-runner tests plus 14/14 vitest across every test file the batch touches, and check-changelog-integrity, check:cycles (418 files), check:provider-consistency (272 REGISTRY entries, 355 canonical providers), check:docs-counts, check:docs-sync (42 locales) and check-file-size all green.
Thanks @pacocartones.
groq/compound and allam-2-7b were absent from the curated Groq registry, so the capability heuristic defaulted them to reasoning-capable and forwarded reasoning_effort verbatim — Groq answers HTTP 400. Declaring supportsReasoning: false makes applyThinkingBudget() strip reasoning_effort, output_config.effort and thinking, same class as #3258. The gpt-oss reasoning models keep the field.
Validated in a combined worktree with all 25 PRs of this batch boarded together: typecheck:core clean, 443/443 node-runner tests plus 14/14 vitest across every test file the batch touches, and check-changelog-integrity, check:cycles (418 files), check:provider-consistency (272 REGISTRY entries, 355 canonical providers), check:docs-counts, check:docs-sync (42 locales) and check-file-size all green.
Thanks @pacocartones.
The playwright:v1.62.0-noble base ships Chromium as a Chrome for Testing build, which extracts to chrome-linux64/chrome. The CMD's find -path '*/chrome-linux/chrome' matched nothing, $chrome_path came out empty, and the container crash-looped on `exec: --headless=new: not found`. Widening the glob to '*/chrome-linux*/chrome' resolves both the legacy and the Chrome for Testing layout.
Validated in a combined worktree with all 25 PRs of this batch boarded together: typecheck:core clean, 443/443 node-runner tests plus 14/14 vitest across every test file the batch touches, and check-changelog-integrity, check:cycles (418 files), check:provider-consistency (272 REGISTRY entries, 355 canonical providers), check:docs-counts, check:docs-sync (42 locales) and check-file-size all green.
Thanks @pacocartones.
* fix(memory): point the rerank-providers dynamic import at the real db module
#11390 landed with a dynamic import of the localDb barrel, which #12052 had
already removed from the base (and which Hard Rule #2 forbids) — the API
Route Typecheck gate reds on the tip with TS2307. getCachedProviderNodes
lives in src/lib/db/readCache.
* chore(quality): ratchet the api-typecheck baseline down (163 stale entries gone)
Regenerated with --update on a faithful npm ci environment (the .113 box)
against the current tip plus the rerank-providers import fix — the gate now
reads OK at 289 pre-existing errors, all baselined. No new entries added.
Adds uncensored.com as two OpenAI-compatible providers mirroring UC's own surfaces: uc, the persona/subscription side over WebSocket with a durable Clerk credential minting a short-lived per-connect token (no API key, un-metered), as a full multimodal port — chat, tools, vision, doc-RAG, image, video, TTS; and uc-direct, the metered Developer API over REST with X-api-key. Same underlying models, two billing surfaces.
Reconciled on merge. 57 files conflicted; only seven carried UC content, the rest was drift from the older release line and took the tip's side.
- executors/index.ts: the tip has since refactored the executor map to lazy dynamic imports, so uc is registered in that shape. uc-direct needs no entry — it routes through the default OpenAI-compatible executor.
- imageGeneration.ts: the branch still carried the retired designerWeb import alongside ucImage; kept only the UC one.
- config/providers/index.ts, webSessionCredentials.ts and web-cookie.ts resolved additively against the MaxAI entries #11461 put on the tip an hour earlier.
- web-cookie.ts: the uc entry declared no serviceKinds, required since #11392, so provider validation would have thrown at load. Declared ["llm"]. uc-direct already declared it at the end of its own entry — an earlier pass of mine added a second one after id and TypeScript caught the duplicate (TS1117); the author's placement is what shipped.
Every count was measured against the merged tree rather than taken from the branch, and all three would have been wrong: reserved prefixes are 406, not 399; APIKEY_PROVIDERS is 237, not 234; providers are 355. PROVIDER_REFERENCE.md regenerated, the count updated across README/AGENTS.md/llm.txt and its 42 mirrors, package.json and 6 SVGs — every changed line in the protected surfaces is a digit substitution and nothing else, verified by masking digits and comparing the removed and added sets (90 lines each, identical). The executor-map golden snapshot went 134 -> 135.
The branch's file-size-baseline.json predates #12411's ratchet re-tightening and was discarded rather than merged; imageGeneration.ts (+12 for the uc-image format branch) was entered against the current baseline under a _rebaseline annotation, and no other cap moves.
Verified: typecheck:core clean, check:provider-consistency OK (271 REGISTRY entries, 355 canonical providers), check:docs-counts exit 0, check-file-size OK, check:cycles OK, 119/119 across the PR's test files, and 2/2 executor-map-golden.
Thanks @arminanton — two providers for two real billing surfaces, rather than one entry pretending to be both, is the right modelling.
MaxAI joins as a first-class signed provider: 13 chat models discovered live from /models/get_config plus 6 image models, routed through the standard /v1 endpoints with per-request X-Authorization signing, browserless onboarding, prompted tool-calling, vision input, image generation and document RAG.
Reconciled on merge — worth reading, because the branch forked 227 commits back and 77 files conflicted. Only five carried MaxAI content; the rest was drift from the older release line and took the tip's side, taking the diff from 113 files to 37 (then 93 as counted against the current base).
- executors/index.ts: the tip has since refactored the executor map to lazy dynamic imports, so MaxAI is registered in that shape rather than the branch's static import.
- imageRegistry.ts: kept only the maxai block. The branch still carried microsoft-designer-web, which #11754 retired.
- models/route.ts: the conflicting hunk was an unrelated Vertex/Anthropic URL change, not MaxAI — tip's side.
- volcengine agent-plan/coding-plan registries: git auto-merged both sides and produced a duplicated supportsVision key, which TypeScript rejects (TS1117). Removed.
One real integration break that only the combined state shows: the MaxAI entry declared no serviceKinds, which #11392 made required a few hours ago. Provider validation threw at load time and check:provider-consistency crashed outright. Declared ["llm"] — the image kinds derive from imageRegistry, per the convention in that PR's backfill.
Every count was measured rather than taken from the branch, and each would have been wrong: reserved prefixes are 402, not the 397 the branch computed from its stale 395 base; providers are 353, not 354. PROVIDER_REFERENCE.md regenerated, the count updated across README/AGENTS.md/llm.txt and its 42 mirrors, package.json and 6 SVGs — every changed line in those files is a digit substitution and nothing else, verified by masking digits and comparing the removed and added sets (90 lines, identical). The executor-map golden snapshot was regenerated: keyCount 133 -> 134.
The branch's file-size-baseline.json predates #12411's ratchet re-tightening, so it was discarded rather than merged — taking it would have silently undone that. The three files this PR grows (proxyFetch.ts +20 for the Windows/firefox_150 TLS profile, imageGeneration.ts +12, models/route.ts +48) were entered against the current baseline under one _rebaseline annotation; no other cap moves.
Verified: typecheck:core clean, check:provider-consistency OK (269 REGISTRY entries, 353 canonical providers), check:docs-counts exit 0, check-file-size OK, check:cycles OK, and 79/79 across the MaxAI suites plus 21/21 reserved-prefix and 2/2 executor-map-golden.
Thanks @arminanton — the provider work itself is thorough; it was the 227 commits of base that needed the attention.
The +30% loosening of 2026-08-10 (fbbef4eaaf) left this gate inert: combo.ts carried a 5,691-line cap against 4,023 real lines and chatCore.ts 7,895 against 5,946. Both god-files grew roughly 600 lines in two weeks without the gate ever firing.
Mechanical check:file-size --update against the tip. No source touched. combo.ts 5,691 -> 4,023, chatCore.ts 7,895 -> 5,946, frozen source entries 178 -> 135 (43 already fit the 1,200 cap), frozen test entries 49 -> 39. From here every 3.8.52 decomposition slice lowers the cap again.
Reconciled on merge, and worth recording because neither PR could see it alone: #11460 (flat-rate cost estimates) landed first and grew CostOverviewTab.tsx from 1,282 to 1,319 lines. This PR had frozen that entry at 1,283 — measured before #11460 existed — so the two together would have turned the tip red while each was green on its own. --update correctly refuses to raise a cap, so the entry was set to the real post-merge LOC with a _rebaseline_2026_09_02_11460_flat_rate_estimates annotation naming #11460 as the growth, following the own-growth precedent already in the file (_rebaseline_2026_08_20_10531_freebuff_provider).
The ratchet invariant is intact and was checked rather than assumed: across the whole baseline, 45 caps decrease and 0 increase; CostOverviewTab.tsx still falls 2,002 -> 1,319.
Verified: check-file-size OK (135 frozen source entries, 4,481 files checked; 39 frozen test entries, 5,338 checked), and prettier clean on the baseline.
Claude Code (claude / cc) is correctly classified as a flat-rate subscription, so the analytics API reports $0 — accurate as billed cost, and useless as a view of what the subscription actually consumed. Neither the Costs nor the Analytics dashboard had a token-price-equivalent view.
The fix keeps both meanings rather than picking one: ordinary analytics callers keep billed-cost semantics ($0 for flat-rate), /dashboard/costs and /dashboard/analytics opt in explicitly via includeFlatRateEstimates=true, the response reports whether estimates were included so a caller cannot mistake them for vendor billing records, and the figures on /dashboard/costs are labelled as flat-rate estimates rather than presented as spend. Omitted, false and unknown values all retain the existing behaviour.
Scope note carried from the description: this is a checkpoint on #11459, not its full closure — the issue stays open.
Verified in a combined worktree with three sibling PRs of this batch: typecheck:core clean, 134/134 focused tests (4 skipped), and i18n UI coverage PASS across all 42 locales for the 43-file locale pass.
One cross-PR interaction worth recording, since it is invisible from either side: this grows CostOverviewTab.tsx from 1282 to 1318 lines, which is fine against the tip's current 2002 cap but exceeds the 1283 that #12411 (file-size ratchet re-tightening) would freeze. Neither PR fails alone. Merged first on purpose so #12411's mechanical --update recomputes against the real post-merge LOC — the cap still only goes down.
Thanks @xiaoyaner0201 — the opt-in contract plus the "were estimates included" flag is the right shape for this.
The contact sheet, the dedup comparator and the drill-down each validated JPEG data-URIs independently. They now share src/lib/guardrails/videoBridgeFrameContract.ts. No behaviour change; the sibling tests that asserted a per-module message were aligned to the shared one. Closes the Standards-4 residue from the 2026-08-18 Video Bridge review.
Verified in a combined worktree with three sibling PRs of this batch: typecheck:core clean, 134/134 focused tests (4 skipped), i18n UI coverage PASS across all 42 locales.
Every job installs through this composite — 36 times per ci.yml run, 8 per
quality.yml run — and each call paid ~80-90 s of npm ci even with setup-node's
npm tarball cache warm (measured 2026-09-01: 3,327 runner-seconds per ci.yml run
just installing). A node_modules cache keyed on runner.os + runner.arch + the
resolved Node version + hashFiles(package-lock.json, .npmrc, postinstall.mjs and
its five helpers) lets an exact hit skip the install entirely.
- No restore-keys, same rule as the ESLint cache (#11600): exact key or a full
npm ci, never a partial tree from another lockfile / Node / postinstall.
- The retry loop is unchanged and remains the miss path; --no-audit --no-fund
because audit:deps is its own gate.
- cache input (default true) lets a caller opt out.
- actions/cache pinned to the v6.1.0 hash already used in nightly-mutation.yml
(zizmor unpinned-uses blanket policy).
- tests/unit/build/npm-ci-retry-composite.test.ts pins the key contents, the
no-restore-keys rule and the miss path.
Refs #8084
serviceKinds now drops .optional() in providerSchema.ts, and check-provider-consistency gains the reverse walk: a canonical provider whose serviceKinds include "llm" must have a REGISTRY entry unless it is in the new KNOWN_CATALOG_ONLY allowlist (providers routed through a connection baseUrl or a specialised executor). That turns "catalog entry outlived its registry entry" — the half-finished provider:remove — into a checkable invariant instead of something a reviewer has to notice.
Reconciled on merge, and worth reading before comparing diffs. The branch's 18 files had landed at the repository ROOT: git diff --name-status showed A gateways.ts, A providerSchema.ts, A check-provider-consistency.test.ts, A backfill-servicekinds.mjs with no directory component. The real provider files, schema, gate and test were never touched, so the +5093/-0 diff was root files AGENTS.md forbids (a test outside tests/, a script outside scripts/) and a no-op for the feature. The content was also 227 commits stale — the root gateways.ts was missing oneminai, among 267 divergent lines.
So each file's actual delta was reapplied onto the current tip rather than copied: the schema one-liner; the gate's KNOWN_CATALOG_ONLY, findCatalogOnlyLlmProviders(), the main() check and the summary line (the branch's copy also repeated the file header and imports at the end — 12 lines of residue from the same accident, dropped); the test's import block and five reverse-walk cases; and backfill-servicekinds.mjs placed at scripts/ad-hoc/, the path its own docstring names, then run against the current catalog: 315 insertions, 352/352 entries declaring serviceKinds, idempotent on a second run.
Two entries the mechanical pass could not get right, both surfaced by doing it against the live tree:
- github in oauth.ts is a single-line object, so the script's id:-per-line regex skipped it — the one failure it reported. Declared ["llm"] by hand, which is what the script's own rule computes.
- magnific came out as ["llm"] but is an image provider (icon: "image", registered in imageRegistry.ts). It is freepik renamed by migration 160, and freepik is in the script's NO_LLM set, so the rename left that set no longer matching. Your reverse walk caught it on its first run — a fair demonstration of why the gate is worth having. Corrected to [], with magnific added to NO_LLM and a note so a re-run cannot reintroduce it.
Verified: check:provider-consistency OK (268 REGISTRY entries, 352 canonical providers, 0 registry-only exceptions, 32 catalog-only), typecheck:core clean, 137/137 across the provider/schema/serviceKinds suites, check-file-size and check:cycles green.
Thanks @Tushar49 — the design is sound and the backfill script did the heavy lifting; only its placement and freshness needed fixing.
On dashboard/memory?tab=engine the Embedding Model quick-select (and the rerank selector) built their lists from a keyword heuristic over the CHAT catalog (AI_MODELS) plus OpenRouter live discovery. Providers whose embedding models are not in that catalog never appeared — mistral, gemini, nvidia nim, groq, vercel-ai-gateway and others that serve embeddings on a standard OpenAI-compatible /embeddings endpoint — and typing such a model by hand failed at runtime with "Unknown embedding provider".
The fix is one generic mechanism rather than a list of per-provider patches: deriveEmbeddingProviderForChatProvider() turns any chat-registry entry with a /chat/completions base into an OpenAI-compatible /embeddings config, with curated EMBEDDING_PROVIDERS entries always winning; the embeddings service resolves a derived config for unknown-but-configured providers instead of rejecting them; deriveRerankProviderForChatProvider() does the same for Cohere-compatible /rerank; and both memory selectors fall back to a free-text provider/model input when no static catalog exists. No provider is special-cased by name, so adding one to the chat registry now makes it embedding- and rerank-capable here automatically.
Verified on the current release tip: merged clean, typecheck:core clean, check:cycles OK across 417 files, and 35/35 across the PR's five new suites (qdrant-quick-select-catalog, memory-provider-listings, rerank-provider-listings, embedding-generic-provider-fallback, rerank-generic-provider-fallback) plus the updated hard-session-lease-bypass-inventory and embeddings-handler.
Note: the base-red disclaimer in the description referenced #9985 against release/v3.8.50 — that window is closed and the current tip carries no open base-red, so nothing was inherited here.
Thanks @rqzbeh — deriving the capability instead of enumerating providers is the version of this that stays correct as the registry grows.
* fix(memory): measure the embedding width instead of waiting for a probe
resolveEmbeddingSource() reports dimensions: null for any source the
hard-coded registry does not describe, and a self-hosted endpoint is by
definition absent from it. Both write paths then deadlocked on that null:
- scheduleVectorUpsert called ensureReady() with the null resolution, which
declines to create vec_memories, and then ignored the {ready:false} answer
and upserted anyway -- straight into the catch, so every memory was stored,
marked needs_reindex, and never vectorized;
- reindexPending refused to embed until the width was known, and the width
could only ever come from an embedding.
Nothing surfaced it: POST /api/memory returned 200 and the health check
stayed green while rowCount stayed at 0.
The comment on EmbeddingResolution.dimensions already calls this a lazy
probe; nobody performed the probe. The upsert path holds a finished vector
when it calls ensureReady, so measure it there, and let reindex spend one
embedding up front to measure -- reusing that vector rather than paying for
it twice. withMeasuredDimensions rebuilds the signature the same way the
resolution did, identity first, so two endpoints serving the same model id
still reindex independently.
scheduleVectorUpsert now also honours a {ready:false} answer instead of
upserting into a table that is not there.
Fixes#12154
* chore(changelog): point the fragment at the real PR number
* fix(memory): extract reindex helpers so the complexity ratchet stays green
runReindexBatch grew past max-lines-per-function and cognitive-complexity
when the lazy-probe path landed. Split measure/ready/item helpers without
changing the #12154 behavior.
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
---------
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
Co-authored-by: Diego Rodrigues de Sa e Souza <diegosouza.pw@gmail.com>
* fix(sse): map normalized xhigh to max for GLM-5.x+, DeepSeek-V4+, and provider aliases
* feat(sse): support native max reasoning effort and per-model clamping
* test(sse): add unit tests for Qwen 3.8, Claude 4.7+, GPT-5.6, and 2026 reasoning models
* fix(sse): align tests and file-size split for native max effort
Keep `max` as a first-class canonical tier. Split the new sanitizer
coverage out of base-executor-sanitize-effort.test.ts so the file stays
under testCap, and update discovery/catalog/vscode assertions to expect
native max instead of the old xhigh alias.
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
* fix(sse): keep combo effort lists and drop unused collectSSE helper
Combo vscode routes still advertise the 5-tier list. Canonical `max` is
preserved in discovery (#9160) and github model metadata. Remove the
unused collectSSE helper that failed the absolute ESLint gate.
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
---------
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
Co-authored-by: Chewji <Chewji9875@users.noreply.github.com>
* feat(ui): enable React Compiler (#67)
Enable reactCompiler: true in next.config.mjs (Next 16 + React 19.2.8).
This automates memoization at build time, removing manual useCallback/useMemo
debt (591 + 283 instances respectively) and preventing stale-closure bugs.
Test results (pre-existing failures unchanged):
vitest UI: 282/295 files pass (13 fail = missing router/ReactFlow mocks)
vitest: 1805/1857 tests pass (52 fail = same pre-existing mock issues)
node:test: api/services/db all pass (except platform-specific
serviceSupervisorSpawnError — Windows spawn("ls") issue)
No new failures introduced by the compiler transform.
Optional cleanup: remove now-redundant useCallback/useMemo in hot components.
* fix(build): add babel-plugin-react-compiler peer dependency (#67)
React Compiler (reactCompiler: true in next.config.mjs) requires
babel-plugin-react-compiler as an explicit peer dependency — Next.js
declares it as optional ("*") and does not auto-install it.
Installed babel-plugin-react-compiler@1.0.0 as a devDependency.
Resolves correctly from both the project root and the next package
context (Turbopack resolution path).
* fix(ci): allowlist babel-plugin-react-compiler for React Compiler
The React Compiler peer is a real npm package (facebook/react, MIT) required
by Next 16 `reactCompiler: true`. Adding it to the anti-slopsquat allowlist
unblocks check:deps and the 6A.8 unit-test gate.
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
* fix(ci): drop unused collectSSE helper that trips ESLint
The helper was leftover from #12151 and fails the absolute
lint:json --max-warnings 0 gate on every PR that includes it.
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
---------
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
Co-authored-by: WebPerson <jonlwheat2-gif@users.noreply.github.com>
* feat(compression): make proactive context-compression threshold a live setting
The proactive compression trigger ratio was a hardcoded COMPRESSION_THRESHOLD =
0.7 in chatCore. Operators could not move compression relative to a client's own
compaction point (e.g. Codex Desktop self-compacts at ~0.85 of its window, so
the 0.7 proxy threshold always preempts the client's compaction with the
proxy's lossier one — see #8932 for what that produced before 3.8.50).
New: key_value namespace 'compression', key 'proactiveConfig',
{"thresholdRatio": 0.7}. Clamped [0.1, 0.99], 30s TTL cache, ipFilter
persistence pattern (#6131), synchronous read stays in the hot path. Default
unchanged; missing/invalid rows fall back to 0.7.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
* test(compression): cover the live proactive-compression threshold (read, validity bounds, fallback, TTL)
Locks in getProactiveCompressionRatio() (src/lib/db/compression.ts), the
key_value-backed replacement for chatCore's hardcoded 0.7:
- shipped default 0.7 when no compression/proactiveConfig row exists
- 30s TTL cache: a fresh DB write stays invisible until the TTL lapses
(clock mocked via node:test mock timers, Date API — the module keeps
its cache private with no reset hook)
- valid override read from key_value, boundary values 0.1/0.99 included
- out-of-range ratios fall back to the DEFAULT (a validity window, not
clamping to the nearest bound — matching the shipped comment)
- broken JSON / non-numeric thresholdRatio: 0.7, without throwing
Guard verified by mutation: switching the window to clamping fails the
out-of-range case.
---------
Co-authored-by: root-cli (Hermes ops) <info@livewellwith.us>
Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
Co-authored-by: Diego Rodrigues de Sa e Souza <diegosouza.pw@gmail.com>
* fix(kiro): do not permanently ban on 'User is not authorized to make this call'
* test(kiro): regression cover the 403 'User is not authorized' non-ban classification
---------
Co-authored-by: Deftera186 <Deftera186@users.noreply.github.com>
Co-authored-by: Diego Rodrigues de Sa e Souza <diegosouza.pw@gmail.com>
* feat(usage): devin-cli agentic quota + openrouter credits in Provider Limits
Two provider families with live quota APIs were missing from the Provider
Limits dashboard because their list entries were absent:
- devin-cli: new usage leaf querying the Codeium seat-management Connect API
(exa.seat_management_pb.SeatManagementService/GetUserStatus, protobuf over
POST with the raw `Basic <token>-<token>` auth header the CLI itself uses).
Surfaces the plan name plus daily/weekly agentic quota percentages with
reset timestamps from the GetUserStatus plan_status payload, via a minimal
hand-rolled protobuf encoder/reader (no proto dependency warranted for two
fixed messages).
- openrouter: the /key + /credits quota fetcher (#6842) was already wired
into the dispatcher but gated out of the bulk sync — add it to
USAGE_SUPPORTED_PROVIDERS and PROVIDER_LIMITS_APIKEY_PROVIDERS so key
limits and account credits actually surface.
* fix(build): externalize tiktoken so tiktoken_bg.wasm resolves at runtime
The vendored ChatGPT Web connector v4.0.7 (#12181) imports tiktoken
(get_encoding) at module level. tiktoken's node build reads
tiktoken_bg.wasm via a __dirname-relative fs.readFileSync during import;
when Next bundles the package the wasm asset is not traced into the server
chunk, and page-data collection for every route reaching the tokenizer
(e.g. /api/providers/[id]/chatgpt-web-codex-doctor) aborts with
"Missing tiktoken_bg.wasm" — breaking the whole standalone build.
Externalize it like the other runtime-resolved native/wasm packages
(sql.js, sqlite-vec, better-sqlite3): the require stays at runtime, where
node_modules/tiktoken/tiktoken_bg.wasm resolves normally.
* fix(openrouter): /credits balance survives a /key failure
OpenRouter is credit-based, not subscription-based: the authoritative
remaining-credits signal is GET /api/v1/credits (total_credits -
total_usage, the documented "get remaining credits" endpoint), while the
/key limit fields are optional per-key caps that most accounts never set.
fetchOpenrouterQuota previously treated /key as mandatory — any /key
failure (429 rate limit, transient error, unexpected shape) discarded the
whole payload and the Usage dashboard showed "OpenRouter (usage endpoint
unreachable)" even though /credits was reachable. Now:
- /key unavailable + /credits OK → credits-only quota (creditBalance =
total_credits - total_usage) instead of null
- /key 401/403 alone no longer means an invalid token; only a double
auth-rejection (both endpoints) does
- null is returned only when both endpoints fail, and the dashboard label
reflects that ("credits endpoint unreachable")
* fix(openrouter): render AI Credits as a USD credit count in Provider Limits
The Provider Limits card's dollar renderer only activates on
isCredits/creditCount rows (QuotaCardExpanded), but openrouter went through
parseGeneric — which drops `currency` and never sets those flags — so the
credits balance rendered as a meaningless "100% left" (the unlimited-credits
row is always 100%) instead of the actual credit count.
Route openrouter's `credits` quota through buildCreditsQuota() like the
DeepSeek/AgentRouter credits rows: label "AI Credits", dollar-formatted
balance. Free-tier request windows keep the generic percentage treatment.
* fix(usage): document DEVIN_SEAT_API_URL and split quota parsers
Keep fetchOpenrouterQuota and decodeProtoFields under the complexity
ratchets, and add the seat-management URL to the env/docs contract.
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
* test(usage): drop duplicated GLM quota-ordering test in provider-limits-ui
* test(usage): drop stale openrouter ACCEPTED_DIVERGENCE
OpenRouter is now in both USAGE_FETCHER_PROVIDERS and
USAGE_SUPPORTED_PROVIDERS, so the recorded aggregator divergence
is no longer real. Add the changelog fragment.
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
---------
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
Co-authored-by: Diego Rodrigues de Sa e Souza <diegosouza.pw@gmail.com>
* perf(compression): memory and OOM mitigations for large payload hashing and token estimation
* fix(compression): implement getMemoStats observability for result memo (#7847)
Adds the missing memo observability layer referenced by
tests/unit/compression/oom-memo-memory.test.ts and the monitoring API:
- resultMemo.ts: lifetime hit/miss counters + bounded time-ordered ring
buffer (10k entries, ~90KB) powering 1m/5m/15m/1h hit-rate windows;
getMemoStats() reports size/capacity/hits/misses/hitRate + windows.
- memoLookup() tags served results with stats.memoHit = true.
- clearMemoStore() also resets counters and the ring.
- compression/index.ts re-exports getMemoStats for the monitoring route.
- types.ts: optional memoHit field on CompressionStats.
- New GET /api/monitoring/compression route exposing the stats snapshot
(lightweight, no DB) for operators to track cache-hit efficiency.
* fix(compression): align memo contract with upstream #11727 — return caller object, reset lookup counter in clearMemoStore
* fix(compression): restore unwrapEventEnvelope in stream payload collector summaries
The OOM-mitigation commit accidentally replaced unwrapEventEnvelope(evt.data)
with asRecord(evt.data) in the summary builders and live push, breaking
translate-mode {event, data} envelope unwrapping (clientPayload type detection)
and failing 2 stream-payload-collector tests. Restored upstream semantics;
kept the jsonLength OOM optimization as the only delta in this file.
* refactor(compression): break down writeValue and writeEncodedString to pass complexity ratchets
Refactors jsonSha256 internal helpers (writeValue, writeEncodedString)
into small, single-responsibility sub-functions under the complexity
threshold (max cyclomatic 15, max cognitive 15). Preserves exact
JSON.stringify parity, circular reference guards on both arrays and
plain objects, and escape behavior (all 530 relevant tests pass).
* test(compression): make oom-memo heap assertion robust without expose-gc
The CI unit-test shard runner does not pass --expose-gc, so global.gc is
undefined and heapUsed can still momentarily hold GC-pending transients
(observed 53.4 MiB after a 3MiB body). Gate the retained-heap assertion
on forced collection being available (3 forced cycles for array buffers)
instead of skipping it silently, and keep it fully active when
--expose-gc is present.
* fix(compression): restore worker-pool offload path in runCompressionAsync
The OOM-mitigation refactor dropped the isCompressionWorkerEligible /
runCompressionInWorker dispatch at the top of runCompressionAsync, silently
removing the base's worker-thread offload for eligible large payloads.
Restore the block exactly as on release/v3.8.51, ahead of the result-memo
path, keeping the memoization and hashing improvements intact.
* docs(api): document GET /api/monitoring/compression and log route errors via pino
Add the new monitoring endpoint to docs/openapi.yaml following the
neighboring System entries, and replace the route's console.error with
the repo-standard pino logger.
* fix(skills): regenerate omni-resilience and add changelog fragment
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
---------
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
Co-authored-by: Andrian Balanescu <AndrianBalanescu@users.noreply.github.com>
Co-authored-by: Diego Rodrigues de Sa e Souza <diegosouza.pw@gmail.com>
* feat(combos): add universal handoff feature flag
Add a default-enabled runtime flag that lets operators disable universal context handoffs globally without changing existing combo configuration or requiring a restart.
* fix(i18n): seed the universal-handoff flag description key across locales
---------
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
- test:scoped:full (documented in the script header since #9143 but never wired) rebuilds
config/quality/test-impact-map.json and then selects.
- select-impacted-tests.mjs gains --stdin so --staged selects from the index; the git-diff
path only ever saw commits, so staged-only runs silently fell back to the heuristic.
- Loader parity with npm run test:unit / quality.yml TIA step (#6787): tests/unit/dashboard/**
under --import tsx (CJS transform), tests/unit/serial/** at --test-concurrency=1, the rest
under tsx/esm. The single tsx/esm invocation false-redded every dashboard test the map
selected ("Unexpected token 'export'").
- CONTRIBUTING.md → Running Tests documents the three modes and the fail-safe exit 1.
Refs #8084
* fix(sse): forward the upstream's real trailing usage in passthrough; estimate only at flush
#12151 injected estimated usage into the finish chunk and dropped the real
trailing usage block that genuine OpenAI upstreams send afterwards — metered
clients got estimates instead of real token counts. The estimate now leaves
via a canonical usage-only chunk at flush, only when the upstream stayed
silent; a real trailing block is forwarded verbatim and wins. The tool_calls
finish_reason normalization now materializes its own rewrite (it piggybacked
on the removed finish-time rewrite), and the dead collectSSE helper goes with
it (subsumes #12324).
* fix(i18n): seed the radarPage limits/training keys the #12320 UI already consumes
RadarCatalogTable.tsx references radarPage.colLimits / trainsOnPrompts /
trainsOnPromptsHelp but #12320 never added them to en.json, so the EN
fallback could not resolve the __MISSING__ markers across 42 locales.
Real translations for pt-BR and vi; the rest resolve via the EN fallback.
* fix(sse): carry the chat stream id into flush-time synthetic chunks
The estimated usage-only chunk emitted at flush used passthroughResponsesId,
which is only ever set on the Responses path — on the chat path the synthetic
chunk shipped id: null, breaking the string-id invariant pinned by
stream-numeric-ids.test.ts. Track the upstream chat-completion id in the
passthrough loop and reuse it (falling back to the Responses id, then a
generated one). Sibling sweep: 74 files importing utils/stream — 603/603.
The provider plugin manifest already exposed usage-fetch (40 providers, #11903); this publishes the second capability, usage-supported, so integrators can tell without reading TypeScript whether the server usage routes accept a provider. #11903closed#11722 after shipping only half of it and said so at the time — this is the follow-up it promised.
The two scopes genuinely differ and the docs now say how: usage-fetch resolves on id or alias (the dispatcher accepts both), usage-supported on id alone, because the runtime guard does a plain USAGE_SUPPORTED_PROVIDERS.includes(providerId) with no alias resolution. 42 providers carry both tags, 4 carry only usage-fetch (opencode, opencode-zen, openrouter, xai) and 3 only usage-supported (adobe-firefly, firefly, xiaomi-mimo-token-plan) — 7 measured differences, so neither tag implies the other. No list mutation, no new route, schemaVersion stays 1.
USAGE_SUPPORTED_PROVIDERS moved out of src/shared/constants/providers.ts into an import-free leaf at open-sse/services/usage/supportedProviders.ts, keeping the manifest's import graph light — the same move fetcherProviders.ts got in #11903, landed on the correct side of the workspace boundary.
Base note: the branch forked 46 commits before kilocode joined the list, so a wholesale take of its providers.ts would have silently dropped that id. Verified against the current release tip before merging — both sides hold the same 46 ids, nothing lost.
Verified on the current tip: typecheck:core clean, check:cycles OK across 417 files (the import-free-leaf claim holds), and 63/63 focused tests across provider-plugin-manifest, usage-fetcher-registration-coverage, adobe-firefly and agentrouter-quota-dashboard-rendering.
Thanks @maxmad64bis — and for finishing the half of #11722 that was left open rather than letting it sit.
onnxruntime-node only ever moves paired with @huggingface/transformers (already
frozen, #9962/#4050) — a solo bump breaks the single-copy ABI contract test and
reds every production-group PR. eslint-plugin-react-hooks stays pinned to 7.0.1
by a contract test until the 7.1.1 rule set is adopted in its own PR (the
#12146 migration completed today, so that adoption is now unblocked).
* fix: resolve compression worker file using runtime anchors instead of import.meta.url
Replace workerUrl() function that used import.meta.url with resolveWorkerFile()
function that uses runtime anchors (process.cwd() and process.argv[1]) to locate
the worker file. This fixes webpack module resolution in Next.js standalone
bundles where import.meta.url is replaced with a stub pointing to build machine
path.
Also update Dockerfile to copy required worker-related scripts and adjust
npm install flags for better compatibility.
* fix(docker): restore base Dockerfile — keep npm ci --ignore-scripts supply-chain guard
Revert every Dockerfile change from this branch back to release/v3.8.51:
the branch dropped --ignore-scripts (reopening install-time script
execution for all transitive deps), swapped the reproducible npm ci for
npm install, invoked the nonexistent 'npm approve-scripts' command, and
broke the better-sqlite3 smoke test with a stray space in ':memory: '.
The worker-file fix does not need any Dockerfile change.
* test(compression): export runtime-anchor helpers and cover worker-file resolution
firstAncestorWith's doc already claimed 'exported for tests' without the
export; export it together with resolveWorkerFile and add unit coverage
for the runtime-anchor resolution: cwd anchor, dirname(argv[1]) anchor,
bounded walk-up (8-level cap boundary), prod-first .js-over-.ts ordering,
dev .ts fallback and the fail-open cwd fallback when nothing exists.
Fixtures live in mkdtemp sandboxes only — the repo tree is never touched.
---------
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
The weight table said stability accounts for "low latency stdDev / error rate". Grep errorRate in scoring.ts and you find it declared on ProviderCandidate and read nowhere — while combo.ts pulls 24 hours of usage history behind a ten-sample floor, falls back to real-time metrics, and hands every candidate an errorRate the scorer ignores. Two candidates, one failing 1% of calls and one failing 99%, scored identically at 0.459486.
This declares reliability as a sixteenth factor: 1 - failureRate, using the same formula, field precedence and rate-bounding speedRanking.ts already applies, so a corrupt reading means "nothing observed" rather than "fails every call". It ships at weight 0, leaving the ranking unchanged to the digit — the honest default, since which weight this deserves is a product call backed by traffic the author does not have. Two declared-but-silent factors already ship (cacheAffinity, resetWindowAffinity), so the pattern is not new. The stability row now describes what that factor actually computes: latency variance.
The rest is the mechanical 15 → 16 across nineteen documents and the forty-two llm.txt mirrors — sourced from check:docs-counts rather than a grep, the first real use of the gate #12316 extended.
Protected-surface note: this PR touches AGENTS.md, llm.txt and its 42 mirrors, and skills/omni-combos-routing/SKILL.md. Every changed line in those 45 files is a digit substitution and nothing else — masking all digits makes the removed and added lines identical, with no sentence added, removed or reworded. Reviewed and approved on that basis before merging.
Verified on the author's rebased head: check:docs-counts green (the gate that now enforces the count this PR moves), typecheck:core clean, and 71/71 focused tests across scoring-reliability-factor, combo-scoring-weights-schema-coverage, check-docs-counts-sync, lkgp-enabled-context, intelligent-routing-options and the combo-matrix auto integration suite.
Thanks @maxmad64bis — shipping the factor at weight 0 and saying plainly that the weight is someone else's call is the right way to land this.
* feat(providers): manual "Clear cooldown" action in the cooling panel
The persisted 429 cooldown (provider_connections.rate_limited_until) is
OmniRoute's local lesson, not upstream truth. When a quota has already
refreshed upstream (daily/weekly reset, provider-side fix), the only
automatic clear paths — Test-button success or Edit-modal key
re-validation — still require an upstream round-trip, so the user waits
out a bench that is already stale.
Adds a per-row "Clear cooldown" button to CoolingConnectionsPanel that
PUTs rateLimitedUntil: null (the route applies backoff reset defaults),
optimistically drops the bench, and refetches. The next request becomes
the real test of the key.
- useProviderConnections: handleClearCooldown + clearingCooldownId
(in-flight guard mirrors the retestingId pattern)
- CoolingConnectionsPanel: optional onClearCooldown/clearingCooldownId
props; button hidden for id-less rows, disabled per-row while clearing
- ProviderDetailPageClient: wires the new handler through
- i18n: en.json keys (clearCooldown, cooldownCleared,
failedClearCooldown, ...) with providerText fallbacks
Tests: CoolingConnectionsPanel.test.tsx — click fires handler with the
row id, disabled + silent while in flight, per-row independence,
read-only when handler omitted, no button without connection id,
renders nothing when empty. Pre-existing
tests/unit/ui/CoolingConnectionsPanel.test.tsx stays green.
* fix(dashboard): dedupe clear-cooldown i18n keys and extract the row button
The providers namespace already carried an (orphaned) clearCooldown /
cooldownCleared / failedClearCooldown key trio, so the new feature keys
re-declared them as duplicate JSON keys ~1200 lines apart. JSON.parse is
last-wins, which silently shadowed the older values and broke ICU
placeholder parity in every locale (EN lost {model} while all 42
translations still carry it). Rename the feature's five keys to a
connection-scoped family instead:
clearConnectionCooldown / clearConnectionCooldownInProgress /
clearConnectionCooldownTitle / connectionCooldownCleared /
failedClearConnectionCooldown
Also extract the per-row action into ClearCooldownButton so the panel
body stays inside the max-lines-per-function ratchet (was 83/80).
* feat(dashboard): mirror the clear-cooldown keys into all 42 locales
Adds the five connection-cooldown keys to every non-EN catalog with the
English value as the runtime fallback (fill-missing-from-en semantics),
and real translations for pt-BR and vi so their strict parity suites
stay meaningful:
pt-BR: Limpar cooldown / Limpando… / Cooldown limpo — a conexão voltou
ao roteamento / Falha ao limpar cooldown
vi: Xóa thời gian chờ / Đang xóa… / Đã xóa thời gian chờ — kết nối
đã tham gia lại định tuyến / Không thể xóa thời gian chờ
---------
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
* fix(guardrails): pass providerId to getResolvedModelCapabilities in checkComboVision (#12112)
Signed-off-by: Minxi Hou <houminxi@gmail.com>
* chore(quality): register combo-vision providerId test in the stryker tap set
---------
Signed-off-by: Minxi Hou <houminxi@gmail.com>
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
* fix(combos): send null to clear an agent feature instead of omitting it
PUT /api/combos/[id] merges its body over the stored record, so an omitted
field means "leave unchanged". The combos editor deleted a cleared agent
feature from the payload, so unchecking context cache protection -- or
emptying the system message or the tool filter -- never persisted: the old
value survived the merge and the editor reopened with the toggle still on.
updateCombo already deletes any key explicitly set to null, which is how
description and context_length are cleared in the same save handler. Use the
same shape for the three agent fields, and make them nullable in
updateComboSchema so the null survives validation.
The clearing logic moves into comboAgentFeatures.ts so it can be tested
directly, matching comboQuotaOnlyFallback.ts next to it.
Fixes#12158
* chore(changelog): point the fragment at the real PR number
When fetching version metadata from npm registry or GitHub APIs, if the remote connection stalls during stream reading, the 10-second AbortController timer aborts the request signal but the underlying body reader stream was not listening to the abort signal. This caused readBoundedJson reader.read() loop to hang until external socket close.
Now readBoundedJson listens to AbortSignal abort events, triggers reader.cancel(), and releases locks immediately on abort.
The Radar feed reports per-model rate limits and whether a provider says it may train on your prompts. Both fields are on RadarMergedEntry and the catalog table rendered neither — grep them in RadarCatalogTable.tsx and the only hits were the type declaration. The training flag is the one that stings: freeModelCatalog.ts documents it as "Surfaced in the UI", a promise the UI did not keep, and thirteen catalog entries carry it today.
Adds a Rate limits column and a badge in the ToS cell when a provider discloses training. No new data, no request, no API change.
Two judgement calls worth keeping: a limit of zero renders as 0/min rather than formatTokens' "rate-only" (right for a monthly budget, nonsense for a ceiling where zero is a real and alarming fact), and the badge condition is === true, since an absent training statement is not a guarantee.
Validation note — read before trusting the green: this PR's own suite (tests/unit/dashboard/radar-catalog-table-limits-training.test.tsx, 14 cases) could NOT be run locally. Vitest fails to resolve react18-json-view, which is declared in package.json and package-lock.json but is not present in this machine's node_modules; two pre-existing .test.tsx files in the same directory fail identically, so the cause is environmental and not this PR. The blocking test-vitest CI job runs npm ci and will execute it.
What was verified locally, in a combined batch worktree with all 11 PRs of this batch: 174/174 node-runner focused tests, typecheck:core clean, complexity 2706/3218, cognitive-complexity 1221/1437, check:cycles and check:docs-counts green, plus both CI i18n gates for the 42-locale pass — check-ui-keys-coverage PASS (all 42 locales at or above 65%) and check-ui-value-drift PASS against the release tip.
Thanks @maxmad64bis.
Free Provider Rankings sorted by the top model's Arena score, so a provider stayed first even if it failed every call — the reliability column from #11546 already showed what each one actually served, but ordering ignored it.
Adds an opt-in ?sortBy=reliability (API) and a "Most reliable first" toggle (page) sharing one comparator in freeProviderRankingsUsage.ts: measured providers first by successRate desc with ELO on ties, then unmeasured in their incoming order. The default is unchanged and locked by tests. successRate is null below MIN_USAGE_REQUESTS = 5 (existing, never zero), and ordering runs before slice(0, limit) so limit counts in the requested order. The toggle composes with the existing sortTypeFirst/groupByType grouping — a stable sort keeps reliability order within each group.
Opt-in is the right default here: the page is for discovery, including providers never called.
13 tests across three files (6 new for the comparator in isolation, plus filter and route coverage including unknown sortBy → 400 and the default path staying off call_logs).
Verified in a combined batch worktree with all 11 PRs of this batch: 174/174 focused tests, typecheck:core clean, complexity 2706/3218, cognitive-complexity 1221/1437, check:cycles and check:docs-counts green. The 42-locale i18n pass was checked with the CI gates: check-ui-keys-coverage PASS (all 42 locales at or above 65%) and check-ui-value-drift PASS against the release tip.
Thanks @maxmad64bis.
docs/reference/FREE_TIERS.md said its numbers were "gathered by web research (confidence tagged per row)". No entry carries one: grep -c confidence on the catalog returns 0, the type does not declare the field, and the API serves nothing of the sort. A reader looking for "how much can I trust this figure" was pointed at a per-row signal that never existed.
Replaced with the two counts the data actually supports — 7 of 446 entries carry hardStopGuaranteed (the field with the strictest sourcing rule in the repo: set only when the provider's own terms document that exceeding the free allowance refuses the request, source in a comment, never defaulted to true) and 13 carry a prompt-training disclosure. check:docs-counts reads both from the catalog at runtime, as required claims, so a reworded or deleted sentence fails rather than passing as "no claim in this file".
The PR deliberately does not add a confidence field — curating one is a product call, and it says so instead of inventing it.
Reconciled on merge: #12316 landed the gate extension underneath, so scripts/check/check-docs-counts-sync.mjs and its test took the tip's side plus this PR's own required-claim additions. Verified afterwards: check:docs-counts green, 48/48 across check-docs-counts-sync and free-catalog-no-confidence-field.
Thanks @maxmad64bis — checking the gate against a number it should reject (7 swapped for 99) is the right way to prove a gate works.
* fix(resilience): per-model 402 on a passthrough gateway no longer terminalizes the whole connection
402 variant of #3027. Passthrough/gateway providers that multiplex many
models behind one credential (kilo-gateway, ollama-cloud, etc.) can 402
on a single PAID model while free models on the same key remain
perfectly usable. Previously any 402 unconditionally set the connection
to a terminal `credits_exhausted` status, which is never auto-recovered
without an operator reset — taking out every remaining model on that
provider, amplified further inside combo routing (measured: one 402
removed 9 of 14 fallback targets in a real combo, dropping success rate
from 98.3% to 74.2% on a fixed load test per the issue report).
Root cause (matches the issue's own analysis):
1. resolveTerminalConnectionStatus() returned "credits_exhausted" for
ANY status === 402, with no per-model/passthrough check.
2. The generic per-model lockout gate (404/429/>=500) excluded 402.
3. The #3027 403-branch is gated on `!terminalStatus` — since (1) already
resolves a terminal status for any 402 before that branch runs, simply
adding 402 to its condition alone would not have fired.
Fix:
- resolveTerminalConnectionStatus() now takes isPerModelQuotaProvider and
skips the connection-wide terminal path for a bare `status === 402`
when true, letting it fall through to the per-model lockout branch
instead. An explicit result.creditsExhausted (a provider's own
classification, independent of HTTP status) is untouched and remains
unconditionally terminal.
- Extended the existing #3027 per-model lockout branch to also handle
402 (reason "credits" vs "forbidden" for 403), reusing the same
cooldown/lockout machinery and log format.
- Single-credential (non-passthrough) providers are unaffected:
isPerModelQuotaProvider is false there, so a 402 still terminalizes
the connection as before — that behavior is deliberate for prepaid
API keys (#5239 / #10616).
Also checked the issue's 4th root cause (terminal statuses never
auto-recovering) against the current codebase: connectionRecovery.ts
already has a 30-minute credits_exhausted reprobe
(isCreditsExhaustedReprobeCandidate) that the issue's report — filed
against v3.8.49 — didn't account for. The other two files it names
(rateLimit.ts's clearStaleCrashCooldowns, tokenHealthCheck.ts's
OAuth-refresh skip) legitimately exclude credits_exhausted for
unrelated reasons and are not bugs. Moot regardless: this fix prevents
credits_exhausted from being set at all for the passthrough case, so no
recovery wait is needed in the first place.
Tests: tests/unit/auth-passthrough-per-model-402-12242.test.ts, modeled
on the existing #3027 precedent test (real DB-backed integration test
via auth.markAccountUnavailable). Covers: paid-model-only lockout with
free model unaffected, a subsequent free-model request succeeding after
a sibling paid-model 402, single-credential 402 still fully terminal,
and no connection-wide backoff escalation on repeated 402s.
Verified:
- node --import tsx/esm --test tests/unit/auth-passthrough-per-model-402-12242.test.ts: 4/4 pass
- All related pre-existing tests (auth-ollama-cloud-per-model-403-3027,
auth-terminal-status, openrouter-free-model-credits-exhausted,
vertex-passthrough-model-lockout, 10347-embed-402-cooldown): 27/27
pass, no regressions
- npm run typecheck:core: 0 errors
- npm run check:cycles: no cycles
- eslint (auth.ts + new test file, with project suppressions): 0 errors
Fixes#12242
* chore(quality): register 402 per-model test in stryker tap and de-ratchet auth.ts
- stryker.conf.json: add tests/unit/auth-passthrough-per-model-402-12242.test.ts
to tap.testFiles in its alphabetical slot
- auth.ts: extract the #12242 connection-wide 402 decision into the pure helper
isConnectionWideCreditsExhausted() so resolveTerminalConnectionStatus stays
within the cyclomatic ratchet (file back to the base's 11 violations)
---------
Co-authored-by: OmniRoute Dev <dev@local>
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
docs/routing/AUTO-COMBO.md documented four mode packs; six ship. It printed 0.14 where modePacks.ts says 0.1333. And nothing was watching: four documents stated a scoring-factor count and check:docs-counts covered none of them. Wiring them up turned the gate red on seven real drifts — ARCHITECTURE.md and REPOSITORY_MAP.md at "9-factor" (code: 15) and "4 mode packs" (code: 6), RESILIENCE_GUIDE.md and SKILL.md at 13, AUTO-COMBO-GUIDE.md at both 5 and 13. ARCHITECTURE.md did not merely have the wrong number: it named nine factors that are not the engine's, and its four "mode packs" were the auto/* request prefixes.
A product fact fell out of writing the table: no pack sets quality, and applying a pack replaces the weight map wholesale (weights = pack in engine.ts, not a merge), so quality carries 0.03 by default and normalizes to 0 under any pack — pick a mode pack and the observed-quality signal stops voting. Documented, not changed.
The gate reads pack names from the module through the tsx subprocess that already reads every other code-derived count, matching the three spellings the docs actually use; on the reference document a missing claim now fails rather than passing. The dashboard was behind too (four of six packs offered); the count is dropped from the strategy label rather than corrected, since nothing reads selector labels and a right-today number goes stale unnoticed.
Verified in a combined batch worktree: 174/174 focused tests across all 11 PRs of this batch, typecheck:core clean, and check:docs-counts green with the four newly-wired documents.
Thanks @maxmad64bis — finding the quality-under-a-pack behaviour while writing a docs table is the kind of thing a table is for.
Two of the fifteen factors calculateScore applies could not be set by anyone. scoringWeightsSchema is a plain z.object, so zod strips what it does not name: PUT a combo with connectionDensity and you get a 200 back with nothing saved, and normalizeScoringWeights then reads the gap as a deliberate zero — switching off anti-concentration and the quality signal. DEFAULT_INTELLIGENT_WEIGHTS, the dashboard's own copy, missed the same two and every non-zero value differed from the engine's; summing to 1.05, validateWeights rejected them outright.
This adds the two keys to both lists and takes the dashboard defaults from DEFAULT_WEIGHTS. The scorer is not touched.
One behaviour change, and it is the point: a combo whose stored weights omitted the two keys was running with them at zero and the other thirteen renormalized upward. It now uses the engine's distribution (quota 0.1549 → 0.1429, health 0.1740 → 0.1605) and a test pins those numbers.
Left alone and documented rather than widened: rounded percentages now total 101% (six factors at 4.76% each render as 5%), and five stale .default() values in the schema that only bite when a config omits the key.
Verified in a combined batch worktree: 174/174 focused tests across all 11 PRs of this batch, typecheck:core clean, complexity 2706/3218, cognitive-complexity 1221/1437, check:cycles and check:docs-counts green.
Thanks @maxmad64bis — the red-before-green note (7 of 8 failing, and naming the one that passes on purpose) is exactly the evidence that makes a behaviour change reviewable.
A passthrough stream could end with no usage even though the client asked for it via stream_options: {include_usage: true}, so providers that do not meter always showed 0 tokens. The fix estimates usage at the finish marker when the upstream stays silent (flagged estimated: true) and drops any duplicate trailing usage chunk so the client never sees two.
open-sse/utils/stream.ts:1982,1749 · open-sse/utils/usageTracking.ts:651,664
Six cases: the predicate (finish without usage but with content, trailing valid, empty response, tool-only) plus two SSE harness cases through createSSEStream passthrough.
Note on base: this branch forked 442 commits back and carried a base-red marker for #12109, which is now closed — the release tip has no open base-red issue. It merged cleanly against the current tip regardless.
Verified in a combined batch worktree: 174/174 focused tests across all 11 PRs of this batch (this PR's stream-passthrough-usage-estimation suite included), typecheck:core clean, check:cycles and check:docs-counts green.
Thanks @maxmad64bis.
With freeAccessPolicy: "strict" the read-only candidate listing silently dropped rows, so an operator could not tell "no free allowance left" from "the quota fetcher is broken" — in a listing whose own module header promises a candidate the routing path would skip "is never dropped". #9133 settled the same question for the resilience filter via a skip opt-out; the zero-cost guard never got one.
It gets it now: the guard is disabled for the inspector build exactly as the resilience filter already is, and every candidate carries freeAccessExclusion — null when satisfied, otherwise one of seven named reasons. The last three (exhausted, state-unknown, no-connection) are the point: they used to look identical because the row just disappeared. STRICT_ZERO_COST.md documents what each asks the operator to do.
Dispatch is untouched and a test pins that. The three existing guard suites were not modified — their 31 cases are the net and still pass. excludeTosAvoid still drops candidates without a reason; documented as a separate question rather than widened into here.
Verified in a combined batch worktree: 174/174 focused tests across all 11 PRs of this batch, typecheck:core clean, complexity 2706/3218, cognitive-complexity 1221/1437, check:cycles and check:docs-counts green.
Thanks @maxmad64bis — the reason table and the honest note about the order change (freshness before status) made this easy to review.
GET /api/free-tier/summary could answer from a Radar overlay built 2026-08-02 while the release ships a catalog curated 2026-08-30 (FREE_CATALOG_CURATED_AT) — totals computed from older data, still tagged catalogSource: radar-overlay. The route now refuses any overlay built before the shipped catalog and falls back to that catalog through the operator's local state.
Tightens #11550 using the generatedAt persisted by #11435.
Verified in a combined batch worktree: 174/174 focused tests across all 11 PRs of this batch (this PR's free-tier-summary-radar-overlay suite included), typecheck:core clean, check-file-size, check-changelog-integrity, check:cycles and check:docs-counts green.
Thanks @maxmad64bis.
usage/fetcherProviders.ts exists, in its own words, "so the registration list can't drift from the dispatcher's switch statement". It drifted: #8006 added adobe-firefly and firefly to the dispatcher and to USAGE_SUPPORTED_PROVIDERS but not to this list, so the connection UI advertised usage support while the provider-plugin manifest, genericQuotaFetcher and the free-access quota cache all reported no fetcher — for two ids getUsageForProvider would happily serve.
Declaring them is what makes the balance actually get fetched (registerGenericQuotaFetchers wires a generic fetcher per declared id, and resolveFreeAccessState stops returning early), which the PR states plainly rather than burying as a side effect.
The test turns the docstring's prose invariant into enforcement: it reads the dispatcher's cases from source and compares both directions, and records each accepted difference against USAGE_SUPPORTED_PROVIDERS with a reason plus a staleness check, so the next drift can't hide among them. xiaomi-mimo-token-plan is left flagged as a real gap rather than widening the PR.
Verified in a combined batch worktree: 174/174 focused tests across all 11 PRs of this batch, typecheck:core clean, complexity 2706/3218, cognitive-complexity 1221/1437, check:cycles and check:docs-counts green.
Thanks @maxmad64bis.
* fix(usage): console-aware Token Plan guidance and subscription hint on bailian 401
The personal Token Plan is sold through two consoles with different portals,
gateway hosts and login tickets. Two operator-facing messages ignored the split:
- The quota guidance always said 'get the cookie at home.qwencloud.com', even
for connections served by the Alibaba Model Studio console — following it
verbatim produces a cookie the gateway rejects (console mismatch →
BailianGateway.Login.NotLogined). The guidance now derives the console from
the provider via resolveConsoleSite, matching what the fetcher will do with
the pasted cookie.
- Key validation mapped upstream 401 to a bare 'Invalid API key'. An expired
Token Plan subscription produces the exact same upstream 401 (observed live
2026-09-01: subscription ended 08-23, the working key started failing), so
the message now names the subscription as a cause worth checking.
* test(providers): align the remaining bailian 401/403 message pins to prefix match
search-provider-validation.test.ts pinned the exact 'Invalid API key' string for
the bailian validator; the message now also names an expired Token Plan
subscription. Same property asserted (401/403 => invalid), prefix match.
The nightly headroom monitor flagged fileSize 🟡 permanently because the worst
frozen file was src/app/docs/lib/openapi.generated.ts — frozen at its emitter's
exact output size in #12212, i.e. ~0% headroom BY CONSTRUCTION (growth is
policed by conscious re-freezes, never by editing the module). Same class:
open-sse/vendor/** (upstream code nobody slims by hand).
isMonitorExemptFile() excludes .generated. modules and vendor/ paths from the
monitor's worst/near-cap accounting only — check:file-size itself still
enforces both. The fileSize row now points at the worst HUMAN-EDITABLE frozen
file (currently tests/integration/skills-pipeline.test.ts at 4.5%, a true
early warning: its baseline note already requires a split rationale for any
further growth).
Refs #12149
* feat(sse): add it/ru/zh caveman output instructions
* fix(sse): expose the dormant terse-prose translations through the catalog
* feat(sse): translate less-code to es/de/fr/it/ru/zh
* feat(sse): translate ponytail to es/de/fr/it/ru/zh
* feat(sse): translate i-have-adhd to es/de/fr/it/ru/zh
* test(sse): anchor wave-2 output-style translations to their own language
* feat(dashboard): offer every output-style language in the default-language selector
* feat(sse): let autoDetect pick the output-style instruction language
* docs(compression): consolidate the output-style tables and record full language parity
* fix(sse): trust finish_reason:length/max_tokens over the reasoning-ratio heuristic in response quality validation
A truncated response with empty content and reasoning_content present was
only rejected by validateResponseQuality() when reasoning consumed >=90%
of completion_tokens. A response truncated at a lower ratio (e.g. 63%)
passed through as "valid" even though the caller received no usable
content and finish_reason was explicitly "length" (or the alternate
"max_tokens" naming some providers use) -- an unambiguous truncation
signal the validator wasn't reading. Reproduced live against
nvidia/nemotron-3-super-120b-a12b: content:null, finish_reason:length,
reasoning_tokens 645/1024 (63%).
Trust finish_reason directly when it's reported, falling back to the
existing token-ratio heuristic only when it isn't. Does not affect the
deliberate-tiny-probe case (e.g. max_tokens:1 connectivity pings) --
those never produce reasoning_content, so the branch this change is in
doesn't run for them.
* docs(changelog): add fragment for #12262
---------
Co-authored-by: brick30llc-ctrl <admin@brick30.com>
Completes the contributor profile: #12198 stopped OmniRoute from assembling the standalone bundle, but Next was still asked to emit one. Making output: "standalone" conditional on OMNIROUTE_BUILD_PROFILE=contributor removes the standalone tracing pass itself, which is where the remaining time went.
Default builds are unaffected — the flag is read from the env at config load and is false everywhere except the contributor profile, so tests/unit/next-config.test.ts still observes output === "standalone" (18/18 green across contributor-build-script, next-config and build-profile-stubs).
Reconciled on merge: CONTRIBUTING.md and scripts/build/backendOnlyPages.mjs already carried this stack's earlier steps on the tip, so both took the tip's side; only the next.config.mjs conditional and its test are this step's delta.
Thanks @rafacpti23 for splitting this into four reviewable steps — it made the whole stack easy to reason about.
Contributor builds only need compilation to type-check; pulling src/instrumentation.ts drags the whole startup graph (DB boot, model-catalog warm, quota fetchers) into the build. stubContributorInstrumentation() swaps both entrypoints for no-ops before next build and hands them to the existing restoreDashboardPages() path afterwards, reusing the same {file, original} shape and the SIGINT/SIGTERM handlers already registered in that block.
Reconciled on merge: the stub originally wrote 'export async function register() {}' into both files, but src/instrumentation-node.ts exports registerNodejs() (register lives in src/instrumentation.ts). Harmless in practice — instrumentation.ts is its only importer and is stubbed at the same time — but the stub misstated the file's contract, so it now emits the right symbol per file and the test asserts it.
Verified: contributor-build-script 3/3 green.
Thanks @rafacpti23.
Makes the contributor profile actually fast: with OMNIROUTE_BUILD_PROFILE=contributor the build stops after next build instead of copying docs/ and running assembleStandalone, which is the expensive half and produces an artifact contributors never ship. Adds isContributorBuild() next to the existing isBackendOnlyBuild() and documents the compile-only contract in CONTRIBUTING.md.
Reconciled on merge: CONTRIBUTING.md's new paragraph was inside the ```bash fence and would have rendered as shell — moved below the closing fence. package.json auto-merged against the tip.
Verified: contributor-build-script + backend-only-smoke-workflows 10/10 green.
Thanks @rafacpti23.
The contributor profile added in #12192 inherited the default Turbopack bundler. Turbopack's native allocator is the documented OOM risk on memory-constrained machines (scripts/build/build-next-isolated.mjs:201-202), and OMNIROUTE_USE_TURBOPACK=0 is the escape hatch the same script already honours at line 139 — the repo's own nightly-compat workflow pins it to "0" for exactly this reason. Setting it on build:contributor makes the fast profile usable on the machines it targets.
Reconciled on merge: the branch was 51 commits behind and #12192 had already landed the script line, so only the env flag is new; the tip's dependency block was kept verbatim rather than taking the stale package.json wholesale.
Verified: tests/unit/build/contributor-build-script.test.mjs 1/1 green.
Thanks @rafacpti23.
The Token Plan console cookie is a browser credential for the operator's
cloud-console account — same class as the ollama/opencode cookies that
sanitizeProviderSpecificDataForResponse already strips — but the four
qwen/alibaba fields (qwenCloudCookie, qwenCloudSecToken, alibabaConsoleCookie,
alibabaConsoleSecToken) were missing from the strip list, so GET /api/providers
returned the operator's console session in the clear to any dashboard session.
The edit modal depended on that leak: it initialized the cookie fields from the
round-tripped response. It now starts them empty, matching the ollama pattern —
the quota-scraping assign skips empty fields and the PUT handler's partial merge
preserves keys the payload does not carry, so 'leave blank to keep the stored
cookie' (already what the field hints promise) holds for real.
Found in the 2026-09-01 audit of the Token Plan quota feature.
getPersistedConnectionCooldownSkipReason() returned a skip for ANY connection
whose testStatus was `unavailable`, with no elapsed-cooldown check:
if (status === "unavailable") return `Skipping ...`;
That is the raw-label anti-pattern AGENTS.md warns about ("check whether code
is reading raw state instead of using getStatus()/canExecute()") — the
resilience layers are meant to recover lazily. The sibling helper directly
above it, getConnectionStatusQuotaCutoffReason(), does require
hasFutureRateLimitUntil() before treating `unavailable` as blocking.
Its stated justification — "Lazy recovery is unaffected: clearAccountError()
resets the status on first success" — does not hold on this path. This gate
runs BEFORE dispatch, so it prevents the very successful request that would
call clearAccountError(). And a row whose rateLimitedUntil is absent cannot be
rescued by the out-of-band recovery job either, because hasElapsedCooldown()
there requires a timestamp to be present.
Net effect reported in #12168: an entire combo pool answering
ALL_TARGETS_SKIPPED with recordedAttempts === 0 — zero upstream attempts, no
path back to healthy.
The original intent (do not burst into a connection AUTH just retired, before
the timestamp lands) is preserved, but bounded: the bare label is honoured only
while lastErrorAt is inside a grace window, mirroring ERROR_LABEL_GRACE_MS in
src/lib/quota/connectionRecovery.ts so the two never disagree about whether a
label is still meaningful. Past the window the request goes through, and one
real attempt either succeeds (clearing the status) or re-arms the cooldown with
a fresh timestamp.
Regression introduced by #11360, shipped in v3.8.50.
Two assertions in repro-combo-persisted-cooldown-preskip.test.ts encoded the
buggy behavior as intended ("skips an unavailable connection whose cooldown
already expired") and are realigned to the corrected contract, plus a case for
the orphan state (unavailable with no timestamps at all).
The 1proxy marketplace integration was decommissioned in v3.8.4; the code
survived only through the localDb barrel, deleted in #12055. Everything below
had zero consumers (grep-proven across src/, open-sse/, bin/, electron/,
scripts/ and tests/):
- src/lib/oneproxySync.ts and src/lib/oneproxyRotator.ts deleted.
- src/lib/db/oneproxy.ts: upsertOneproxyProxy, getOneproxyProxyById,
getOneproxyProxyForRotation and markOneproxyProxyFailed removed (their only
consumers were the two deleted modules); listOneproxyProxies and the record
interface stay — open-sse/utils/proxyFallback.ts still uses them.
- src/shared/validation/oneproxySchemas.ts and the unmounted
settings/components/OneproxyTab.tsx deleted (no importer anywhere; the live
UI is the FreePool* tabs over /api/settings/free-proxies).
- ONEPROXY_ENABLED feature flag removed (readerless since oneproxySync died —
the toggle no longer controlled anything); flag-count contract test aligned
54 → 53.
- Docs: PROXY_GUIDE (component rows, env rows, the three omniroute/
oneproxyRotator snippet sections), CODEBASE_DOCUMENTATION, REPOSITORY_MAP,
ENVIRONMENT (ONEPROXY_* rows), FEATURE_FLAGS, .env.example — canonical +
pl/zh-CN/zh-TW mirrors.
- The 308 compat redirects under /api/settings/oneproxy/ stay (deliberate API
compat), as do the live free-proxy provider and proxy_registry rows.
check:dead-code drops 424 → 417 (baseline kept at the velocity-phase 500 —
banking shrinks is paused until v4.0, headroom grows to 16.6%).
check:docs-all, check:env-doc-sync, typecheck:core and the 8 free-proxy/
proxy-fallback test files are green.
Closes#12091
`node_modules/.bin/dpdm` is an npm shell wrapper, so `node <that path>` crashed with `SyntaxError: missing ) after argument list` and the advisory circular-deps gate in ci.yml (job quality-extended) reported an error instead of a result on every run. Pointing DPDM_BIN at `node_modules/dpdm/lib/bin/dpdm.js` restores it: the gate now completes and reports circularDeps=154 (exit 0).
Scope reduced during merge — the branch was 522 commits behind and carried three base-drift files that were reconciled back to the release tip: open-sse/services/combo.ts (reverted routing code + a @/lib/localDb barrel import, Hard Rule #2), config/quality/eslint-suppressions.json (dropped ~45% of the frozen suppressions), and tests/unit/cli-env-inline-comment-10100.test.ts (replaced a working module import with new Function() source scraping, Hard Rule #3). Rationale documented in the PR discussion.
Verified: check:circular-deps crashes on the pure tip and completes on the merged branch; tests/unit/cli-env-inline-comment-10100.test.ts 5/5 green against the restored version.
Thanks @benzntech for catching the dpdm breakage.
UI completa do Orchestration Canvas sobre o modelo da parte 1: página /dashboard/orchestration com abas em URL (Agents=grafo vivo via FlowCanvas, Routing=ComboLiveStudio intocado, Overview=contadores+kanban com totais reais sob cap), drawer de detalhe com approve/cancel (unwrap por fonte verificado contra as rotas reais, erros client-safe, prUrl https-only), i18n com traduções REAIS em 43 locales, entrada no sidebar. Ciclo SDD: 9 tasks TDD com review por task (Task 15 com fix round: Critical A2A unwrap + rewrite de lint + guard XSS), review final whole-branch (Ready to merge, 0 Critical/Important, refactor de complexity provado behavior-preserving), 3 fixes de CI validados RED→GREEN. CI: tudo verde. Crédito do conceito visual: design da PR #11815.
PR #11770 (2026-09-01) added a CLAUDE.md section instructing every AI agent to
clone and execute a third-party setup script; a merge campaign swept it into
the release branch with no human risk review (reverted in #12249). Review
focus now carries the rule: PRs touching CLAUDE.md / AGENTS.md / GEMINI.md /
llm.txt / skills SKILL.md files are HOLD until explicit per-PR operator
approval — CI validates code, not instruction-surface intent.
Gate check:mutation-test-coverage --strict red→verde local (registro dos 2 testes turn-pin no tap.testFiles, drift da mesma classe do #12170). O único check vermelho desta PR (Unit shard 4/4) é o base-red dos próprios testes turn-pin desalinhados pelo #12247 — corrigido pela #12259, mergeada na sequência. Reds circulares: cada PR só está vermelha no item que a outra corrige.
Gap surfaced by OmniCopilot#16: the Chaos Mode dashboard page, its per-key
chaosModeEnabled permission and both dispatch endpoints had no setup doc at
all (only the auto/chaos table line existed), and AUTO-COMBO.md never stated
that weighted is a proportional draw where zero-weight steps are never drawn.
New docs/guides/CHAOS-MODE.md (registered in meta.json + docs/README.md) and
a 'weighted semantics' subsection under the strategy table, both written from
the code (chaosConfig.ts, chaosExecutor.ts, both routes, targetSorters.ts,
targetResolution.ts). check:docs-all exits 0.
* chore(quality): register native-codex-turn-pin tests in stryker tap.testFiles
The mutation-test-coverage gate (--strict) fails on the release tip: the two
native-codex-turn-pin suites (#10379 merge wave) cover open-sse turn-pin code
and src/shared/utils/circuitBreaker.ts but were not listed in
stryker.conf.json tap.testFiles, so their mutant kills would not count. Adds
both files; the gate now passes clean (4728 test files scanned, no drift).
* style: prettier pass on stryker.conf.json
* test(sse): align turn-pin suites to the provider-cooldown window gate
The two native-codex-turn-pin suites landed via the #10379 merge wave after
PR #12247 forked, so #12247's green CI never saw them: they set up 'provider
in global cooldown' with a single recordProviderCooldown call, the pre-#12247
contract. Since the window gate, a provider only counts as cooling after
providerFailureThreshold failures inside the window — the setup now loops to
the profile threshold (same alignment the tracker's own legacy suite got in
Sibling sweep: all 7 suites touching recordProviderCooldown pass (60/60).
providerFailureThreshold / providerFailureWindowMs / providerCooldownMs shipped
in PROVIDER_PROFILES with no runtime consumer (2026-08-31 docs audit, P0.1).
Provider-level entries in providerCooldownTracker now honor them: the whole
provider only counts as cooling after providerFailureThreshold failures inside
providerFailureWindowMs, then cools for providerCooldownMs. Connection-level
entries keep the pre-existing exponential backoff, and the layer stays opt-in
(PROVIDER_COOLDOWN_ENABLED, default off) — default behavior is unchanged.
TDD: tests/unit/provider-cooldown-window-gate.test.ts written first (4 red on
the old behavior), then the wiring; legacy tracker suite aligned to the new
contract (23/23 green). Docs: AGENTS.md breaker section + RESILIENCE_GUIDE
opt-in layer subsection; executors soft-drift refresh (104 -> 106).
Merging --admin with red discrimination (merge-gates §4). The only failing check is Fast Quality Gates → `mutation-test-coverage`, which cannot be caused by this PR: the diff touches exactly one file, `CLAUDE.md` (13 deleted lines, zero .ts). The same gate is red on #12166, #12167 and #12169 — three unrelated PRs — confirming inherited base drift rather than a PR-introduced defect.
dispatchWithCooldownRetry arms a loop-safety timer (setTimeout, 10 minutes by
default) on every setTry iteration, so a combo that never produces a terminal
response still answers with a 504 instead of hanging. The only clearTimeout in
the whole file sat inside the `if (anySuccess)` branch — the comment said so
verbatim: "clear the safety timer on the happy path".
Every error exit therefore returned the response to the client while leaving a
600s timer pending, its closure retaining orderedTargets and the exhausted
provider/connection sets: all_targets_skipped, all_accounts_inactive, the
aggregated-status return, the final fallback, and the global-timeout branch.
The timer is also re-armed per setTry iteration with no clear in between.
Field evidence from the issue: two requests that failed quality validation
returned 502 to the client immediately, and "Combo loop safety timeout ...
force-terminating" was logged for both exactly 600 seconds later — the leaked
timers firing long after the requests were gone.
Fixed structurally rather than by sprinkling clearTimeout across the five
return sites: the handle is hoisted to function scope and released in a
finally, so a future `return` added to this function cannot silently
reintroduce the leak. The 504 backstop itself is unchanged.
Note the timer already called .unref(), so it never held the event loop open —
this is a memory-retention leak, not a hang.
Closes the src/ side of the campaign (no eslint-disable, no new suppressions;
the 37 matching react-hooks/* entries are removed from
config/quality/eslint-suppressions.json — only the 5 CI-divergent entries in
tests/unit/ui remain, frozen by design, see #12144):
- set-state-in-effect (30×): fetch-on-mount and sync-setter effects wrapped in
the async-continuation pattern (await Promise.resolve() for pure-sync
bodies), preserving semantics exactly.
- refs/purity (ResilienceConnectionsClient): render now reads stopReason state
instead of stoppedRef; the receivedAt fallback Date.now() in JSX was dead
(every setData stamps receivedAt) and became 0.
- exhaustive-deps (ApiTab, SessionInfoCard, useLiveDashboard): clearResults
wrapped in useCallback; missing t dep added; channels array stabilized via
channelsKey + useMemo so connect deps are statically checkable.
- global-error: locale/messages load moved into one async continuation (also
renames the import binding to mod per @next/next/no-assign-module-variable).
Refs #12146
Two drifts left behind when the prime-agent runtime entry landed:
- CLI_PRIME_AGENT_BIN (src/shared/services/cliRuntime.ts, defaultCommand
"prime-agent") was in neither ENVIRONMENT.md nor .env.example. The env-doc-sync
gate does not resolve envBinKey values, so it could not catch this.
- CLI-TOOLS.md's summary table still counted 9 CLI Agents while the catalog holds
10 — the section-2 heading and the README breakdown were already correct, only
that cell lagged.
A full sweep of every envBinKey in cliRuntime now shows all of them documented in
both files.
* docs(api): document every implemented route in openapi.yaml (276 -> 692 paths)
Follow-up nº 3 of the 2026-08-31 docs audit: 416 implemented routes had no
OpenAPI entry (gamification, radar, skills, webhooks, mcp, a2a, tunnels,
version-manager and plugins were absent entirely). Adds a minimal, honest
entry for each — real methods parsed from every route.ts's exports, a group
tag and a neutral path-derived summary; no invented semantics. Rich schemas
remain hand-curated in the existing entries.
Generated by scripts/ad-hoc/gen-openapi-missing-paths.mjs, which enumerates
routes with the same lib check:api-docs-refs uses — the spec now covers
692/692 real routes and the gate verifies every spec path has a real route.
* docs(api): security tiers on generated paths, regenerated API skills, size baseline
The first CI round caught three real contract gaps in the generated coverage:
- Generated operations on LOCAL_ONLY routes now carry x-loopback-only (and
x-always-protected for ALWAYS_PROTECTED_API_PATHS), resolved through the real
src/server/authz/routeGuard.ts at generation time. The
openapi-security-tiers guard now also accepts LOCAL_ONLY_API_PATTERNS —
param-shaped routes (/api/providers/{id}/login) are classified by regex in
the runtime and were invisible to the prefix-only check.
- The API agent skills are generated FROM the spec: 18 SKILL.md files
regenerated via generate-agent-skills --apply so the generator stays 46/46.
- src/app/docs/lib/openapi.generated.ts grew with the spec (171 -> 1347 lines,
emitted by gen-openapi-module): frozen in file-size-baseline.json with a
_rebaseline justification — shrink by slimming the spec, never by editing
the generated module.
* fix(oauth): keep Claude personal and Team organizations apart
One Anthropic identity reaches its personal workspace and every Team
organization it belongs to with the same email AND the same accountUUID,
each with its own tokens, plan and rate limits. The OAuth dedup matched on
email alone for every provider except Codex, so authenticating the second
organization overwrote the first connection instead of adding one: only the
most recent organization stayed usable. organizationUUID is the field that
separates them (cliUserID cannot be used, it changes on every login).
Disambiguate on organizationUUID, mirroring how Codex uses
workspaceId/chatgptUserId (#7737):
- findExistingOAuthConnectionMatch routes claude through a new
isSameClaudeAccount helper, so a login only merges into an existing row
when the organization agrees;
- isMatchingOauthIdentity gains organizationUUID as a third optional
disambiguator, compared strictly two-sided;
- createProviderConnection passes the incoming organizationUUID, closing the
same hole on the create path.
Rows stored before Claude returned organizationUUID keep the bare-email
match, so re-authenticating an existing connection still updates it in place
instead of forking a duplicate. No behaviour change for other providers.
* docs(oauth): changelog fragment for #12222
* fix(oauth): mark empty Antigravity projectId as degraded (#11284)
The #11284 gate only fired when projectDiscoveryOutcome was set. Paste
credentials, persistOAuthConnection, and agy CLI import could persist
projectId="" as testStatus=active, so the dashboard showed Connected
while fetchAvailableModels returned 403.
Degrade on empty projectId itself. Keep the refresh token stored so
request-time bootstrap can still self-heal.
Signed-off-by: Minxi Hou <houminxi@gmail.com>
* fix(oauth): clear stale degrade fields and bind CLI imports to builtin client
persistOAuthConnection left errorCode/lastError on the row when a later
connect discovered a Cloud Code projectId. agy CLI import also kept a
leftover custom: oauthClient marker from dashboard OAuth, so the next
refresh hit the operator web client instead of the public desktop client.
Signed-off-by: Minxi Hou <houminxi@gmail.com>
* fix(oauth): null error fields on healthy create paths too
Update already cleared errorCode/lastError* when a projectId appeared.
Create payloads still omitted the keys; match the update shape so a
fresh row cannot keep a leftover degrade marker.
Signed-off-by: Minxi Hou <houminxi@gmail.com>
* test(oauth): pin healthy create/upsert nulling of degrade fields
Forge flagged create payloads omitting errorCode/lastError* when a
projectId is present. Production already writes explicit nulls; the
reader strips them via cleanNulls, so pin both the payload shape and
the upsert path that must overwrite a leftover degrade marker.
Signed-off-by: Minxi Hou <houminxi@gmail.com>
* fix(oauth): type create payload from AntigravityDegradedProjectState
The persistence helper duplicated a subset of the degrade type and
dropped warning. Align the parameter so the HTTP-only warning field
cannot drift from the exported type.
Signed-off-by: Minxi Hou <houminxi@gmail.com>
* fix(oauth): persist degrade status through a single override helper
OAuth exchange/poll-callback spread the whole degrade object, which
wrote warning into the SQLite row and left healthy updates as {}.
Centralize testStatus/errorCode/lastError* so they always win over a
spread tokenData payload.
Signed-off-by: Minxi Hou <houminxi@gmail.com>
---------
Signed-off-by: Minxi Hou <houminxi@gmail.com>
Refresh the existing MIT-licensed miuuyy/codex-chatgpt-web vendor snapshot and its OmniRoute integration as one reviewable change.
Co-authored-by: backryun <backryun@daonlab.local>
* feat(usage): add Kilo Code balance and Kilo Pass quotas
* feat(usage): add Kilo Pass dashboard meter
* test(usage): cover Kilo Code quota integration
* docs(usage): document Kilo API endpoint override
* perf(sse): defer cloneLogPayload until after SSE collector cap check
Dropped SSE events no longer pay the structuredClone cost. The clone now
runs only for events that survive the maxEvents/maxBytes cap, eliminating
~9,800 wasted deep clones per streaming response (65-71% faster push).
Reducer snapshot isolation restored:
- OpenAI reducer stores first-chunk primitives instead of a chunk reference
- Responses reducer snapshots only needed fields, deep-cloning nested output/metadata
- getEvents() keeps defensive-copy semantics via cloneLogPayload
* chore: add changelog fragment for #12241
Local process execution failures (ENOENT spawn errors, binary missing, EPIPE, exit codes) were incorrectly treated as upstream provider failures, opening provider circuit breakers and cooling down valid connections. Added `isLocalExecutionError` guard to skip circuit breaker trips and connection disables when local host execution fails.
* fix(memory): honest probe-driven FTS5 keyword status + memory_id rowid sync
The "no such module: fts5" complaint on FTS5-less runtime builds (sql.js/WASM
under a global install) was masked by a hardcoded keyword.available=true in
engineStatus and an unsanitized FTS5 MATCH path. Address root cause:
- engineStatus(): probe runtime via supportsFts5(db) instead of hardcoding
available=true; keywordEngineStatus() reports the true backend (FTS5 vs
none) with a reason. Schema, OpenAPI, dashboard chip updated to match.
- store.ts: sync memory_id to the SQLite rowid on insert (+ self-heal legacy
NULL rows). Migration 023 keys the FTS5 external-content trigger off
memory_id, but plain INSERT left it NULL so the JOIN returned 0 rows —
keyword/hybrid search silently returned nothing on FTS5-capable builds.
- retrieval.ts: apply sanitizeFts5Query() to the preview MATCH path.
Tests updated/added across memory-engine-status, memory-retrieve-preview,
memory-schemas-roundtrip, memory-store, and the integration engine-status
test (dropping the hardcoded "always available" assertion). 66 unit tests
pass; lint and typecheck clean.
* fix(memory): sanitize FTS5 queries for memory retrieval
Prevent SQLite FTS5 syntax errors by sanitizing query terms and replacing FTS control operators with double-quoted tokens.
Declare the two answers to "is it free?": counting may use the
Radar-overlaid catalog, deciding reads only the shipped FREE_MODEL_BUDGETS
plus :free suffix / zero pricing / grantsFreeAccess. No production behavior
change. A static-import guard discovers every non-client consumer of
freeModels.ts and asserts none reaches getRadarCatalog / getRadarCache,
mirroring client-bundle-no-server-only-10692 on the server arc.
Co-authored-by: Max <maxmad64@gmail.com>
resolveWorkerPath() had two return branches: a process.cwd()-anchored
primary path and an import.meta.url-relative fallback. Turbopack's
dev-mode static worker-chunk detector partially resolves the
new URL(literal, import.meta.url) construct in the fallback branch
independent of which branch actually runs at runtime, producing an
inconsistent module graph node. turbo-tasks then panics on startup
with either 'inner_of_upper_lost_followers...' (aggregation_update.rs)
or 'there must be a path to a root...' (module_graph/mod.rs), and
Restart=on-failure just silently retries forever.
git bisect (443c96d28 good .. b7a0c5413 bad, 418 commits, 9 steps)
isolated this to 657d3a484 (#11732). Confirmed by isolation probe:
removing the Worker construct, or inlining a single non-branching
new Worker(new URL(literal, import.meta.url)) at the call site, both
avoid the panic; only the two-branch resolver does not.
The import.meta.url fallback was also silently dead in production:
the standalone bundle (webpack) freezes import.meta.url to the
build-machine path -- the same app-wide gotcha already documented on
GATE_DEP_REL in llmlingua/worker.ts's fail-open probe. Dropping that
branch fixes both problems with the same change: process.cwd() alone
is correct in every real runtime layout this app uses (dev via
run-next.mjs, and the production standalone bundle, where
outputFileTracingIncludes already copies the worker script preserving
its process.cwd()-relative path).
Verified live:
- Dev (Turbopack): npm run dev reaches '[Next] dev server listening on
...' cleanly; previously panicked and looped under Restart=on-failure.
- Standalone build: npm run build produced a clean webpack compile and
a working .build/next/standalone/open-sse/lib/deepseek-pow-worker.mjs
at the traced path; running the real solveDeepSeekPowAsync from cwd =
the standalone dir spawned the worker and returned the correct nonce
for a fabricated DeepSeekHashV1 challenge.
- node --test tests/unit/deepseek-pow-js-only.test.ts
tests/unit/deepseek-web.test.ts: 44/44 passing.
- eslint and tsc --noEmit: clean on the touched file (tsc's remaining
errors are pre-existing, in unrelated test files).
When a combo is duplicated or imported, its inner data JSON blob may
retain a stale id from the template. withRowId previously kept the inner
string id instead of prioritizing the database primary key (row.id),
causing GET /api/combos to return mismatched ids and breaking subsequent
DELETE / PUT operations with 404.
Also add an error notification branch to handleDelete in the combos page
so failed delete requests surface actionable feedback instead of failing
silently.
Signed-off-by: Minxi Hou <houminxi@gmail.com>
Every Codex quota read goes through throttleQuotaFetch() — the #6009/#6058
gate that spaces genuine upstream calls so many accounts behind one IP do not
fire in the same second, which is the pattern documented to have got a Codex
OAuth token revoked. The auto-ping scheduler called getCodexUsage() directly,
so the one Codex path that runs unattended every 60s per connection was the
one skipping the mitigation written for Codex.
The tick walks connections sequentially but without spacing, so N enabled
connections still produce N upstream usage requests within a few hundred ms.
Gate the read on the same throttle, injected through deps like every other
effect in this module. Placed after the skip checks so a connection filtered
out by the circuit breaker, a cooldown or the failure cache does not consume
a slot and delay the connections that do reach the network.
This does not change the polling cadence. Codex sets pingWhenResetAtSlides
because its resetAt slides forward while the window is idle, so the per-tick
re-fetch is deliberate and is left alone.
Closes#11904
The leading system message was read as `typeof content === "string" ?
content : ""`, so a Chat-Completions content-part array — valid for
`system`, and what every prompt-caching client sends — collapsed the
whole system prompt into an empty `instructions`. Upstream accepted the
request and reported a normal prompt_tokens count, so the model answered
with no instructions at all and nothing in the response said so.
Mid-conversation system turns already handled the array shape (#7056);
only the first one did not. Reuses buildResponsesTextParts() and joins
the text parts, since `instructions` is a string rather than a part array.
Co-authored-by: Vadim Zhyvylo <zhyvylo@involve.software>
The onboarding diagram (TierFlowDiagram.tsx) still drew the legacy 3-tier
cascade (Subscription -> Cheap -> Free). Redrawn for the real model —
Tier 1 Subscription -> Tier 2 API -> Tier 3 Cheap -> Tier 4 Free — keeping
each theme's existing visual language (new cyan family for the API tier),
exact 4-column geometry on the 800x420 canvas, a Linux-safe font stack and
the accessibility floor (role/aria-label/title/desc). Rendered and verified
via svg-studio (validator pass, diagram checker 0 violations); canonical
numbers (352/19/110) remain covered by check:docs-counts.
mcp-tools-107.{mmd,svg} -> mcp-tools.{mmd,svg} and
auto-combo-12factor.{mmd,svg} -> auto-combo-scoring.{mmd,svg}: filenames that
embed a canonical count fossilize the moment the count moves (107 -> 110,
12 -> 15 factors already happened). All referencers updated — diagrams index,
AUTO-COMBO.md (+ its pl/zh-CN/zh-TW mirrors' links) and the check:docs-counts
file list.
45 violations across 27 files fixed at the source (no eslint-disable, no new
suppressions; the 45 matching react-hooks/* entries are removed from
config/quality/eslint-suppressions.json):
- set-state-in-effect (fetch-on-mount effects): async continuation wrapper.
- Prop/state sync effects (EditMemoryModal, radar/setup, EvalsTab): adjust
during render with prev tracking.
- purity/refs (ActivityFeedClient, ProviderQuotaWidget, ReasoningCacheTab):
Date.now() snapshots moved to state set from the fetch path; rendered refs
converted to state.
- immutability (useCodexResetCreditRedemption): ref-store writes extracted to
module-level helpers.
- exhaustive-deps (RequestLoggerV2, HomePageClient): COLUMN_SORT_MAP hoisted to
module scope; openDetail/closeDetail wrapped in useCallback and added to the
dependent hooks; versionInfo destructured to locals; baseUrl now reads
location.origin via useSyncExternalStore (hydration-safe, no effect).
Refs #12146
* fix(api): keep registry width and type on embedding models
* docs: changelog fragment for embedding registry fix
* test(api): cover embedding width and type merge
Exercises /v1/models rather than the registry in isolation: a synced
model colliding with an embeddingRegistry entry must keep the width the
registry states, and a synced model the registry names must be typed as
an embedding model.
Fails on catalog.ts before e7fbb62 (2 failures), passes after.
Refs #11759
Add a direct low-level mode for users who require explicit control over provider selection. score selects the highest configured weighted score directly while reusing the existing exploration rate.
Exact ties preserve configured candidate order. rules and all other strategies remain unchanged.
- streamReadiness: reset deadline on each received chunk (keepalive = alive)
with a hard maxTimeoutMs ceiling so truly-dead connections still fail fast.
Preserves operator's 20s/100s intent for dead pulls while allowing slow-but-alive
upstreams (reasoning warm-ups) to survive.
- chatCore + codexIdentity: auto-detect Claude Code CLI via user-agent/originator
headers and enable model echo for it. The response field now echoes
the originally-requested alias/combo (e.g. ) instead of the
resolved upstream id (e.g. ), so restores
cleanly without 'could not be restored' errors.
Refs: opensource-elearning/omniroute-fixes#1, diegosouzapw/OmniRoute#12185
OmniRoute's SSE teardown aborts in-flight legs with
`Error [AbortError]: request_signal_aborted` on client disconnects
(open-sse/utils/streamHandler.ts getClientAbortReason), and fetch/DOM
cancellation surfaces as AbortError with an abort-flavoured message.
isClientAbortError() only matched message 'aborted'/'Aborted' plus errno
codes, so these shapes fell through shouldSwallowUncaught() and were
re-thrown from the process-level uncaughtException/unhandledRejection
handlers — killing the whole server on a routine client disconnect
(observed as repeated exit-code-7 crashes with
'uncaughtException: Error [AbortError]: request_signal_aborted').
Match AbortError by name when the message is abort-flavoured; genuine
errors that merely mention 'abort' (e.g. TypeError) still crash loudly.
Tests: new unit cases for the SSE/DOM AbortError shapes, a child-process
regression proving the process survives both benign emissions with the
production no-logger install shape, and a child-process test proving
genuine errors keep crash semantics.
Fase 2 da auditoria código×docs 2026-08-31: ~90 divergências corrigidas em README/llm.txt(+42 espelhos)/SVGs/AGENTS.md/25+ docs; correções semânticas (breaker 8/12/2 + DEGRADED, webhooks sem eventos fantasma, ROUTE_GUARD_TIERS completo, API_REFERENCE sem fantasmas, reasoning 200); gate check:docs-counts endurecido (versão em prosa, patterns anti-evasão, superfície +llm.txt/mcp-server/omni-mcp/tier-flow) +5 testes; fonte do gerador de agent-skills corrigida (107/32 → 110/33) e mesma família varrida do Copilot prompt, 39 locales, skills/README, CONTRIBUTING e 2 guias.
* chore(lint): batch 5 of #12146 — resolve the react-hooks compiler violations in combos, endpoint, provider-stats, api-manager and costs
40 violations across 14 files, all real refactors (no eslint-disable, no new
suppressions; the areas' react-hooks entries are deleted from the freeze):
- set-state-in-effect (30): fetch-on-mount effects moved behind an async
continuation (usePools, usePoolUsage, useApiKeyUsageLimits, Notion/Obsidian
source cards, A2A/MCP dashboards, ComboControlCenterClient, provider-stats,
combos modal loaders, ApiManager initial load); prop/state sync converted to
state adjustment during render with prev tracking (ApiKeyUsageLimitCard,
PoolWizard dimensions/reset/group snap, combos sortMethod, builder reset,
builder stage guard, single-provider default, stale intelligent selection);
localStorage reads became lazy useState initializers (combos usage guide).
- immutability / TDZ (8): effects that scheduled fetchers declared below them
moved after the declarations (EndpointPageClient, ApiManagerPageClient,
combos mount load); fetchData relocated below the per-key fetchers it calls.
- static-components (7): provider-stats SortIcon hoisted to module level.
- preserve-manual-memoization (2): ApiManager blockedModels dep destructured to
a local; provider-scope derivation memoized so downstream memos see a stable
dependency.
Validation: eslint (CI command with suppressions, --max-warnings 0) clean on the
14 files; dashboard typecheck within baseline; mutation gate no drift; area
tests 208/208 (node) + 29/29 (vitest).
Refs #12146
* chore(lint): batch 5 follow-up — hoist the render-adjustment predicates so the new-code complexity gates stay flat
The render adjustments added one cyclomatic branch to combos/page.tsx and one
cognitive point to PoolWizard (caught by the new-code gate on the committed
work); the compound conditions now live in pure module-level predicates.
* chore(lint): batch 5 follow-up 2 — PoolWizard render adjustments live in two small hooks
One consolidated hook tripped max-lines-per-function (>80) and the cognitive
budget; the dimensions and open/close adjustments now live in two focused hooks
with a shared WizardSetters type, and the group snap stays inline (one branch).
complexityNewCode=0, cognitiveComplexityNewCode=0.
* test(quota): repoint the two PoolWizard structural pins at the render-adjustment hook
quota-edit-opens-wizard anchored the pre-fill block on the old '} else if (editPool)'
effect literal and quota-pool-wizard-edit expected a bare 'if (editPool)' that only
existed there; both now anchor on the batch-5 structure (submit still branches via
if (!editPool)).
Resolves all 28 react-hooks/* compiler violations (24 set-state-in-effect,
4 refs) across the 18 dashboard/providers files of batch 2 and removes their
suppression entries — no eslint-disable, no new suppressions.
Techniques per file:
- Fetch-on-mount loaders (CustomModelsSection, ProviderCcAliasSection,
ProviderInterceptionSection, ProviderParamFilterSection, page.tsx,
useProviderConnections, useProviderSettings, CliproxyAccountHealthCard,
DarioAccountPanel, NinerouterModelList): network/parse/error concerns
extracted to module-level helpers returning error-as-value; the async glue is
defined INSIDE each effect with every setState after the await. Loaders that
handlers still need (refresh/retry buttons, exposed hook API) remain as
callbacks; spinner flags moved into the button handlers.
- Loading flags for provider-keyed sections derived from a loadedProviderId
marker instead of synchronous setLoading(true) resets.
- Modal init/reset effects (EditConnectionModal, EditCompatibleNodeModal,
AddCompatibleProviderModal, VolcengineConnectModal state reset,
useProviderUrlFilters hydration, page.tsx display-mode fallback,
useProviderSettings per-provider flag reset): converted to render-phase
adjustments guarded by the previously-seen prop/marker (react.dev "adjusting
state when a prop changes").
- VolcengineConnectModal: phone prefill via localStorage lazy initializer;
server-side session cancel + poll stop moved to the cleanup of an
open-scoped effect reading a session ref mirror.
- ModelCompatPopover refs: render-time ref mirrors removed — headerRowsRef is
maintained by an applyHeaderRows writer used by all handlers, paramTargetRef
is mirrored in an effect, and blockText/allowText mirrors were already kept
in sync by their single writer (applyParamFields).
- ModelCompatPopover state: header-row loading and value-visibility resets
moved from [open, protocol] effects into the open/protocol/outside-click
gesture handlers; the closed-popover rect reset was dropped (render is gated
on open and the rect is recomputed pre-paint on reopen).
- useRiskAcknowledged: localStorage mirrored via useSyncExternalStore with a
module-level listener set notified by acknowledgeProviderRisk.
- useProviderModels: loading for the empty-providerId case derived at the
return site instead of a synchronous setLoading in the effect.
Validation: scoped eslint with the suppressions file passes with 0 problems;
check-dashboard-typecheck.mjs OK; node --test batch (14 files) and vitest
batch (9 files, 47 tests) green.
Refs #12146
* chore(lint): batch 3 of #12146 — resolve the react-hooks compiler violations in dashboard/settings
Resolves the 25 react-hooks/* React Compiler violations frozen in
config/quality/eslint-suppressions.json for the dashboard/settings area
(24 set-state-in-effect, 1 immutability), plus the adjacent
react-hooks/exhaustive-deps in ProviderAccountRoutingCard, and removes
their suppression entries. No eslint-disable added anywhere; one
pre-existing eslint-disable-line (AccessTokensTab) removed.
Techniques used:
- ResilienceTab (8×): the "sync draft state from prop via useEffect"
cards now use the documented adjust-state-during-render pattern
(prevValue state + conditional setState in render) instead of an
effect.
- PricingTab visibleCount reset: same render-adjustment pattern keyed
on the filters tuple, replacing the reset effect.
- Fetch-on-mount loaders only used by the effect (IPFilterSection,
ModelCapabilityOverridesTab, PayloadRulesTab*, RoutingStrategyCard):
loader inlined into the effect as an async IIFE with a cancelled
flag; every setState now happens after the first await.
- Loaders reused by handlers/intervals (AccessTokensTab, AuthzSection,
FallbackChainsEditor, MitmProxyTab, ModelsDevSyncTab, OneproxyTab,
PayloadRulesTab, PoliciesPanel, PricingTab,
ProviderAccountRoutingCard, SystemStorageTab, GlobalConfigTab,
SubscriptionTab): split into a module-level pure fetcher + a
useCallback applier; the effect awaits the fetcher and applies after
the await (cancellation-guarded), while handlers keep the original
named loader (sync setState is fine there) built from the same
fetcher/applier — no logic duplication, identical error-message and
loading semantics.
- OneproxyTab keeps the spinner-on-filter-change behavior via the same
render-adjustment pattern (filtersKey → setLoading(true)).
- AccessTokensTab: the L() fallback helper is now memoized with
useCallback([t]), which also let the old
eslint-disable-line react-hooks/exhaustive-deps be removed.
- ProviderAccountRoutingCard: save's dependency array now includes
load (the frozen exhaustive-deps violation).
Suppressions: all react-hooks/* entries for the 17 batch files removed
(19 rule entries, 26 violation counts). Entries for other rules/files
untouched.
Refs #12146
* chore(lint): batch 3 follow-up — extract useOneproxyData so the cyclomatic gate stays flat
The first pass grew OneproxyTab past the complexity threshold (caught by the
new-code gate on the PR); the data-loading state now lives in a dedicated
useOneproxyData hook. Also registers search-432-plan-limit-cooldown in
stryker tap.testFiles (base drift the gate flagged on every batch).
* chore(lint): batch 1 of #12146 — resolve the react-hooks compiler violations in dashboard/cli-code
Real refactors, no suppressions — the 42 frozen react-hooks/* entries for the
12 dashboard/cli-code files (plus Antigravity's exhaustive-deps one) are
removed from config/quality/eslint-suppressions.json and the files now lint
clean under the React Compiler rules.
Techniques, per pattern:
- set-state-in-effect ("default API key" effects — Antigravity, Claude, Cline,
Codex, Droid, GrokBuild, Kilo, OpenClaw): the setState-in-effect that copied
apiKeys[0].id into the selection state is deleted; an `effective*` value is
derived during render (`selected || apiKeys[0]?.id`) and used by the select
and the submit handlers. Behavior identical, one less render pass.
- immutability ("accessed before declared") + set-state-in-effect on the
expand-time loaders (all tool cards): the fetchers (checkXStatus,
fetchModelAliases, fetchBackups, fetchProfiles, loadSavedMappings) are
hoisted above the effect as useCallback with correct deps, listed in the
effect deps, and invoked through an async continuation
(`void (async () => { await Promise.all([...]) })()`) so no setState runs
synchronously in the effect body.
- set-state-in-effect ("init form from fetched status" effects — Claude,
Cline, Codex, Droid, OpenClaw): the status-parsing effects are deleted and
their logic now runs inside checkXStatus right after the fetch resolves
(setState after await), keeping the same one-time ref guards. Codex's config
parser became syncFormFromStatus(), called on both success and error paths.
- HermesAgentToolCard: Date.now() in render (purity) is snapshotted once via a
lazy useState initializer; the batchStatus seeding effect is replaced by a
derived `displayRoles` (useMemo over batchStatus with currentRoles taking
precedence); the collapse-reset effect moved into the header toggle handler.
- ClaudeClassifierCompatToggle / CliProfileAutoSyncToggles / Cliproxyapi /
GrokBuild: mount/expand loads wrapped in the same async continuation.
- DroidToolCard's isOmniRouteEntry helper hoisted to module scope (pure).
Validation: eslint with suppressions --max-warnings 0 on the 12 files (clean),
scripts/check/check-dashboard-typecheck.mjs (OK, within frozen baseline),
vitest UI suites for the touched cards (15 files / 57 tests green, plus the 3
quarantined #8618 files run explicitly: 27 tests green), and the node-native
cli-code tests (61 tests green).
Refs #12146
* chore(lint): batch 1 follow-up — hoist the settings-init helpers so the cognitive gate stays flat
The first pass folded the one-time form init into the status fetchers, which pushed
sonarjs/cognitive-complexity to 1 in Claude/Cline/OpenClaw tool cards (caught by the
new-code gate on the PR). The init logic now lives in module-level helpers
(initXFormFromSettings + defaultKeyId); complexityNewCode=-1, cognitiveComplexityNewCode=0.
check:mutation-test-coverage --strict verde local e no CI (Fast Quality Gates pass, 18/18 checks). Registro de 1 linha em tap.testFiles cobrindo accountFallback.ts e auth.ts, drift introduzido pelo #12139. Desbloqueia o gate para todas as PRs contra release/v3.8.51.
* chore(lint): batch 4 of #12146 — resolve the react-hooks compiler violations in shared/components
Real refactors (no suppressions, no eslint-disable) for the 21 react-hooks/*
violations across the 11 src/shared/components files of this batch:
- set-state-in-effect (prop/state mirror or modal open/close reset):
replaced with guarded render-time adjustments (react.dev "You Might Not
Need an Effect" prev-tracking pattern) — KiroAuthModal,
ModelSelectModal, ProxyConfigModal, OAuthModal (provider-change, close
and open resets; ref invalidation split into ref-only effects),
RequestLoggerDetail.sections (liveDetail mirror),
ComboCompressionModeSelect (initialCompressionMode mirror).
- set-state-in-effect (fetch+set effects calling component-scope
functions): moved the async loader inside the effect (ModelSelectModal
fetchCombos/fetchProviderNodes/fetchCustomModels, PricingModal
loadPricing, useProviderDailyUsage fetchRows — now with a cancelled
guard) or wrapped the call in an effect-local async runner
(ReasoningRoutingRules load, UsageStats fetchStats, OAuthModal
startOAuthFlow) with every setState on the async path.
- OAuthModal device-code countdown: deviceCodeSecondsRemaining state
deleted and derived from deviceCodeExpiresAt plus a `now` tick state
updated by the interval (re-anchored when polling starts).
- Sidebar localStorage hydration: reads moved into useSyncExternalStore
snapshots (server snapshot null) applied via render-time adjustment;
skipInitialActiveExpansion ref converted to state; the active-section
expansion effect became a render-time adjustment keyed on the old
effect deps; persistence consolidated into one saveToStorage effect
(removes the saves that ran inside setState updaters and drops a
pre-existing eslint-disable for exhaustive-deps).
- immutability (use-before-declare): PricingModal loadPricing inlined
into its effect; ProxyConfigModal resetFields hoisted above the load
effect as a dependency-free useCallback.
- exhaustive-deps (ProxyConfigModal): effect now depends on the stable
resetFields and on hoisted translated strings (socks5HiddenError,
levelGlobalLabel) instead of the `t` identity.
- preserve-manual-memoization (UsageStats sortedAccounts): optional
chains destructured into locals so the memo deps match the usage.
config/quality/eslint-suppressions.json: removed every react-hooks/*
entry for the 11 files (other-rule entries preserved).
Validation: eslint gate (--suppressions-location, --max-warnings 0) green
on all 11 files; typecheck:core clean; node unit sweep 373/373; vitest
sweep 547/550 with the 3 fails being 5s-timeout flakes under parallel
load (all pass isolated 8/8, one in an untouched file).
Refs #12146
* test(mutation): register search-432-plan-limit-cooldown in tap.testFiles
The test (merged with the DuckDuckGo cooldown fix) covers accountFallback.ts and
auth.ts but was not listed, so check:mutation-test-coverage --strict reds any PR
whose merge ref includes it. Base also merged in.
Typed CallLogRow / PayloadEnvelope views over the raw rows and payload envelopes;
(assert as any).equal back to assert.equal. Suppression entry for the file removed —
the gate now watches it for real. eslint (CI command) clean, suite 15/15.
Refs #12146
A global ratchet ("total ≤ baseline") reds an innocent PR whenever the base
drifted, and lets a PR that adds 10 violations pass as long as someone else
removed 11 — both happened this week. On pull_request events quality.yml now
passes --base-ref <PR base SHA> to check:complexity-ratchets and check:dead-code
(file-size already had it); in that mode the gate compares HEAD with the
merge-base RESTRICTED to the files the PR touched:
- blocking: violations / dead exports the PR added in files it changed
(complexityNewCode=, cognitiveComplexityNewCode=, deadExportsNewCode=)
- advisory: the global total vs the frozen baseline (re-frozen at release,
watched by the nightly headroom job)
scripts/check/newCodeMode.mjs holds the git side (merge-base, changed files,
throwaway `git worktree` of the base with node_modules linked — no stash, no
checkout) and the pure comparison helpers (13 unit tests). ESLint runs only on
the changed files in both trees (~20 s); knip runs twice (~70 s).
Exercised locally against the last 8 merges: complexity flagged
src/lib/credentialHealth/scheduler.ts (2→3, cognitive 1→2) and dead-code flagged
src/lib/resilience/settings.ts:CredentialHealthCheckSettings — findings the
global totals were hiding under the relaxed baselines.
workflow_dispatch, the release-green sweep and the headroom job have no PR base
and keep the absolute comparison. Docs: QUALITY_GATES.md → "New-code mode".
* fix(ci): clear the base-reds the 2026-08-30 afternoon merge batch left on release/v3.8.51 (round 5)
- docs-counts / check-docs-counts-sync test: #12103 (Perplexity Agent) made it 352
providers; README, AGENTS.md, llm.txt (+42 i18n mirrors), package.json description
and the 4 README diagrams still said 351.
- api-route-typecheck: #11971 passes a third `{ featureEnabled }` argument to
appendNoThinkingVariants() that the helper never accepted (TS2554 — and the flag
silently did nothing); the helper now honours it. src/lib/skills/interception.ts
narrowed a mapped object with a `Record<string, string>` predicate (TS2677) —
predicate typed with the actual element shape.
Gates: check:docs-counts OK (test 28/28), check:docs-sync PASS, check:api-typecheck
OK (289 frozen). Refs #12103, #11971
* docs(env): document RATE_LIMIT_EXECUTION_MAX_WAIT_MS (#12027 added it to .env.example only)
* fix(ci): round 5b — freeze the react-hooks compiler-rule violations, align 7 tests to merged contracts
No new ESLint warnings: the exact CI command (lint:json --max-warnings 0) reports 278
problems on the tip — 226 from eslint-plugin-react-hooks 7 compiler rules
(set-state-in-effect 167, immutability 36, refs/static-components/purity/
preserve-manual-memoization) that were masked until the lockfile change of
dfc84ba030 invalidated the ESLint cache, plus 46 no-explicit-any in
tests/unit/call-log-cap.test.ts (#12026). Velocity phase: frozen with
`eslint --suppress-all` (+668 suppressions); the 5 now-unused
`eslint-disable react-hooks/immutability` directives and one unused import removed.
Verified: lint:json --max-warnings 0 → 0 problems.
Tests aligned to contracts merged this afternoon (all reproduced red on the pure tip):
- providers-constants-split: 235 → 236 (Perplexity Agent, #12103)
- sse-auth: a forced pin outside allowedConnections now yields no credential
instead of silently falling back (#12080)
- with-chat-admission-10786: withInjectionGuard(postHandler, { logger: null }) (#12117)
- hard-session-lease-bypass-inventory: classify src/app/api/oauth/codex/import/route.ts (#12116)
- usage-service-hardening: OpenCode Go official usage API shape (#12124)
- i18n placeholder parity: apiManager.restrictedToConnections rewritten as a plain
ICU plural (`{count, plural, one {# connection} other {# connections}}`) in en,
vi, pt-BR and the 40 __MISSING__ mirrors — the parity extractor counts every
`{word}` including the old literal `{s}`
Refs #12103, #12080, #12117, #12116, #12124, #12026
* fix(ci): run the ESLint warnings job on the box with an 8 GB heap; reserved-prefix set 398 → 400
The cold full lint with the react-hooks 7 compiler rules is killed on the 7 GB hosted
runner with no message (status null → exit 1, JSON never written) — it only looked
green while the ESLint cache was warm. tests/unit/provider-node-reserved-prefix.test.ts
aligned to the two prefixes the afternoon batch registered (#12103).
* test(ci): document the lint-guard runner exception; #9147 event-loop gap 400 → 800 ms
quality-rail-gate-membership pinned lint-guard to ubuntu-latest; the cold full lint is
OOM-killed there, so the job now runs on omni-light with an 8 GB heap — the test keeps
fast-gates pinned and asserts the documented exception. With the catalog at 352
providers the hosted shards measure 410–633 ms gaps on 9147-catalog-eventloop-yield
(3 runs); 800 ms still fails a true pin. Re-tighten with the v4.0 catalog split.
* chore(quality): summarize the ESLint report on failure — a red lint:json printed nothing
--format json --output-file swallows every problem; a red 'No new ESLint warnings' job
gave zero output (three blind debugging rounds in #12144), and a killed process (OOM,
status null) was equally silent. On any non-zero exit the runner now prints the problem
count and the first 60 'file:line rule — message' lines from the report.
* chore(lint): freeze react-hooks/immutability for the 5 UI test harnesses in the suppressions file
The rule fires for these files in CI but not locally (compiler analysis divergence),
so the inline eslint-disable directives read as 'unused directive' warnings locally.
A suppressions entry is symmetric: suppressed where the rule fires, tolerated as
unpruned (--pass-on-unpruned-suppressions) where it does not. Found via the new
lint:json failure summary.
Aumenta o teto do sticky round-robin limit para 1000, com teste próprio (3/3 verdes). Fiz cherry-pick só dos 2 commits reais direto na tip atual: a branch original carregava 4 commits antigos de drift do ciclo (release/electron/CI, já resolvidos de outras formas) que geravam conflito redundante contra `.github/workflows/electron-release.yml`. Nenhum conteúdo seu foi perdido — força-pushed a branch limpa (autoria preservada). Obrigado!
Honra um `context_length` definido pelo operador em tempo de requisição no roteamento do combo (supersede #12014, que estava incluída nos mesmos commits). Boa cobertura de testes, incluindo o refactor de `resolveComboContextLimit` para módulo próprio. Validado no worktree combinado (13/13). Obrigado!
Intervalo de checagem de saúde de credencial configurável pelo operador, com boa cobertura de testes. Validado no worktree combinado (20/20).
Corrigi o import de `getCachedSettings` em `src/app/api/resilience/route.ts` e `src/lib/credentialHealth/scheduler.ts`, que apontava para `@/lib/db/settings` (path antigo antes do split para `@/lib/db/readCache`, já na tip). Resolvido também um conflito de tradução vi.json entre chaves duplicadas de outra feature (exclusive lease), sem relação com esta PR — mantida a versão já mergeada. Obrigado!
Adiciona suporte ao GLM-5.3-Flash Coding Plan (endpoint OpenAI-compatible, tiers de esforço low/high/max via reasoning_effort). Boa cobertura de testes. Validado no worktree combinado.
Dois problemas resolvidos antes de mergear:
1. **Duplicata silenciosa de "glm-5.3-flash"** em `src/shared/constants/modelSpecs.ts` e `open-sse/config/glmProvider.ts` (#11830, já mergeado nesta sessão, e sua PR inserem a mesma entrada em pontos diferentes do arquivo — git não detecta como conflito textual). Removida a duplicata, preservando a ordem que o teste pré-existente `open-sse/mcp-server/__tests__/glmCodingProviderConfig.test.ts` espera (glm-5.3-flash primeiro no array `GLM_SHARED_MODELS`).
2. Conflito real em `zai/index.ts`, `default.ts`, `pricing/shared-tiers.ts` e no teste de catálogo — todos aditivos, resolvidos mantendo ambos os lados.
27/27 + 10/10 (vitest) testes focados verdes. Obrigado!
Mantém o erro do call-log quando o limite de tamanho corta os bodies, com `preserveErrorForSizeLimit` (UTF-8-safe, preserva o valor original quando cabe, trata erro circular/não-serializável) — implementação mais robusta que a alternativa que já estava na tip (via #12027, que resolvi combinando: mantive a camada extra "errorOnly" do #12027 usando o helper mais seguro deste). Testes próprios + os de #12027 todos verdes (30/30) no worktree combinado. Obrigado!
Emite `web_search_call` nativo para o fallback de web_search da Responses API, com boa cobertura (integração + unitário). Validado no worktree combinado.
Corrigi 3 problemas no próprio `tests/integration/skills-pipeline.test.ts` desta PR antes de mergear: faltava `encodeSkillToolName` no import (usado em 3 lugares, causava `ReferenceError` que se propagava como 502 no teste "matching tool calls execute the registered skill") e 2 asserções comparavam nomes decodificados (`decodeSkillToolName`) contra valores re-codificados (`encodeSkillToolName`) — copy-paste do helper usado para montar o mock. 30/30 testes focados verdes após a correção.
Adiciona compatibilidade de renomeação de migração para 056/073/077/101. Teste próprio atualizado (7/7 verde no worktree combinado + isolado).
Fiz cherry-pick só dos 2 commits reais da PR (o fix + o ajuste do teste) direto na tip atual: a branch original carregava 3 commits antigos de drift do ciclo (release-workflow/electron, já mergeados de outras formas) mais um commit de auto-resolução de merge seu, que juntos geravam conflito redundante contra `.github/workflows/electron-release.yml`. Nenhum conteúdo seu foi perdido — força-pushed a branch limpa (autoria preservada). Obrigado!
Desacopla a expiração de execução do rate-limit do orçamento de espera na fila, e preserva erros em artefatos de call-log oversized. Testes próprios (`call-log-cap.test.ts` + atualizações em `rate-limit-execution-timeout-message-4165.test.ts`/`ratelimit-admission-control-6593.test.ts`). Validado no worktree combinado. Obrigado!
Migra o quota fetcher do OpenCode Go para a API oficial de uso, com refactor substancial que remove ~1850 linhas de código legado e atualiza a suíte de testes existente inteira para o novo contrato. Validado no worktree combinado (typecheck limpo, testes focados verdes). Obrigado!
Corrige o bulk-import do Codex apagando `providerSpecificData`/`tokenExpiresAt`/duplicando `priority` de conexões existentes ao fazer upsert num match — mescla o payload importado sobre o estado existente em vez de substituir tudo, igual ao caminho de import single-file já fazia. Findings 4, 6 e 7 do #12113. Teste próprio (224 linhas). Validado no worktree combinado. Obrigado!
Refresca o manifest do plugin a partir do disco ao ativar, para que instalações pré-existentes ganhem hooks novos adicionados por schema updates (ex.: `onStreamComplete` do #11825/#11934 nunca chegava a plugins já instalados antes do upgrade, pois o manifest persistido no DB era stripado pelo schema antigo). Finding 3 do #12113. Teste próprio (259 linhas). Validado no worktree combinado. Obrigado!
Restaura o log do injection-guard nas 13 rotas não-chat (embeddings, images, audio, moderations, etc.) — a correção de log duplicado anterior (#11936) silenciou completamente o único emissor de log dessas rotas, deixando tentativas de injeção sem rastro nenhum em modo warn, e sem log mesmo quando bloqueadas em modo block. Achado de segurança real (Finding 2 do #12113). Teste próprio (141 linhas). Validado no worktree combinado. Obrigado!
Corrige o kill do processo inteiro do plugin quando um handler fire-and-forget de `onStreamComplete` demora >10s — hook documentado como fire-and-forget não deveria derrubar o processo a cada stream completo. Finding 5 do #12113. Teste próprio (214 linhas). Validado no worktree combinado. Obrigado!
Corrige vazamento de colunas de conexão (email/nome/etc.) através do cast em `getExclusiveConnectionLeaseStatus` — a projeção agora fica restrita às colunas de lease, evitando que um futuro consumidor sirva PII sem querer via o tipo `ExclusiveConnectionLease`. Documentado como Finding 8 do seu próprio bug-audit (#12113). Teste próprio (103 linhas). Validado no worktree combinado. Obrigado!
Vincula o refresh OAuth do Google ao client que emitiu o token, com teste próprio (`google-oauth-client-binding.test.ts`). Validado no worktree combinado. Obrigado!
Adiciona o provider Perplexity Agent API, com dois arquivos de teste próprios (provider + sanitização de chatCore). Validado no worktree combinado. Obrigado!
Mantém conexões forçadas indisponíveis com escopo correto, com teste próprio ampliado (`forced-connection-fallback.test.ts`). Validado no worktree combinado. Obrigado!
Alinha o body do combo e o acesso legado por chave, com testes atualizados (CLI api-generator + row parsers). Validado no worktree combinado. Obrigado!
Deriva as modalidades do auto-combo a partir do pool de targets efetivo, com teste próprio robusto (174 linhas). Validado no worktree combinado. Obrigado!
Injeta a tag de usuário obrigatória nas requisições de inferência do provider Nous, com teste próprio ampliado. Validado no worktree combinado. Obrigado!
Corrige o achatamento incondicional de conteúdo de mensagem no cloudflare-ai — a restrição #2539 é model-scoped, não global, e estava bloqueando entrada de imagem em modelos de visão da Cloudflare. Teste próprio atualizado. Validado no worktree combinado. Obrigado!
Usa a lista de publishers v1beta1 do Model Garden para descoberta de modelos Vertex Anthropic, com teste próprio. Validado no worktree combinado. Obrigado!
Mantém tokens de cache-write no formato de usage do OpenAI, com teste próprio (`cache-write-openai-shape.test.ts`) e atualização do teste existente de tokens detalhados. Validado no worktree combinado (typecheck limpo, 351/351 testes focados). Obrigado!
Owner decision (2026-08-30): shipping speed matters more than holding the debt line
until the v4.0 LTS modularization; the base was going red on every merge batch and
each red baseline cost a sweep.
Relaxation (one auditable pass, scripts/quality/relax-baselines.mjs):
- quality-baseline.json metrics: lower-is-better ×1.2, higher-is-better ÷1.2
(coverage floor 60 kept; eslintErrors stays 0; eslintWarnings 0 → 1050 = 20% of
the 5,247 frozen suppressions). Adds `_policy {phase: velocity, until: 4.0.0,
relaxPct: 20, requireTighten: false}` + a `_relax_velocity_2026_08_30` note
listing every before → after.
- complexity count 2681 → 3218; duplication 5.72 → 6.86; file-size cap/testCap
1000 → 1200 and all 127 frozen caps ×1.2; api/dashboard/open-sse typecheck
per-file counts ×1.2; openapi-coverage THRESHOLD 36 → 30.
- check-quality-ratchet: --require-tighten is advisory while _policy.requireTighten
is false (2 new tests); nightly bank-ratchet-shrinks pauses during the phase (it
would bank the measured shrink and undo the headroom every night).
Monitoring (scripts/quality/baseline-headroom.mjs, npm run quality:headroom):
measures each numeric gate the way CI does, prints live / baseline / headroom per
gate (ok ≥10%, warn <10%, critical <0); the new nightly `baseline-headroom` job
posts the table to the living issue "📈 Baseline headroom (velocity phase)" and
toggles the `headroom-alert` label. 6 unit tests on the pure helpers.
Also aligns the remaining red tests on the tip to contracts already merged:
#11775 (FREE lease-capable connections are ordinary capacity: gate inventory 48/97/99,
sse-auth selection, warmup scheduler), #11794 (dual-loopback readiness probe), and the
8 vi strings #11775 left as __MISSING__.
Docs: QUALITY_GATES.md → "Velocity phase" (what changed, tooling, how to close the
phase at 4.0), AGENTS.md quick reference.
- api-route-typecheck: 56dddfce34 (antigravity loadCodeAssist metadata) made
getAntigravityLoadCodeAssistMetadata() return Record<string, number> while
onboardAntigravityUser() still typed the parameter Record<string, string> —
TS2345 in src/lib/oauth/providers/antigravity.ts, gate red on every PR. The
parameter now derives from the getter's return type.
- env-doc contract: 0b19c5a09b (#11852, 5dive configure target) reads
CLI_5DIVE_BIN and CLI_5DIVE_STATE_DIR without documenting them —
Docs Gates red on every PR. Added to .env.example and ENVIRONMENT.md.
Gates: check:api-typecheck OK (289 frozen), check:env-doc-sync OK,
antigravity oauth tests 14/14.
Refs #11852
#11923 (22011437f8) deliberately routes OrcaRouter chat requests to
https://api.orcarouter.ai/v1/chat/completions; the golden snapshot in
tests/unit/provider-translate-path-golden.test.ts still pinned /v1, so
Unit Tests fast-path (2/4) is red on the release tip for every PR.
Regenerated with UPDATE_GOLDEN=1 — the only delta is the orcarouter block.
Refs #11923
* fix(dashboard): make RequestLoggerDetail loadable outside Next — CSS via globals.css, CJS/ESM interop for react18-json-view
Origin: #11703 (5684589ce7) imported `react18-json-view/src/{style,dark}.css` at
module level in RequestLoggerDetail(.sections).tsx and relied on the bundler's
default-import interop. Next is fine with both, but every test that renders the
component died on the release tip:
- node:test / tsx: ERR_UNKNOWN_FILE_EXTENSION ".css" — request-log-detail-layout,
request-log-detail-stream, request-logger-detail-copy-all,
request-timeline-lane-allocation (4 unit shards red on every PR).
- esbuild bundle-safety check (media-page-client-browser-bundle): cannot resolve the
.css specifiers.
- node ESM resolves the package's CJS `main` (no `exports` map), so the default
import is the module namespace: "Element type is invalid … got: object".
Fix at the source: the two stylesheets are @imported from src/app/globals.css
(same as material-symbols / fumadocs), and the component unwraps `mod.default ??
mod` like redisQuotaStore/keytar already do. Also adds the 5 vi strings #11703
introduced (requestLogger.detail.{collapseAllLevels,collapseOneLevel,
currentExpandLevel,expandOneLevel,expandAllLevels}) — vi has strict parity.
Refs #11703
* refactor(dashboard): move the react18-json-view interop into shared/components/jsonView.ts
RequestLoggerDetail.tsx is frozen by check:file-size (1111 lines, cannot grow); the
inline interop pushed it to 1118. One tiny module serves both components and keeps
the CSS-import warning in a single place.
- complexity-baseline.json: 2774 -> 2681 (npm run quality:ratchet-style --update, measured on release/v3.8.51 tip after this session's merge batch)
- quality-baseline.json cognitiveComplexity: 1223 -> 1197 (same measurement)
- file-size-baseline.json: gateways.ts 1347 -> 1348, the #11771 rebaseline that only ever landed in a scratch worktree, never in the merged commit
Prompted by PR #11847's stale ratchet-shrink numbers (measured on release/v3.8.50, both below what the current tip actually measures — 2351 vs 2681 real, 1060 vs 1197 real) — remeasured directly instead of merging the stale values.
Remove a reserva estática global e passa a gatear o roteamento pela ocupação real do lease exclusivo ativo, com boa cobertura de testes (5 arquivos, 54 casos, todos verdes no worktree combinado). Typecheck limpo.
Dois ajustes feitos por cima antes do merge:
1. **26 arquivos `.pyc` órfãos removidos** (`scripts/ops/__pycache__/…`, `tests/unit/ops/__pycache__/…`) — cache compilado do Python sem relação com o fix de lease, provavelmente commitado sem querer do ambiente local.
2. **Conflito em `src/app/api/keys/[id]/route.ts`**: mantida a checagem mais ampla desta PR (`instanceof ApiKeyPolicyInvariantError || código LEASE_KEY_POLICY_INVALID`), que é um superset da versão anterior — cobre o caso original e o novo caminho de erro do lease.
Obrigado pela contribuição!
Feature grande e bem construída: exportação contínua de call logs para destinos plugáveis (BigQuery primeiro). Revisei especificamente o tratamento de segredos (`src/lib/logExport/secrets.ts`) e a migração — encryption gate real (`requiresEncryptionKey` recusa gravação em texto plano quando `STORAGE_ENCRYPTION_KEY` não está setada), redação antes de qualquer resposta de API, e a migração cria a tabela com `enabled=0`/`include_bodies=0` por padrão (opt-in, sem exportar nada até o operador configurar). 62/62 testes focados verdes, typecheck limpo.
Resolvido o conflito com o barrel `src/lib/localDb.ts` (removido nesta mesma sessão, #11795 fase 5 — todo consumidor já migrado para `src/lib/db/*`); a PR só adicionava um re-export nele, que não é mais necessário. Obrigado pela contribuição!
Mostra a % de cache nos logs de requisição, com teste próprio (`request-logger-cache-percentage.test.ts`). Validado no worktree combinado do lote.
Pequeno ajuste feito por cima: `formatCachePercentage` movida de `RequestLoggerV2.tsx` para `src/shared/utils/formatting.ts`. `RequestLoggerV2.tsx` importa `RequestLoggerDetail`, que desde o #11703 (mergeado nesta mesma sessão) importa CSS bruto de `react18-json-view` — algo que o Node native test runner não consegue carregar. O teste original importava a função direto do componente e quebrava por causa dessa cadeia de import, não por bug na PR. Movida a função (pura, sem dependências) para o utils compartilhado; ajustado o import do componente e do teste. Typecheck limpo, teste passando (6/6).
Adiciona 5dive como configure target, com teste de regressão próprio (`tests/unit/cli/setup-5dive.test.ts`) e strings i18n em 12 locales. Validado no worktree combinado (typecheck limpo, 26 testes focados).
Nota: um dos subtestes desse arquivo ("falls back to the local server when no context") depende de não haver contexto CLI ativo em `~/.omniroute/` — nesta máquina de desenvolvimento compartilhada existe um contexto real configurado, então o teste lê a config real em vez do fallback via `PORT`. Confirmado que é vazamento de ambiente do devbox (não do CI): reproduzido isoladamente, rastreado até `resolveActiveContext()` lendo `~/.omniroute/*.json` antes de cair no fallback de `PORT`. Não bloqueia o merge, mas fica registrado — o teste merece ficar hermético (mockar/isolar o data dir) numa limpeza futura.
Corrige divergência de scoring no relatório de saúde do auto-combo, com testes atualizados em `combo-resolve-auto-strategy-split.test.ts` e `combo-scoring-inspector.test.ts`. Validado no worktree combinado. Obrigado!
Dá aos alvos de extended-thinking o orçamento de prontidão de reasoning, com cobertura de teste ampliada em `stream-readiness-policy.test.ts`. Validado no worktree combinado. Obrigado!
Envia metadata completo do loadCodeAssist (ideType/platform/pluginType como enums numéricos) para o Antigravity, com teste de regressão atualizado. Validado no worktree combinado. Obrigado!
Fix de contraste no tooltip do gráfico de custos (fundo opaco + cor de texto legível). Mudança isolada de CSS/classe, validada no worktree combinado. Obrigado!
Resolução de links relativos entre Fumadocs e a wiki do GitHub, com dois arquivos de teste novos e bem focados (`docs-link-resolver.test.ts`, `sync-wiki.test.ts`). Validado no worktree combinado. Obrigado!
Fix correto — CLI agora sonda IPv4 e IPv6 no probe de prontidão do servidor, com teste de regressão próprio (`tests/unit/cli-waitForServer.test.mjs`). Validado no worktree combinado (typecheck limpo, teste focado verde). Obrigado!
Correção pequena e correta — `passthroughModels: true` para o Vercel AI Gateway. Validado no worktree combinado do lote (typecheck limpo, gates estáticos verdes). Obrigado!
Resynced onto release/v3.8.51 (originally targeted main; retargeted since the default branch is release/v3.8.51). One real conflict in open-sse/handlers/chatCore.ts, but it was entirely unrelated to this PR's actual purpose: the antigravity-aware lockExactModel branching and deferAntigravityQuotaStateToCaller state exist on main but haven't been synced to release/v3.8.51 yet (confirmed by diffing your branch against its own main merge-base — the only change there was a Prettier reformat, not new logic). Discarded that unrelated drift and kept the release tip's current quota-lock shape; the onStreamComplete plugin wiring itself is untouched and intact. typecheck:core clean, 13/13 plugin delivery tests pass. Thanks for the thorough three-layer root-cause writeup.
Boarded with 8 other PRs in one combined worktree: typecheck:core, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles, check-native-deps all green; 75/75 focused tests pass. Live-reproduced root cause (byte-for-byte reproduction/removal of the malformed schema) is solid evidence. Thanks for tracing this to the builtin skill schemas rather than stopping at "provider outage".
Boarded with 8 other PRs in one combined worktree: typecheck:core, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles, check-native-deps all green; 75/75 focused tests pass. Fixes a genuinely confusing failure mode — matches the documented TROUBLESHOOTING.md symptom exactly. Thanks for the preflight check and clear diagnostics.
Boarded with 8 other PRs in one combined worktree: typecheck:core, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles, check-native-deps all green; 75/75 focused tests pass. Discovery-only as claimed — nothing reads the new tag yet, dashboard quota widget stays gated by USAGE_SUPPORTED_PROVIDERS. Thanks.
Boarded with 8 other PRs in one combined worktree: typecheck:core, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles, check-native-deps all green; 75/75 focused tests pass. Clean, well-scoped env-override with correct blank-value handling and a startup log naming the resolution source. Thanks.
Boarded with 8 other PRs in one combined worktree: typecheck:core, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles, check-native-deps all green; 75/75 focused tests pass. Reviewed the security fencing closely — the status query fences on lease_owner_hash + api_key_id + generation + state=ACTIVE + not-expired, gated behind the existing lease:exclusive scope check. configuredConnectionName() correctly excludes email-derived fallback labels from the response. Test coverage explicitly verifies foreign key / different owner / stale generation all fail closed with 409, and no metadata leaks for released/expired/invalidated/missing leases. Thanks for the careful privacy-safe design.
Boarded with 8 other PRs in one combined worktree: typecheck:core, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles, check-native-deps all green. Retargeted from main to release/v3.8.51. Confirmed ollamaTransform.ts was the only streaming transform not using a persistent { stream: true } decoder — matches the pattern already established in responsesTransformer.ts. TDD repro included. Thanks for finding an unreported bug.
Boarded with 8 other PRs in one combined worktree: typecheck:core, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles, check-native-deps all green. Retargeted from main to release/v3.8.51. One-line baseUrl fix with a live endpoint probe documenting the exact 404→401 transition — solid verification. Thanks.
Boarded with 8 other PRs in one combined worktree: typecheck:core, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles, check-native-deps all green; 75/75 focused tests pass. Verified the console fallback → null fix removes the duplicate plain-text line while the structured pino log is unaffected. Thanks.
Boarded with 8 other PRs in one combined worktree: typecheck:core, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles, check-native-deps all green; 75/75 focused tests pass. Trivial, correct log-level fix — both conditions are already handled gracefully by callers. Thanks.
Phase 3 creates the GitHub Release with the curated notes right after the tag push,
so softprops always finds an existing body; generate_release_notes must be false
on every event, not only on workflow_dispatch (v3.8.48 shipped with the auto block
appended; the body sits ~3 KB under the 125,000-char cap). Same hunk as main (#12086);
the #12085 squash did not carry it.
Refs #12084
Boarded with #11954/#11953/#11951/#11952 in one combined worktree: typecheck:core, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles all green; 85/85 focused tests pass. Verified both halves of the gap directly: isCodexFreePlan() (open-sse/executors/codex/tools.ts) only checks workspacePlanType, while codexImport.ts normalizes the JWT plan into providerSpecificData.chatgptPlanType — confirmed imported free-plan accounts would bypass the existing guard. Thanks for tracing the full import-to-guard path.
Boarded with #11954/#11953/#11951/#11948 in one combined worktree: typecheck:core, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles all green; 85/85 focused tests pass. Confirmed the Antigravity Gemini path only forwarded aspectRatio into generationConfig, dropping the requested size tier entirely. Thanks for the fix and the 3:4/2K regression coverage.
Boarded with #11954/#11953/#11952/#11948 in one combined worktree: typecheck:core, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles all green; 85/85 focused tests pass. Confirmed the codex registry entry was missing forceStream: true while every other JSON-only-client provider (cline, clinepass, ghe-copilot, kimi, zed-hosted, chatgpt-web-codex) already has it. Clean reuse of the existing bridge, no Codex-specific response handling needed. Thanks!
Boarded with #11954/#11951/#11952/#11948 in one combined worktree: typecheck:core, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles all green; 85/85 focused tests pass. Verified the exact gap: invalidateDbCache("connections") after _updateConnectionRow() (src/lib/db/providers.ts:599) is only reached inside the retired-provider special-case branch (line 610-617) — the common return path (line 619) skips it entirely, confirmed. Thanks for the precise fix.
Boarded with #11953/#11951/#11952/#11948 in one combined worktree: typecheck:core, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles all green; 85/85 focused tests pass. Verified the root cause directly: createProviderConnection() matches existing rows via provider_specific_data.workspaceId (src/lib/db/providers.ts:462-470), but codexImport.ts only emitted chatgptAccountId — confirmed re-import would miss the intended stable-identity match. Thanks for the careful diagnosis.
Validated: actionlint clean on all three touched workflows, check-api-typecheck.mjs OK (289 pre-existing, all frozen) after boarding on top of #12094. Confirmed the .trivyignore justification against the documented CVE Variance process (docs/security/SUPPLY_CHAIN.md) — has tracking issue #12084, expiry before the v3.8.51 tag, and a real technical reason the .so can't be rebuilt in this repo. Scorecard branch guard correctly targets the actual default branch (release/vX.Y.Z), not a hardcoded main.
Validated: check-api-typecheck.mjs OK (289 pre-existing, all frozen), typecheck:core clean, 8/8 check-api-typecheck.test.ts pass. Spot-checked two of the six fixes directly — the webhooks/[id]/test/route.ts duplicate import is confirmed removed (real ESM defect), and the volcengine-plan strict-boolean-narrowing fix (`validation.success === false` vs `!validation.success`) is behaviorally identical since `.success` is a strict boolean. This unblocks every other open PR into release/v3.8.51 that was landing red on the new API Route Typecheck gate — including #12085.
Resynced onto the release tip — the FREE_CATALOG_CURATED_AT bump conflicted with a later bump already on the tip; resolved to today's date since real content is landing. typecheck:core clean, 23/23 focused tests pass (free-model-catalog, free-models). Verified live against OpenRouter's own /api/v1/models pricing as claimed. Thanks for the new free-tier entry.
Resynced onto the release tip. Two fixes applied during boarding: (1) the branch forked before the recent optionalDependencies placement of @huggingface/transformers and onnxruntime-node — its own diff re-added both into "dependencies" as duplicates alongside the real new dependency (react18-json-view); removed the duplicates, ran npm install to sync the lockfile. (2) config/quality/dependency-allowlist.json referenced the wrong package name (react-json-view-lite, an earlier iteration per the PR body) — the code actually imports react18-json-view; fixed the allowlist entry to match. RequestLoggerDetail.tsx crossed its frozen file-size cap (1018->1111); rebaselined with a note — the PR does split out the new logic (RequestLoggerDetail.sections.tsx, JsonTreeExpandControls.tsx, useTimestampTitles.ts, jsonTreeExpandStore.ts, all well under cap), the growth here is irreducible wiring. typecheck:core, check:dashboard-typecheck, check:file-size, check-deps all green after resync; 8/8 vitest + 11/11 native tests pass. Nice, well-structured 6-commit feature with full i18n and good test coverage. Thanks!
* test(cli): align the nodes --base-url contract test with #12033#11860 asserted that `nodes add/update/validate` must NOT register `--base-url`
(reserved for the global server target); #12033 (issue #11999) then registered
it on purpose so `omniroute nodes add --provider p --base-url <url>` stops being
rejected by Commander's global option. Both PRs landed and the older test turned
the base red on unit shard 2/4 (`Unit Tests fast-path (2/4)`, run 33293442568).
The test now asserts the current contract: both flags are registered and each
parses into its own option; the server-target/payload separation keeps its own
test right below.
* test(mutation): register lkgp-stale-pin-exhaustion-11911 in tap.testFiles
38e2baa879 (#11911) added a unit test covering src/shared/utils/circuitBreaker.ts
without listing it in stryker.conf.json tap.testFiles, so check:mutation-test-coverage
--strict (Fast Quality Gates) is red on the release tip.
* fix(db): drop three consumer-less 1proxy exports the deleted localDb barrel was masking
50bc8ab8aa (#12055) removed the @/lib/localDb barrel; its re-exports were the only
thing keeping getOneproxyStats / deleteOneproxyProxy / clearAllOneproxyProxies (and
the private mapStatsRow + OneproxyStats type) 'used' for knip. The 1proxy routes
are 308 compat redirects to /api/settings/free-proxies since v3.8.4, so nothing
calls them: check:dead-code went 413 -> 419 on the release tip (baseline 416).
Back to 416 with typecheck:core, eslint and check:db-rules green.
* chore(mutation): drop the duplicate tap.testFiles entry — #12082 already registered it
Boarded with #11756 (a duplicate fix for the same underlying issue #11650). Compared both implementations directly: this one is technically superior — guards the json_extract() call with json_valid(metadata) so malformed/legacy metadata returns no match instead of throwing a 500, and covers genericBackend.ts/obsidianBackend.ts in addition to sqliteBackend.ts. #11756 only touched SQLite and had no malformed-JSON guard. Closing #11756 with credit. Resynced onto the updated release tip: the test file's `await import("../../src/lib/localDb.ts")` broke after #12055 deleted the barrel earlier this session (your branch forked before that migration) — fixed to import updateSettings directly from @/lib/db/settings, matching the pattern already used by other integration tests. typecheck:core, check:dashboard-typecheck, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles, check-deps all green; 4/4 integration + 35/35 vitest pass after resync. Thanks for the thorough, well-tested fix.
Boarded in a combined worktree: typecheck:core, check:dashboard-typecheck, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles, check-deps all green. Clean, self-contained addition (5 new files, 0 modifications to existing code) that mirrors the existing dashboard-typecheck baseline-ratchet pattern. Thanks for closing a real coverage gap — API routes had no dedicated typecheck gate.
Boarded in a combined worktree: typecheck:core, check:dashboard-typecheck, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles, check-deps all green; 10/10 focused tests pass. Real SSRF gap confirmed — the default "block-metadata" guard mode fell through to the unchecked parseOutboundUrl() while 3 other call sites of the same guard mode already routed through parseAndValidateNonMetadataUrl(). Good catch that the existing test suite only ever exercised "public-only" explicitly. Retargeted from the stale release/v3.8.50 base to release/v3.8.51. Thanks for closing a real cloud-metadata SSRF exposure.
Boarded in a combined worktree with 6 other PRs: typecheck:core, check:dashboard-typecheck, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles, check-deps all green. Verified the root-cause diagnosis directly against the code: isConnectionUnavailableToAuxiliaryActivity() does return true for any connection reachable by an active exclusive lease regardless of whether the lease is actively serving a request, confirming the fix's scoping is correct. The change is surgically limited to providerLimits.ts's live-usage-fetch path — the shared isolation function and its other call sites (warmupScheduler, quotaAutoPing, modelTestRunner, etc.) are untouched. Well tested (214 lines across 3 test files). Thanks for tracking this down.
Boarded with #11741 (a duplicate fix for the same underlying issue #11739). Compared both implementations directly: this one is technically superior — a dedicated resolveIncomingCorrelationId() helper that strips CRLF (header-injection prevention) and bounds length to 1-256 chars, with 4 unit tests covering those edge cases. #11741's simpler `header || generateRequestId()` has no sanitization. Closing #11741 with credit. Validated in a combined worktree: typecheck:core, check:dashboard-typecheck, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles, check-deps all green; 84/84 + 43/43 focused tests pass across this batch. Thanks for the careful sanitization work.
Boarded with #12082 in one combined worktree: typecheck:core, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles all green; 77/77 focused tests pass. Genuinely conservative as described — dev-only Tailwind/webpack scanning bounds, production chunking untouched. The later phases of #12074 (2/3/4/4b) are being held for a dedicated review given their combined architectural weight (DB init graph, credential refresh, process lifecycle, network dispatch boundary) — flagged separately on those PRs. Thanks for the clean Phase 1 baseline.
Boarded with #12075 in one combined worktree: typecheck:core, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles all green; 77/77 focused tests pass. CI-contract-only reconciliation as described — no production behavior change, and the referenced files (lkgp-stale-pin-exhaustion-11911.test.ts, cli-nodes-commands.test.ts) confirmed already present and correctly aligned. Thanks for keeping this separate from the dev-bundler phase PRs.
Boarded with #12003/#11841/#11840 in one combined worktree: typecheck:core, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles all green; 24/24 focused tests pass. relayMode is opt-in and defaults to standard, so this is backward-compatible as claimed — verified the plumbing through resolveUniversalHandoffConfig/resolveContextRelayConfig/selectMessagesForSummary. Thanks for the clean, well-tested addition.
Boarded with #12003/#11841/#11839 in one combined worktree: typecheck:core, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles all green; 24/24 focused tests pass. Contained fix — preserves the unstripped model string for passthrough providers (cline/kilocode) only when the combo actually redirected to a passthrough provider. Thanks for the regression coverage.
Boarded with #12003/#11840/#11839 in one combined worktree: typecheck:core, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles all green; 24/24 focused tests pass. Clean, well-contained addition mirroring the existing systemTransforms hot-reload pattern, tested for both set and cleared states. Thanks for the tidy runtime-config feature.
Boarded with #11841/#11840/#11839 in one combined worktree: typecheck:core, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles all green; 24/24 focused tests pass. Both fixes are surgical and well-reasoned: explicit ensureDbInitialized() call for MCP stdio (verified the function exists at src/lib/db/core.ts:1496) avoids a startup race, and the reasoningContent fallback prevents empty message.content when only reasoning was returned. Thanks for tracking down both root causes.
Resynced onto the release tip after #12051/#12052/#12053 landed. Same LKGP-clear conflict as #12053 (kept the current clearStaleLKGP() helper at both call sites). One additional issue this final phase's combined-worktree validation surfaced: clearStaleLKGP() itself (added by #12013, which none of the 4 phase PRs could have seen since it landed after they were authored) still had a dynamic `await import("@/lib/localDb")` — a real break once this PR deletes the barrel. Fixed to `await import("@/lib/db/settings")`, matching the direct-import pattern used at every other call site. typecheck:core, check-db-rules, check:cycles, and the eslint-import-boundaries regression test (3/3, including "G14 rejects localDb barrel imports") all green after resync — zero barrel-importing production files remain. Nice clean 5-phase migration, and thanks for taking on the full #11795 cleanup.
Resynced onto the release tip after #12051/#12052 landed. One real conflict in open-sse/services/combo.ts at both LKGP-clear call sites (handleComboChat + round-robin path): the release tip already has #12013's clearStaleLKGP() helper, which this PR's branch predates — kept the current helper call at both sites, discarding the pre-refactor inline pattern. typecheck:core and the open-sse test suite (vitest, 9/9 on volumeDetector) both green after resync. Thanks for the well-scoped Phase 4 migration.
Confirmed the bug is real and unfixed on the current tip before merging: `scripts/build/prepublish.ts` line 332 was still calling `execFileSync(NPX_BIN, ...)` directly (raw win32 npx.cmd spawn), the exact CVE-2024-27980 shim pattern the file's own header warns about. Root cause, fix, and evidence match — routing through the existing `runBuildTool()` helper. Thanks for catching the one call site the earlier refactor missed.
Boarded together with Phases 2, 4, 5 (#12051, #12053, #12055) and validated in one combined worktree: typecheck:core, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles, check-db-rules all green. Mechanical import-path migration only, no behavior change. Thanks for the phased, well-tested cleanup.
Boarded together with Phases 3-5 (#12052, #12053, #12055) and validated in one combined worktree: typecheck:core, check:file-size, check:changelog-integrity, check:complexity, check:cognitive-complexity, check:cycles, check-db-rules all green. Mechanical import-path migration only, no behavior change. Thanks for the phased, well-tested cleanup.
When an auto/*/lkgp combo target failed into exhaustion (e.g. an unauthenticated free-tier 401) or was skipped pre-dispatch (cooldown, model lockout, unavailability), the Last Known Good Provider pin was never cleared — so subsequent requests kept re-selecting the same dead provider, causing repeated failures and mass-skipping instead of falling through to a healthy target. Centralizes invalidation into clearStaleLKGP(), invoked from both handleComboChat and handleRoundRobinCombo on exhaustion, pre-dispatch skip, and body-specific 400 termination.
Real production incident (2026-08-29): the resource-pressure guard ratioed raw cgroup v2 memory.current (which counts reclaimable page cache) against memory.max, so a busy host with ~3GiB of page cache latched a global 503 across every model for 26 minutes even though PSI/OOM/memory.events all showed zero real pressure — the kernel would have reclaimed those pages instantly. Fix: ratio the working set (current - file) for the trip/recovery check, falling back to the raw ratio when memory.stat is missing/stale/zero (never clamping to a false zero-pressure reading).
12 new tests including direct incident reproduction (raw 95%/workingset 32% stays normal) + bug-injection round trips. Full resource-pressure + admission suites green (48/48, re-verified in this batch together with the other 3 PRs: 57/57).
Commander's top-level global --base-url option was shadowing the flag on `omniroute nodes add/update/validate`, rejecting the command with "required option '--endpoint <url>' not specified" even when --base-url was correctly supplied. Now both flags are accepted on all three subcommands, falling back to whichever the user passes.
Real production incident (2026-08-29): the crash guard #11556 introduced defaulted its logger to `log ?? console` — console is an object, not a function, so a burst of client aborts (ECONNRESET) reaching the process-level guard threw TypeError inside the uncaughtException handler itself and killed the server, twice in three minutes. Fix: default to console.warn.bind(console).
Bug-injection round trip confirms the new test fails on the old default and passes on the fix. Existing guard suite stays green: 9/9 (verified together with the new test).
Extracts videoBridge.ts's per-part loop body, whole-result cache identity/key helpers, and describeWithVisionModel into a new videoBridgePipeline.ts with explicit port boundaries (VideoMediaBrokerPort, VideoAudioTranscriptionPort, VideoDrilldownPort). videoBridge.ts shrinks 820→255 lines, now only handling request traversal, policy resolution, aggregation, and response payload. Moved as whole blocks, parameterized rather than rewritten — byte-for-byte traceable to the pre-extraction code.
Rebased onto the tip after sibling #12009 (FU-05 core) landed first and bumped the result-cache version v4→v5 in videoBridge.ts — that same bump (plus its explanatory comment) is now carried into the extracted videoBridgePipeline.ts instead. Re-validated: 20/20 focused tests, typecheck clean.
FU-06 (Audio Bridge STT orchestration): one download, two extractions — takes already-downloaded video bytes and extracts bounded mono 16kHz PCM WAV via the loopback broker's new mode=audio operation, sharing the exact same queue/deadline/byte budgets as the frame path. Dual opt-in (operator setting default false + per-request), only reaches the STT call when both are on.
Rebased onto the tip after sibling #12011 (subtitle mode) landed first, both touching the same broker route/client — combined additively so frames/audio/subtitles all share the one extractionQueue singleton. Re-validated: 59/59 focused tests pass.
Reconciles the Video Bridge FU-01..09 backlog docs against verified code and GitHub state (ground truth established first, per this repo's Documentation accuracy rule), correcting a real gap in GUARDRAILS.md: the transcript source field was documented as validated without noting OmniRoute didn't yet verify server-side extraction — exactly the gap #11652 (now merged as #12009) closes. Refs #11661, not Closes — truthfully closing it needs the sibling PRs' actual landed state folded back in, left as an explicit follow-up.
FU-08 (Refs #11655): drill-down producer/consumer lifecycle on top of the existing cache substrate, without modifying it — new VideoDrilldownLifecycle (opaque sha256 handles, principal-bound resolve/delete with no existence oracle, preview/standard/detail multiresolution variants, 8-frame/32MiB page budget) plus a new authenticated remote-consumer route, both opt-in (default false).
FU-07/FU-09 promotion-evidence harness (Refs #11656): delivers the manifest schema, deterministic fixture recipes, metrics aggregator, and promotion-verdict evaluator #11656 asks for — deliberately does NOT deliver the promotion verdicts themselves (they require real models against real fixtures on a live host, HOLD with explicit reason instead of any fabricated result). New files only, no collision with sibling PRs.
FU-05 subtitle adapter (Refs #11659 — deliberately not Closes: the adapter is not yet wired into the live describeVideoPart path, that composition point is sibling #12009 which just landed): server-owned, loopback-only ffprobe/ffmpeg subtitle extraction that legitimately earns the "embedded" provenance label, mirroring the existing frame-extraction lifecycle. Broker route now also serves ?subtitles=1, stamped with the shared broker fingerprint so the client-side adapter can verify the payload actually came from the trusted process. Bounded, ReDoS-safe WebVTT parser.
FU-05 core (closes#11652): caller-supplied Video Bridge transcripts had no bounded, deterministic contract — a client could self-assert source: "embedded"/"audio-bridge" and it was accepted verbatim. normalizeVideoTranscript gained a code-only trustedSource seam unreachable from request-body JSON; without it, any cue declaring embedded/audio-bridge is reclassified to client. Added budgets (256 cues, 4096 code units/cue, 4KiB/cue, 64KiB total), malformed-surrogate rejection, focus-window scoping, deterministic cross-source reconciliation, and bumped the result-cache version v4→v5 so old-contract cache entries can never serve new-contract requests.
All 187 videoBridge* tests pass (185 pass, 2 unrelated pre-existing skips).
Adds opt-in modelVisibilityAllowlist/modelVisibilityDenylist settings so an operator can curate exactly which models GET /v1/models advertises, mirrored into auto/* combo candidate pools (the same trap #6512 fixed for hidePaidModels). Default off, no behavior change for anyone who doesn't opt in.
TDD: 4 new test files, 22/22 passing (16 node:test + 6 vitest) + regression sweep across virtual-auto-combo/hide-paid/hide-auto-no-think suites (21/21).
Rebased onto the updated tip (a sibling #9133 landed first, same file) — kept both rebaseline annotations in file-size-baseline.json and set the value to the real measured line count after both merged.
Fixes three defects in the "update doesn't restart the running process" bug class: CLI update guidance now detects a live server and tells the operator to restart instead of implying the update is already live; the dashboard's Update button tries OmniRoute's own PID-file supervisor before falling back to pm2 instead of hardcoding pm2 and silently skipping; getLatestVersionFromNpmCli now uses --prefer-online (same fix pattern as #4376). TDD throughout, 63/63 targeted regression tests pass.
Fixes a copy-paste label typo (Hermes-4-405B mislabeled "7B") in both the registry and the free-model catalog data, spotted in the #11861 comment thread. TDD: 3/3 tests, generic parameter-size consistency check + exact regression guard.
Local no-API-key providers (ollama-local, lm-studio, vllm, etc.) were invisible in the Qdrant embedding-model dropdown because configuredProviders required a real apiKey or OAuth. Extended the filter to also include providerAllowsOptionalApiKey(connection.provider) — the same canonical helper already used for the identical check elsewhere. TDD: 21/21 integration tests pass (was 20/21 before the fix).
Fixes the blocking Lint job's own ci.yml cache: PR #11963 removed the stale restore-keys fallback from quality.yml but left ci.yml's two "Restore ESLint file cache" steps carrying the same prefix-match fallback that lets a cache from a different lint config report stale per-file verdicts. Byte-level parity with #11963's already-merged fix.
Deliberately half of #11600 — the other half (run-eslint-json.mjs) is covered by PR #11983 from a parallel session, so the two don't collide on the same file.
Turbopack had 31 GB on omniroute-113-6 and still panicked
(TurbopackInternalError: there must be a path to a root, run
33253576569). The same tree's arm64 webpack build on hosted ARM
succeeded. Dockerfile already documents webpack as the Docker
escape hatch. Keep amd64 on the one omni-build slot (#12048).
* docs(ops): the .113 heavy-build ceiling is one runner, not two
Two concurrent next-builds (15.4 GB + 17.2 GB RSS) OOM-killed one on 2026-08-29 17:26 UTC;
systemd booked the kill on the other runner's unit and its job died with the same
"shutdown signal" text a hosted-runner OOM shows. omni-build now lives on
omniroute-113-5 only; 113-6 keeps omni-release. The janitor ceiling counts every
listener on the box (4 OmniRoute + OmniHeuris + OmniMind = 6). The second heavy slot
returns when the Proxmox VM gets more RAM; the exact command is in the doc.
* docs(ops): apply the single-heavy-slot text (previous commit only carried formatting)
* fix(release): the packaged-app smoke verifies the database opened, not a driver line the primary path never prints (release/v3.8.51 twin of #12032)
Same change as #12032 on main: the packaged app opens SQLite during the smoke but
its primary open path prints no "[DB] Driver: …" line (only the recovery path and
the sql.js fallback do), so the #7592 assertion failed every Linux release leg. The
guard rejects the sql.js fallback line, accepts a native driver line, and otherwise
accepts demonstrable database activity; after readiness the smoke requests
/api/monitoring/health and waits for that activity outside the readiness loop.
electron-smoke-script suite 10/10.
* fix(release): reapply the smoke rework on top of release/v3.8.51's own copy of the script
The previous commit copied main's file wholesale and dropped this branch's
ensureSmokeEnvDirs(currentPlatform) fix and its tests; this reapplies only the
DB-open evidence change as a patch. electron-smoke-script suite green.
* fix(ci): stop hosted docker-publish OOM and unpaint Build (advisory)
docker-publish was firing 8 concurrent hosted builds on every merge
storm; each died ResourceExhausted in npm run build (#11976). One
publish per ref, webpack instead of Turbopack so native RSS stays
inside the V8 heap we can cap. Build (advisory) is skipped: continue-on-error
still reports FAILURE and was painting every fork PR red.
Closes#11976
* fix(ci): run docker-publish amd64 on omni-build and share the heavy lane
The .113 box is 31 GB / 32 cores — enough for one next-build. Hosted
ubuntu-24.04 is ~7 GB and ResourceExhausted every publish (#11976).
amd64 now targets [self-hosted, omni-build] (Turbopack) when
USE_VPS_RUNNER is on, joins the existing heavy-build-main group so it
queues beside ci.yml Build instead of becoming a third heavy, and
falls back to hosted + webpack if the VPS is off. arm64 stays on
ubuntu-24.04-arm with webpack (no ARM box).
* test(ci): align the advisory-build contract with the hosted-OOM skip
if: ${{ false }} tripped zizmor obfuscation (194→195). Bare if: false
skips the job without a new finding. The #7307 test now pins the skip
and keeps the job body as the restore recipe.
* fix(release): resync the electron lockfile, build a dispatch from a repaired ref, keep curated notes, attach the SBOM on dispatch (release/v3.8.51 twin of #11982 + #12020)
Same four changes as #11982 and #12020 on main, applied to this branch's own copies:
- electron/package-lock.json regenerated (271 -> 284 entries): the optional
electron-builder-squirrel-windows subtree was missing and `npm ci` refused the lock
(EUSAGE) on the Linux and macOS legs; a clean `npm ci --ignore-scripts` on the
result exits 0.
- electron-release.yml: `build_ref` dispatch input (default: the version tag) and
`generate_release_notes` only on the tag push (a re-attach dispatch appended
GitHub's auto notes to the curated body on v3.8.50).
- npm-publish.yml: the SBOM attaches to the GitHub Release on workflow_dispatch
publishes too, whenever a release for the tag exists.
actionlint and prettier clean; electron-release-desktop-channel-8949,
electron-release-efficiency, electron-release-latest-yml.repro, check-workflows
and npm-publish-artifact-provenance suites pass.
* fix(release): validate build_ref in the validate job before any checkout uses it
CodeQL (actions/cache-poisoning/poisonable-step, high) on release/v3.8.51 — the
default branch: a raw dispatch input checked out next to setup-node's npm cache is a
cache-poisoning vector. The input now goes through the validate job's regex
allowlist (main or release/vX.Y.Z, empty = the version tag) and every build job
checks out needs.validate.outputs.build_ref, never the input itself.
* fix(release): drop the build_ref input — a dispatch builds the ref it is dispatched on
CodeQL (actions/cache-poisoning/poisonable-step) tracks the input through the
validate job's output regardless of the regex allowlist: an input-controlled
checkout next to setup-node's npm cache on the default branch is a cache-poisoning
vector. The ref is not an input any more; the checkouts use github.ref, so
`gh workflow run electron-release.yml --ref v3.8.50 -f version=v3.8.50` rebuilds
the tag and `--ref main` builds the repaired line. The tag-push path is unchanged.
* fix(sse): stop the auto-combo candidates inspector from dropping blocked rows (#9133)
prepareVirtualAutoComboInputs applied filterResilienceBlockedCandidates
before the #7819 read-only candidate inspector ever saw the pool, so a
model-locked or cooled-down candidate silently disappeared from
/auto-combo/*/candidates instead of showing up as reachable:false with a
reason (modelLocked/connectionCooldown/breakerState were dead fields by
construction). Add an opt-in `skip` parameter so the inspector builds its
own unfiltered pool; routing (createVirtualAutoCombo/createBuiltinAutoCombo
called without a prepared override) is unchanged. Also aligns
isModelLocked's model argument to the bare model id, matching every lock
writer and the routing-side filter, instead of the "provider/model" string.
Regression test: tests/unit/auto-combo-candidates-locked-model-visible.test.ts
(red before the fix — locked account's row silently missing; green after).
* chore(quality): register the #9133 regression test in stryker tap.testFiles
tests/unit/auto-combo-candidates-locked-model-visible.test.ts covers
open-sse/services/accountFallback.ts (via isModelLocked) but wasn't listed,
so its mutant kills wouldn't count toward mutation coverage.
---------
Co-authored-by: Markus Hartung <mail@hartmark.se>
* fix(cli): drop the never-produced dist/index.cjs requirement from prepublish's opencode-plugin skip check (#11787)
* test(build): resolve tsup/npm portably in the #11787 regression test instead of a hardcoded .bin path
The old test assumed @omniroute/opencode-plugin/node_modules/.bin/tsup
already existed. A fresh checkout (CI's npm ci never installs this
standalone package's own deps) has no such node_modules at all, so the
test failed with MODULE_NOT_FOUND in CI while passing locally on a devbox
that had installed it before. Mirror scripts/build/prepublish.ts's own
install-then-resolveLocalBinEntry approach.
---------
Co-authored-by: Markus Hartung <mail@hartmark.se>
enforceCodexResponsesLiteParallelToolCalls() forces parallel_tool_calls:false
at the top of CodexExecutor.execute(), but transformRequest() early-returns
the body before its RESPONSES_API_ALLOWLIST field filter only when
_nativeCodexPassthrough is set. Any request that reaches the codex
executor via the translated (non-native-passthrough) path never gets that
flag, so the allowlist filter silently deleted parallel_tool_calls right
before the fetch body was sent, reproducing the reported upstream
rejection ('X-OpenAI-Internal-Codex-Responses-Lite requires
parallel_tool_calls to be false') for every model.
Add parallel_tool_calls to RESPONSES_API_ALLOWLIST so the value survives
the translated path too. Update the sibling #2608 allowlist test that
previously asserted parallel_tool_calls gets stripped like other Chat
Completions-only fields -- it is a legitimate Responses API field that
must now survive.
Co-authored-by: Markus Hartung <mail@hartmark.se>
* fix(db): rate-limit Arena ELO fetch-failure warnings on repeated timeouts (#11500)
* test(quality): split the #11500 fetch-failure-dedup tests into their own file
tests/unit/arena-elo-sync.test.ts crossed the 1000-line new-test-file cap
(file-size gate, PR mode). The two new tests don't need the file's DB
fixture (fetchArenaLeaderboards() never touches the DB), so they move to a
self-contained sibling file instead of growing the frozen suite.
---------
Co-authored-by: Markus Hartung <mail@hartmark.se>
flux/kontext is catalogued with isMarket: true, so handleKieImageGeneration
routed it through KIE's unified Market createTask endpoint with
model: "flux/kontext". KIE does not expose Flux Kontext through the Market
catalog at all -- it lives under a dedicated API tree
(POST /api/v1/flux/kontext/generate, poll GET /api/v1/flux/kontext/record-info,
models flux-kontext-pro/flux-kontext-max) -- so the Market endpoint rejected it
with "model name not supported", matching the reporter's exact error text.
Special-case flux/kontext ahead of the isMarket branch so it hits the
dedicated endpoint/payload shape instead of being treated as a Market entry.
z-image/4.0-*/4.5-* remains intentionally untouched (still blocked on
reporter/live confirmation per the existing in-code comment).
Co-authored-by: Markus Hartung <mail@hartmark.se>
2026-08-29 08:10:13 -03:00
1912 changed files with 139303 additions and 31659 deletions
--body "Living tracker for the velocity-phase baseline budget (docs/architecture/QUALITY_GATES.md → Velocity phase). One comment per nightly run; the newest comment is the current state." \
API routes follow a consistent pattern: `Route → CORS preflight → Zod body validation → Optional auth (extractApiKey/isValidApiKey) → API key policy enforcement → Handler delegation (open-sse)`. No global Next.js middleware — interception is route-specific.
**Combo routing** (`open-sse/services/combo.ts`): 19 public strategies (priority, weighted, fill-first, round-robin, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline). Each target calls `handleSingleModel()` which wraps `handleChatCore()` with per-target error handling and circuit breaker checks. The `fusion` strategy is the exception: it fans out to a panel of models in parallel, then a judge model synthesizes one final answer (`open-sse/services/fusion.ts`). See `docs/routing/AUTO-COMBO.md` for the 15-factor Auto-Combo scoring + the full strategy table and `docs/architecture/RESILIENCE_GUIDE.md` for the 3 resilience layers.
**Combo routing** (`open-sse/services/combo.ts`): 19 public strategies (priority, weighted, fill-first, round-robin, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline). Each target calls `handleSingleModel()` which wraps `handleChatCore()` with per-target error handling and circuit breaker checks. The `fusion` strategy is the exception: it fans out to a panel of models in parallel, then a judge model synthesizes one final answer (`open-sse/services/fusion.ts`). See `docs/routing/AUTO-COMBO.md` for the 16-factor Auto-Combo scoring + the full strategy table and `docs/architecture/RESILIENCE_GUIDE.md` for the 3 resilience layers.
---
@@ -110,26 +110,36 @@ upstream/service level, so one unhealthy provider does not slow down every reque
| Agent features | `src/lib/{acp,memory,skills,cloudAgent}/` | [`docs/frameworks/AGENT_PROTOCOLS_GUIDE.md`](docs/frameworks/AGENT_PROTOCOLS_GUIDE.md), [`docs/frameworks/SKILLS.md`](docs/frameworks/SKILLS.md) |
@@ -254,13 +264,13 @@ Read the nearest `AGENTS.md` and the linked deep-dive before making a non-trivia
## File placement & repo-root hygiene
- **Test files**: ALL unit tests, integration tests, ecosystem tests, or Vitest files MUST strictly be placed within the `tests/` directory (e.g., `tests/unit/`, `tests/integration/`). NEVER create test files in the project root (`/`).
- **Scripts and utilities**: ALL maintenance, debugging, generation, or experimental scripts (`.cjs`, `.mjs`, `.js`, `.ts`) MUST be placed strictly inside one of the `scripts/` subfolders (`build/`, `dev/`, `check/`, `docs/`, `i18n/`, `ad-hoc/`, `quality/`, `release/`, `ci/`, `ops/`, `perf/`, `research/`, `sre/`, `vps/`, `homolog/`, `raycast/`, `skills/`, `test/`, `cli/`, `compression/`, `compression-eval/`, `devin-bridge/`, `docker/`, `features/`, `router-eval/`). One-shot or experimental code goes under `scripts/ad-hoc/`. NEVER dump loose scripts in the project root (`/`) or the top-level `scripts/` folder.
- **Scripts and utilities**: ALL maintenance, debugging, generation, or experimental scripts (`.cjs`, `.mjs`, `.js`, `.ts`) MUST be placed strictly inside one of the `scripts/` subfolders (`build/`, `dev/`, `check/`, `docs/`, `i18n/`, `ad-hoc/`, `quality/`, `release/`, `ci/`, `ops/`, `perf/`, `research/`, `sre/`, `vps/`, `homolog/`, `packs/`, `skills/`, `test/`, `cli/`, `compression/`, `compression-eval/`, `devin-bridge/`, `docker/`, `features/`, `router-eval/`). One-shot or experimental code goes under `scripts/ad-hoc/`. NEVER dump loose scripts in the project root (`/`) or the top-level `scripts/` folder.
When creating _any_ validation tests or one-off logic scripts, default to `scripts/ad-hoc/` or `tests/unit/` according to your goals. Do not pollute the `/` root context.
@@ -289,8 +299,7 @@ When creating _any_ validation tests or one-off logic scripts, default to `scrip
### Database
- **Always** go through `src/lib/db/` domain modules — **never** write raw SQL in routes or handlers
- **Never** add logic to `src/lib/localDb.ts` (re-export layer only)
- **Never** barrel-import from `localDb.ts` — import specific `db/` modules instead
- **Never** barrel-import from `localDb.ts` — import specific `src/lib/db/*` modules
- DB singleton: `getDbInstance()` from `src/lib/db/core.ts` (WAL journaling)
- Migrations: `src/lib/db/migrations/` — versioned SQL files, idempotent, run in transactions
@@ -334,6 +343,7 @@ Documentation must describe verified behavior, not plausible behavior.
### Adding a New Provider
0. Check `docs/reference/REMOVED_PROVIDERS.md` first — providers removed at their operator's request must never be reintroduced (guarded by `tests/unit/removed-providers-blocklist.test.ts`)
1. Register in `src/shared/constants/providers.ts` (Zod-validated at load)
2. Add executor in `open-sse/executors/` if custom logic needed (extend `BaseExecutor`)
3. Add translator in `open-sse/translator/` if non-OpenAI format
@@ -355,19 +365,18 @@ Documentation must describe verified behavior, not plausible behavior.
1. Create `src/lib/db/yourModule.ts` — import `getDbInstance` from `./core.ts`
2. Export CRUD functions for your domain table(s)
3. Add migration in `src/lib/db/migrations/` if new tables needed
4.Re-export from `src/lib/localDb.ts` (add to the re-export list only)
5. Write tests
4.Write tests
### Adding a New MCP Tool
1. Add tool definition in `open-sse/mcp-server/tools/` with Zod input schema + async handler
2. Register in tool set (wired by `createMcpServer()`)
3. Assign to appropriate scope(s)
4. Write tests (tool invocation logged to `mcp_audit` table)
4. Write tests (tool invocation logged to the `mcp_tool_audit` table)
4. Add OAuth/credentials handling if needed (`src/lib/oauth/providers/`)
@@ -387,7 +396,7 @@ Documentation must describe verified behavior, not plausible behavior.
1. Create installer in `src/lib/services/installers/{name}.ts` modeled on `ninerouter.ts` (use `runNpm` from `installers/utils.ts` — no shell interpolation, hard rule #13).
2. Register the service in `src/lib/services/bootstrap.ts` (add to `SERVICES[]` array and extend `buildSpawnArgsFactory()`).
3. Add a DB seed row for the new service in `src/lib/db/migrations/` (`version_manager` table, `status='not_installed'`, `auto_start=0`).
4. Create 7 API endpoints under `src/app/api/services/{name}/` (`_lib.ts`, `install`, `start`, `stop`, `restart`, `update`, `status`, `auto-start`). All delegate errors through `createErrorResponse()`. The shared `logs` endpoint is already wired via `[name]/logs/route.ts`.
4. Create 8 API endpoints under `src/app/api/services/{name}/` (`_lib.ts`, `install`, `start`, `stop`, `restart`, `update`, `status`, `auto-start`, `auto-restart-adopted`). All delegate errors through `createErrorResponse()`. The shared `logs` endpoint is already wired via `[name]/logs/route.ts`.
5. Verify `/api/services/` is in `LOCAL_ONLY_API_PREFIXES` in `src/server/authz/routeGuard.ts`; add a test asserting `isLocalOnlyPath()` returns `true` for the new prefix if you add one (hard rule #17).
6. Add a UI tab in `src/app/(dashboard)/dashboard/providers/services/tabs/` reusing `ServiceStatusCard`, `ServiceLifecycleButtons`, `ServiceLogsPanel`.
7. Document in `docs/frameworks/EMBEDDED-SERVICES.md` (update §1 service table + §4 API reference) and `docs/openapi.yaml`.
@@ -399,6 +408,9 @@ Documentation must describe verified behavior, not plausible behavior.
@@ -482,6 +495,12 @@ Why this matters: fixing bug A while opening bug B is worse than not fixing at a
pipeline, and A2A skills.
- Do not close a contributor pull request after using its code; merge it through GitHub so
the contributor receives credit.
- **Never merge a PR that touches an agent-instruction surface without explicit operator
approval** — `CLAUDE.md`, `AGENTS.md`, `GEMINI.md`, `llm.txt` (+ mirrors) and
`skills/**/SKILL.md` are executed as authority by every AI session; a merged instruction
compromises every future agent run. Check with `gh pr diff <N> --name-only` before any
merge. Incident record: PR #11770 (2026-09-01) told agents to execute a third-party
setup script and was swept in by a merge campaign; reverted in #12249.
---
@@ -627,8 +646,8 @@ focused checks, and use a Conventional Commit message (for example, `docs: slim
## Environment
- **Runtime**: Node.js ≥22.0.0 <23 || ≥24.0.0 <27, ES Modules. This is the **only supported** runtime for the published `omniroute` CLI, the server, and the test suites (`node:test` + vitest) — `engines.node` is authoritative and end users never need Bun. A **best-effort `bun:sqlite` compatibility path** exists so a global Bun install (`bun install -g omniroute`) can start without `better-sqlite3` (driver adapter + Bun-aware process spawning); it is **not** a supported runtime — no support guarantees — and every Bun-specific runtime change MUST preserve the Node driver/fallback chain and ship a Bun test (`test:bun:db`) or an explicit reason why the path is Node-only.
- **Bun (build/dev script runner + compatibility smoke only)**: Bun `1.3.14` is pinned as an **exact devDependency** (provisioned through the existing `npm ci` via the lockfile's `@oven/bun-*` platform binaries — no `setup-bun`/ad-hoc install). It is used **only** to execute a small, allow-listed set of TypeScript **gate/generator scripts** (replacing `node --import tsx` for startup speed): the CI checks `check:provider-consistency`, `check:compression-budget`, `check:known-symbols`, and the non-CI `gen:provider-reference`, `bench:compression` — plus the focused `test:bun:db` compatibility smoke suite for the best-effort `bun:sqlite` path. **Do NOT** widen Bun to `npm install`, the build (`build:cli*`), `check:pack-artifact`, the supported published runtime, or the main test runners — those stay on Node. Any new Bun-invoking gate/generator script must be validated byte-identical against its `node --import tsx` output first. After pulling the lockfile change, run `npm install` so `bun` resolves locally (a stale `node_modules` will fail those scripts with `bun: not found`).
- **Runtime**: Node.js ≥22.22.2 <23 || ≥24.0.0 <27, ES Modules. This is the **only supported** runtime for the published `omniroute` CLI, the server, and the test suites (`node:test` + vitest) — `engines.node` is authoritative and end users never need Bun. A **best-effort `bun:sqlite` compatibility path** exists so a global Bun install (`bun install -g omniroute`) can start without `better-sqlite3` (driver adapter + Bun-aware process spawning); it is **not** a supported runtime — no support guarantees — and every Bun-specific runtime change MUST preserve the Node driver/fallback chain and ship a Bun test (`test:bun:db`) or an explicit reason why the path is Node-only.
- **Bun (build/dev script runner + compatibility smoke only)**: Bun `1.4.0` is pinned as an **exact devDependency** (provisioned through the existing `npm ci` via the lockfile's `@oven/bun-*` platform binaries — no `setup-bun`/ad-hoc install). It is used **only** to execute a small, allow-listed set of TypeScript **gate/generator scripts** (replacing `node --import tsx` for startup speed): the CI checks `check:provider-consistency`, `check:compression-budget`, `check:known-symbols`, and the non-CI `gen:provider-reference`, `bench:compression` — plus the focused `test:bun:db` compatibility smoke suite for the best-effort `bun:sqlite` path. **Do NOT** widen Bun to `npm install`, the build (`build:cli*`), `check:pack-artifact`, the supported published runtime, or the main test runners — those stay on Node. Any new Bun-invoking gate/generator script must be validated byte-identical against its `node --import tsx` output first. After pulling the lockfile change, run `npm install` so `bun` resolves locally (a stale `node_modules` will fail those scripts with `bun: not found`).
- **[TS7] fix(types): validate default executor pool config** ([#9849](https://github.com/diegosouzapw/OmniRoute/pull/9849), original [#9795](https://github.com/diegosouzapw/OmniRoute/pull/9795)) — thanks @backryun
- **[TS7] fix(types): preserve The Old LLM proxy contracts** ([#9850](https://github.com/diegosouzapw/OmniRoute/pull/9850), original [#9793](https://github.com/diegosouzapw/OmniRoute/pull/9793)) — thanks @backryun
- **[TS7] fix(types): normalize Gemini Business credentials** ([#9851](https://github.com/diegosouzapw/OmniRoute/pull/9851), original [#9792](https://github.com/diegosouzapw/OmniRoute/pull/9792)) — thanks @backryun
- **[TS7] fix(types): preserve Claude thinking body contracts** ([#9852](https://github.com/diegosouzapw/OmniRoute/pull/9852), original [#9791](https://github.com/diegosouzapw/OmniRoute/pull/9791)) — thanks @backryun
- **fix(response): strip internal reasoning placeholder from all reasoning fields** ([#9853](https://github.com/diegosouzapw/OmniRoute/pull/9853), original [#9790](https://github.com/diegosouzapw/OmniRoute/pull/9790)) — thanks @adevwithpurpose
@@ -888,7 +887,7 @@ _By commits in `ed2db6cb19..v3.8.50`, author identities consolidated via `.mailm
- **fix(combo):** defer the known-context-overflow hard rejection for compressible requests so compression runs before the final context gate, instead of a raw-body estimate 400'ing generic Responses clients targeting a large model before OmniRoute can shrink it ([#10225](https://github.com/diegosouzapw/OmniRoute/issues/10225))
- **fix(api):** deleting a manually-added custom model no longer tombstones a provider-synced model that shares its id. `DELETE /api/provider-models` is addressed by `provider` + `model` alone, so when both a custom row and a synced row existed for one id it removed both and wrote `isDeleted:true`. `replaceSyncedAvailableModelsForConnection` then filtered that id out of every subsequent re-import, so the provider could never resync — model sync kept reporting `added: N` while the catalog stayed empty and `/v1/models` never listed the model again, even though routing to it still worked. The custom row is now removed first and its presence is treated as the operator's intent, leaving the synced sibling importable; a synced-only delete still tombstones as before (#3199, #3782 unaffected) ([#10228](https://github.com/diegosouzapw/OmniRoute/pull/10228)) — thanks @Neuron-Mr-White
- **Audio Bridge:** fix production transcription self-loop uploads so real audio reaches the configured STT provider instead of falling back to an unavailable-provider stub ([#10229](https://github.com/diegosouzapw/OmniRoute/pull/10229)).
- **fix(api):** DeepSeek V4's native `max` reasoning tier is now reachable. DeepSeek accepts `reasoning_effort``low`/`high`/`max` and maps `medium`/`xhigh` down to `high`, while OmniRoute's canonical vocabulary collapses `max` onto `xhigh` — so `{"effort":"max"}` silently resolved to `high` and the catalog never advertised a `max` tier (or its `<model>-max` variant). Following the existing `extendCodexGpt56EffortValues` precedent, the native tier is now preserved for `deepseek`/`ds` V4 models only; the global effort vocabulary is unchanged, routed namespaces (`openrouter/deepseek/…`,`tllm/deepseek_v4`,`oc/deepseek-v4-flash-free`) keep the canonical behavior, and an explicit client `reasoning_effort` still wins ([#10230](https://github.com/diegosouzapw/OmniRoute/pull/10230)) — thanks @Neuron-Mr-White
- **fix(api):** DeepSeek V4's native `max` reasoning tier is now reachable. DeepSeek accepts `reasoning_effort``low`/`high`/`max` and maps `medium`/`xhigh` down to `high`, while OmniRoute's canonical vocabulary collapses `max` onto `xhigh` — so `{"effort":"max"}` silently resolved to `high` and the catalog never advertised a `max` tier (or its `<model>-max` variant). Following the existing `extendCodexGpt56EffortValues` precedent, the native tier is now preserved for `deepseek`/`ds` V4 models only; the global effort vocabulary is unchanged, routed namespaces (`openrouter/deepseek/…`, `oc/deepseek-v4-flash-free`) keep the canonical behavior, and an explicit client `reasoning_effort` still wins ([#10230](https://github.com/diegosouzapw/OmniRoute/pull/10230)) — thanks @Neuron-Mr-White
- **fix(providers):** FreeAIAPIKey now targets `api.freeaiapikey.com`, the host upstream names in its `410 endpoint_moved` response — every request through the provider was failing — and its catalog is resynced to the 10 models the live `/v1/models` actually serves ([#10233](https://github.com/diegosouzapw/OmniRoute/pull/10233))
- **fix(providers):** MonsterAPI's deprecation now actually applies — the flag was written as `isDeprecated`, a key no consumer or schema reads, so the provider kept rendering as healthy in the dashboard, the onboarding wizard and the generated provider reference ([#10234](https://github.com/diegosouzapw/OmniRoute/pull/10234))
- **fix(cliproxy):** read platform/arch at runtime via `os.platform()`/`os.arch()` in `binaryManager` so the embedded installer selects the Windows/ARM assets even when the release bundle is built on a Linux runner (fixes #10244)
@@ -3000,7 +2999,6 @@ _Living section — regenerated 2026-07-19 from all 306 cycle commits (bump 2c62
- Stream model health probes for slow providers ([#7377](https://github.com/diegosouzapw/OmniRoute/pull/7377)) — thanks @JxnLexn
- Add reasoning-based model and effort routing ([#7607](https://github.com/diegosouzapw/OmniRoute/pull/7607)) — thanks @JxnLexn
@@ -3442,7 +3440,7 @@ _Living section — bullets land here as PRs merge into `release/v3.8.47` (paral
- **fix(cli):**`omniroute serve` no longer hangs silently on a readiness timeout ([#6321](https://github.com/diegosouzapw/OmniRoute/issues/6321)) — the child server's stdout was piped to `"ignore"` whenever `--log`/`OMNIROUTE_SHOW_LOG` wasn't set (the default), discarding any debug output, and `runWithSupervisor`'s `waitForServer(...).then((up) => { if (up) {...} })` had no `else` branch, so a boot that never became ready produced zero further output after "⏳ Starting server...". Stdout is now buffered alongside stderr (`ServerSupervisor.getRecentLog()`), and a timeout prints a clear diagnostic plus the buffered output instead of staying silent. Does not by itself explain why boot never completes on a given machine — see the issue for further reproduction. Regression guard: `tests/unit/cli-serve-readiness-timeout-6321.test.ts`.
- **fix(pricing):** Pricing Sync dashboard no longer stuck on "Next Sync: Never" / "Synced Models: 0" ([#6325](https://github.com/diegosouzapw/OmniRoute/issues/6325)) — `pricingSync.ts` kept sync state (`lastSyncTime`, `lastSyncModelCount`) in module-level vars, but the background periodic sync (`instrumentation-node.ts`) and the dashboard status route (`/api/pricing/sync`) each import the module from separate Next.js standalone webpack chunks, giving each its own independent state; `getSyncStatus()` read the (empty) API-route instance's vars. Sync status is now additionally persisted to a new `pricing_sync_status``key_value` namespace and `getSyncStatus()` falls back to it when the local module instance never ran a sync itself. Regression guard: `tests/unit/pricing-sync-cross-instance.test.ts`.
- **fix(api):** stop spuriously 403-ing "Invalid request origin" on `POST /api/providers/health-autopilot/actions` for Docker/LAN dashboard requests ([#6277](https://github.com/diegosouzapw/OmniRoute/issues/6277)) — the route carried a duplicate per-route `validateBrowserMutationOrigin` check re-added by the v3.8.42 release squash after PR #5278 centralized origin enforcement in the authz pipeline; the pipeline strips `PEER_IP_HEADER` before forwarding, so the stale duplicate check could no longer resolve the LAN "direct-local-host" candidate and rejected legitimate same-origin LAN mutations (e.g. clicking "remove cooldown" when accessed via a LAN IP). Removed the duplicate check — origin validation is now solely enforced by the centralized pipeline check, which already handles this case correctly. Regression guard: `tests/unit/serial/provider-health-autopilot.test.ts`.
- **fix(resilience):** a bare, unrecognized `403` from a no-credential (`authType:"none"`) provider like mimocode or theoldllm no longer permanently bans the connection ([#6315](https://github.com/diegosouzapw/OmniRoute/issues/6315), [#6345](https://github.com/diegosouzapw/OmniRoute/issues/6345)) — `classifyProviderError()`'s 403 branch only exempted `apikey` providers from the terminal `FORBIDDEN` classification, so these free/stateless proxies (no real account/credential to revoke) fell through to `FORBIDDEN` on the first unmatched 403 and got `isActive:false, testStatus:"banned"` with no cooldown or retry. The exemption now also covers `authType:"none"` providers, returning `null` (recoverable) so the existing connection-cooldown/retry layer handles it. Regression guard: `tests/unit/errorClassifier-noauth-403-6315.test.ts`.
- **fix(resilience):** a bare, unrecognized `403` from a no-credential (`authType:"none"`) provider like mimocode no longer permanently bans the connection ([#6315](https://github.com/diegosouzapw/OmniRoute/issues/6315), [#6345](https://github.com/diegosouzapw/OmniRoute/issues/6345)) — `classifyProviderError()`'s 403 branch only exempted `apikey` providers from the terminal `FORBIDDEN` classification, so these free/stateless proxies (no real account/credential to revoke) fell through to `FORBIDDEN` on the first unmatched 403 and got `isActive:false, testStatus:"banned"` with no cooldown or retry. The exemption now also covers `authType:"none"` providers, returning `null` (recoverable) so the existing connection-cooldown/retry layer handles it. Regression guard: `tests/unit/errorClassifier-noauth-403-6315.test.ts`.
- **fix(providers):** the **Auggie (Augment CLI)** executor no longer fails on Windows with `spawn EINVAL` ([#6304](https://github.com/diegosouzapw/OmniRoute/issues/6304)) — the global-npm install exposes `auggie` as a `.cmd` shim, which Node's `child_process.spawn` cannot launch on win32 without `shell: true`. Both spawn sites (streaming + the `auggie --version` test) now go through a shared `buildAuggieSpawnOptions()` that sets `shell: process.platform === "win32"`; the argv (built by `buildAuggieArgs()` with a registry-validated `model` and a trailing `--` end-of-options marker) is unchanged, so the argument-injection surface stays closed on non-Windows. Regression guard: `tests/unit/auggie-win32-spawn-6304.test.ts`.
- **fix(api):** the dashboard **"Test model"** action is now a clean connection test ([#6240](https://github.com/diegosouzapw/OmniRoute/issues/6240)) — `modelTestRunner` sent its probe request without an explicit compression override, so whenever the operator's global `compression.enabled` flag was on the test call inherited compression (and any Output-Styles system prompt), polluting the result. The internal test requests now send `X-OmniRoute-Compression: off`, and `chatCore` honors an explicit `off` header even when `compression.enabled` is globally true. Regression guards: `tests/unit/model-test-runner-compression-off-6240.test.ts`, `tests/integration/test-model-compression-off-6240.test.ts`.
- **fix(startup):** an update/restart could crash the whole server at boot with `TypeError: Cannot create property 'message' on string 'Database closed'`, masking the real failure and 500-ing every request until manually restarted ([#6560](https://github.com/diegosouzapw/OmniRoute/issues/6560), plausibly the root cause of #6594's post-upgrade 500) — `driverFactory.ts::preInitSqlJs()` cached its sql.js WASM adapter per file path in a `globalThis`-backed map for idempotency, but never checked whether the cached adapter had since been closed (e.g. by `gracefulShutdown`/`resetDbInstance` racing a reload); reusing that dead handle made the very next query throw sql.js's own bare string `"Database closed"` (not an `Error`) straight out of `instrumentation-node.ts`'s previously-unguarded `ensureDbInitialized()` call, and Next.js's internal `registerInstrumentation()` wrapper unconditionally does `err.message = ...` on whatever `register()` rejects with — assigning `.message` on a primitive string throws in strict mode, so the secondary `TypeError` is what actually crashed the process. Fixed in two parts: `preInitSqlJs()` now evicts a closed cached adapter and creates a fresh one instead of returning it; a new `ensureDbReadyForBoot()` wraps the DB-init call, normalizes any non-Error throw via `normalizeBootError()`, and retries once specifically for a transient "database closed" message (now succeeding against the fresh adapter) before re-throwing anything else as a real `Error`. Regression guard: `tests/unit/instrumentation-database-closed-6560.test.ts`.
@@ -3998,7 +3996,6 @@ Thanks to everyone whose work landed in v3.8.45:
- **feat(api):** expose a read-only provider plugin manifest at `GET /api/v1/provider-plugin-manifest` for sidecar/relay discovery. ([#6001](https://github.com/diegosouzapw/OmniRoute/pull/6001)) (thanks @KooshaPari)
- **feat(sidecar):** advertise the provider manifest URL to Bifrost/CLIProxyAPI via the `X-OmniRoute-Provider-Manifest-Url` header (`OMNIROUTE_PROVIDER_MANIFEST_URL`). ([#6007](https://github.com/diegosouzapw/OmniRoute/pull/6007)) (thanks @KooshaPari)
- **feat(autoCombo):** add a latency/speed-optimized routing mode (shared `rankBySpeed` scoring core) plus the `omniroute_pick_fastest_model` MCP tool. ([#6011](https://github.com/diegosouzapw/OmniRoute/pull/6011)) (thanks @KooshaPari)
- **feat(providers):** refresh The Old LLM (Free) model catalog ([#5181](https://github.com/diegosouzapw/OmniRoute/issues/5181)) — seed the current free `/api/chatgpt` tier (GPT-5/5.1/5.2/5.3/5.4, o3/o4-mini, Gemini 3 Pro / 2.5 Pro / 2.0 Flash / 1.5 Flash, Claude 4.6 Opus/Sonnet & 4.5 Haiku, GPT-4o, Grok 4, DeepSeek V3/R1, Sonar Pro) while keeping the legacy alias IDs for saved-preference compatibility. Also fixes a latent routing bug: `mapModel()` now passes known upstream IDs through unchanged, so Gemini/o-series/Grok/DeepSeek/Sonar models no longer silently collapse onto `GPT_5_4`. Regression guard: `tests/unit/theoldllm-model-refresh-5181.test.ts`. (thanks @WslzGmzs)
- **feat(resilience):** surface Codex **banked reset credits** per connected account ([#5199](https://github.com/diegosouzapw/OmniRoute/issues/5199)) — the Codex quota parsers (`buildCodexUsageQuotas`, `parseCodexUsageResponse`) now additively read `rate_limit_reset_credits.available_count` (+ optional `rate_limit_reached_type`) from the `/wham/usage` payload OmniRoute already fetches, and the provider-limits dashboard renders a **"Banked Reset Credits"** row when a positive count is present. Display-only and **fail-open** — the field is eligibility-gated, so accounts without it are unaffected (parsers never throw on absent/garbage shapes); redemption (an unofficial mutating endpoint) is intentionally out of scope. Regression guard: `tests/unit/codex-banked-reset-credits-5199.test.ts` (8). (thanks @ofekbetzalel)
- **feat(providers):** add sign-up geo-restriction notices for **SenseNova** and **StepFun** ([#5462](https://github.com/diegosouzapw/OmniRoute/issues/5462)) — the provider add-form now warns that SenseNova's console appears to require a Chinese (+86) phone number with no documented international path, and that StepFun's default endpoint is its China platform while a global StepFun Open Platform (`platform.stepfun.ai`, operated by Sparkling AI Pte. Ltd., Singapore) with email/Google/Discord login exists for international users. Informational `notice` only — neither provider is disabled. Regression guard: `tests/unit/regional-provider-cn-notices-5462.test.ts`. (thanks @chirag127)
- **feat(usage):** add on-demand period-scoped usage-data reset (Settings → System Storage) with a purge API and time-window selector. ([#5831](https://github.com/diegosouzapw/OmniRoute/pull/5831))
@@ -5644,7 +5641,6 @@ Thanks to everyone whose work landed in v3.8.43:
- **fix(sse): map `reasoning_effort` to DeepSeek V4's native `{high, max}` vocabulary** — DeepSeek V4 only understands `high`/`max` reasoning levels, so other `reasoning_effort` values are mapped onto its native vocabulary instead of being rejected. ([#4219](https://github.com/diegosouzapw/OmniRoute/pull/4219))
- **fix(glm): default `max_tokens` and an extended timeout for GLM-5.2+ thinking** — GLM-5.2+ thinking responses are slow and need headroom, so OmniRoute now sets a sensible default `max_tokens` and a longer timeout for them. ([#4255](https://github.com/diegosouzapw/OmniRoute/pull/4255) — thanks @dhaern)
- **fix(antigravity): default `includeThoughts` for modern Gemini models** — modern Gemini models on the Antigravity path now default to including thoughts so reasoning isn't silently dropped. ([#4180](https://github.com/diegosouzapw/OmniRoute/pull/4180) — thanks @dhaern)
- **fix(provider-registry): add correct `contextLength` to theoldllm models** — fills in accurate context-window sizes for theoldllm's models. ([#4184](https://github.com/diegosouzapw/OmniRoute/pull/4184) — thanks @herjarsa)
- **fix(models): expose combo model token limits** — `/v1/models` now reports token limits for combo models. ([#4189](https://github.com/diegosouzapw/OmniRoute/pull/4189) — thanks @megamen32)
- **fix(combo): keep the passthrough quota fallback scoped** — prevents the passthrough quota fallback from leaking across unrelated targets. ([#4194](https://github.com/diegosouzapw/OmniRoute/pull/4194) — thanks @Svetznaniy33)
- **fix(combo): opt proactive-fallback compression into the TV1 bail-out (no silent target drop)** — proactive-fallback compression now participates in the TV1 bail-out so a target is never silently dropped. ([#4228](https://github.com/diegosouzapw/OmniRoute/pull/4228))
@@ -6304,7 +6300,6 @@ Thanks to everyone whose work landed in v3.8.43:
- **fix(catalog):** Codex CLI model-catalog refresh no longer errors — `GET /v1/models` now returns a top-level `models: []` array for Codex clients (detected via the `originator` / `user-agent` = `codex_*` headers it sends on `GET /v1/models?client_version=...`), so `codex_models_manager` stops failing to decode the OpenAI-standard response and no longer logs `failed to refresh available models` on every startup. The array is intentionally empty: Codex replaces its built-in per-model agent prompt (`base_instructions`, ~21k chars) with whatever a populated entry carries for the selected model, so emitting our catalog would break Codex's agent behaviour — an empty list keeps Codex on its built-in model info (same inference as before, minus the error). Non-Codex OpenAI clients receive the unchanged `{object,data}` response. ([#3481](https://github.com/diegosouzapw/OmniRoute/pull/3481) — thanks @diegosouzapw)
- **fix(provider):** Cursor's Responses-API-shaped bodies on `/chat/completions` are detected and handled — a body with `input` but no `messages` is now classified as `openai-responses` (instead of forcing `openai` and building from undefined `messages` → upstream 400); standard OpenAI clients are unaffected by the `messages===undefined` guard. ([#3490](https://github.com/diegosouzapw/OmniRoute/pull/3490) — thanks @borodulin)
- **fix(sse):** numeric provider IDs normalized to strings across 4 more surfaces — extends #3427 to the Responses-API SSE passthrough (`response_id`/`item_id`/`call_id`), the buffered/flush path in `stream.ts`, the dedup-key builders, and `sseParser.ts`, preventing `undefined` lookups when IDs arrive as numbers. ([#3451](https://github.com/diegosouzapw/OmniRoute/pull/3451) — thanks @disafronov)
- **fix(theoldllm):**`X-Request-Token` generated server-side, dropping the Playwright dependency — replicates the site's client `rie()` token (djb2 hash + `oldllm-client-2026` seed + UA prefix + 8-hex `crypto.randomUUID` suffix) directly, so The Old LLM no longer needs a headless browser to mint tokens. ([#3491](https://github.com/diegosouzapw/OmniRoute/pull/3491) — thanks @borodulin / @diegosouzapw)
- **fix(combo):** parallel pre-screen + circuit-breaker fast-exit for priority combos — provider profiles and model availability for all targets are pre-screened concurrently (max 5), and targets whose circuit breaker is OPEN are skipped immediately, reducing first-token latency on multi-target priority combos. ([#3169](https://github.com/diegosouzapw/OmniRoute/pull/3169) — thanks @pizzav-xyz)
- **fix(authz):** URL-tokenized client endpoints (`/api/v1/vscode/<key>/...`) authenticate again when the caller sends its own non-OmniRoute `Authorization` header — a non-`Bearer <token>` header (e.g. VS Code Copilot's own, or an empty `Bearer `) no longer short-circuits auth; it falls through to the path-scoped URL token (still validated downstream), instead of 401'ing under `REQUIRE_API_KEY=true`. ([#3504](https://github.com/diegosouzapw/OmniRoute/pull/3504) — thanks @zhiru / @diegosouzapw)
- **fix(playground):** the dashboard provider Test playground works under `REQUIRE_API_KEY=true` — it previously sent the **masked** key (`sk-xxxx****yyyy`) as a bearer (always invalid → 401). It now authenticates via the dashboard session and sends only the key **id** (`x-omniroute-playground-key-id`); the gateway resolves the secret server-side, honored **only** for an authenticated session and never putting the key secret on the wire. ([#3503](https://github.com/diegosouzapw/OmniRoute/pull/3503) — thanks @zhiru / @diegosouzapw)
@@ -6337,7 +6332,7 @@ Thanks to everyone whose work landed in v3.8.43:
- **fix(translator):** Vertex AI tool calls no longer fail with `400 Unknown name "id"` — the OpenAI-style `id` field is stripped from `functionCall`/`functionResponse` parts for `vertex`/`vertex-partner`; the public Gemini API still receives `id` as required for Gemini 3+ signature matching. ([#3457](https://github.com/diegosouzapw/OmniRoute/pull/3457) — thanks @nullbytef0x / @diegosouzapw)
- **fix(claude):** Claude Code `claude-opus-4-8` tool calls no longer break with `tool call could not be parsed` — OmniRoute no longer force-injects `interleaved-thinking` / `advanced-tool-use` / `effort` beta flags the client never negotiated; clients sending their own `anthropic-beta` header control those betas themselves. ([#3458](https://github.com/diegosouzapw/OmniRoute/pull/3458) — thanks @Forcerecon / @diegosouzapw)
- **fix(catalog):** imported/custom models on no-auth providers (e.g. The Old LLM) now appear in `GET /api/v1/models` and the Playground model selector — the eligibility gate required a DB connection row which no-auth providers never have, silently dropping every imported model for them. ([#3463](https://github.com/diegosouzapw/OmniRoute/pull/3463) — thanks @tjengbudi / @diegosouzapw)
- **fix(catalog):** imported/custom models on no-auth providers now appear in `GET /api/v1/models` and the Playground model selector — the eligibility gate required a DB connection row which no-auth providers never have, silently dropping every imported model for them. ([#3463](https://github.com/diegosouzapw/OmniRoute/pull/3463) — thanks @tjengbudi / @diegosouzapw)
- **fix(browser):** optional `cloakbrowser` import no longer causes bundle errors when the package is absent — the import is now wrapped in a dynamic require so the build succeeds on environments that don't install the optional dep. ([#3460](https://github.com/diegosouzapw/OmniRoute/pull/3460) — thanks @rdself)
- **fix(claude-web):** claude-web session handling cleanup — corrects an edge case where session cookies were not properly refreshed after a Turnstile challenge, and removes stale wrapper code left over from the provider split. ([#3449](https://github.com/diegosouzapw/OmniRoute/pull/3449) — thanks @androw)
- **fix(analytics):** SQL named params are now scoped per query context — a shared params object was being mutated across concurrent analytics queries, causing `SQLITE_MISUSE: named parameter not found` errors under load. ([#3447](https://github.com/diegosouzapw/OmniRoute/pull/3447) — thanks @ReqX)
@@ -6513,8 +6508,7 @@ Thanks to everyone whose work landed in v3.8.14:
- **fix(dashboard):** Agent Bridge page (`/dashboard/tools/agent-bridge`) no longer crashes with "Internal Server Error" — the page replaced its well-shaped state with the raw `/api/tools/agent-bridge/state` response (`{ server, agents }`), leaving `serverState` undefined and throwing `Cannot read properties of undefined (reading 'running')`. A shared `normalizeAgentBridgeState()` now maps the route shape into the page contract (incl. `server.certExists → certTrusted`) and always returns safe defaults, used by both the SSR loader and the polling hook. (#3318 — thanks @tycronk20)
- **fix(codex):** strip client-only params (`prompt_cache_retention`, `safety_identifier`, `user`) on the native `codex/``/v1/responses` passthrough — Codex upstream rejects them with `400 Unsupported parameter`, which broke Factory Droid and any client injecting those fields. The chat-completions path already stripped them; the responses→responses passthrough now does too. (#3317 — thanks @tycronk20)
- **fix(theoldllm):** stop the `[502]: Body is unusable: Body has already been read` error on the cached-token path — the executor read the same upstream `Response` body with `.text()` twice; it now reads it once and only re-reads after a token-rejection refetch. (#3296 — thanks @onizukashonan14-png)
- **fix(dashboard):** keep no-auth providers (opencode, duckduckgo-web, theoldllm, veoaifree-web) visible under the "Show configured only" filter — they never create a connection row (`stats.total === 0`) but are always usable and already appear in `/v1/models`, so the filter now treats `displayAuthType === "no-auth"` as configured. (#3290 — thanks @uniQta)
- **fix(dashboard):** keep no-auth providers (opencode, duckduckgo-web, veoaifree-web) visible under the "Show configured only" filter — they never create a connection row (`stats.total === 0`) but are always usable and already appear in `/v1/models`, so the filter now treats `displayAuthType === "no-auth"` as configured. (#3290 — thanks @uniQta)
- **fix(dashboard):** refresh the connection list after a Codex/Claude/Gemini auth import — the import modals called `fetchData()` (which only reloads provider metadata), so a freshly-imported connection stayed invisible until a manual reload; they now call `fetchConnections()`. ([#3320](https://github.com/diegosouzapw/OmniRoute/pull/3320) — thanks @zhiru)
- **fix(cli):**`omniroute update` no longer always fails on a global install — `getCurrentVersion()` and `createBackup()` now resolve `package.json`/`bin` relative to the script (`import.meta.url`) instead of `process.cwd()` (the user's working dir on a global npm/brew install → _"Could not determine current version"_), and the backup copies the `cli` directory with `cpSync({recursive:true})` instead of `copyFileSync`, which threw a swallowed `EISDIR` → _"Failed to create backup. Aborting"_. (#3295 — thanks @uniQta)
- **fix(sse):** harden the passthrough stream against empty upstream responses — emit a synthetic retry chunk on an empty `choices: []` (fixes a Copilot Chat crash) and log empty post-`tool_calls` completions; also registers **MiniMax M3** (1M context) across 8 provider tiers. ([#3297](https://github.com/diegosouzapw/OmniRoute/pull/3297), #3110 — thanks @wilsonicdev)
@@ -6606,7 +6600,6 @@ Thanks to everyone whose work landed in v3.8.12:
### ✨ New Features
- **theoldllm:** add The Old LLM — a free, Playwright-backed provider with dual-mode operation (cached browser token + direct fetch) bridged through a Vercel relay (#3217 — thanks @oyi77)
- **codex:** add Codex login via OpenAI's browser-driven device authorization flow, exposed as a shareable "Adicionar Externo" public link (`/connect/codex/{token}`) so a third party can complete the OpenAI device login without dashboard access (#3195 — thanks @zhiru)
- **proxy:** per-connection proxy distribution — `proxy_enabled` DB schema + Zod-validated resolution backend, automatic proxy-fallback selection when provider validation hits a network error, and a dashboard UI with per-connection toggles and a tag-filtered "Distribute Proxies" button (#3170, #3171, #3172 — thanks @pizzav-xyz)
- **api:**`/v1/images/generations` and `/v1/images/edits` now resolve a bare combo/alias model name (e.g. `image`) to its single image target, and `/v1/images/edits` forwards multipart edits to custom OpenAI-compatible providers' `{base_url}/images/edits` (also accepting JSON/data-URL edit input) instead of rejecting everything but chatgpt-web (#3214, #3215 — thanks @ngocquynh85)
# Smoke check native database driver used by Bun (bun:sqlite)
RUN bun -e "import { Database } from 'bun:sqlite'; const db = new Database(':memory:'); db.query('SELECT 1 AS ok').get(); db.close(); console.log('bun:sqlite smoke: OK');"
@@ -58,7 +56,7 @@ ENV NODE_ENV=production
RUN bun run --quiet build
# ── Runner Base stage (100% Bun Native Production Runtime) ──────────────────
FROMoven/bun:1.3.14-slimASrunner-base
FROMoven/bun:1.4.0-slimASrunner-base
LABEL org.opencontainers.image.title="omniroute"\
org.opencontainers.image.description="Unified AI proxy — route any LLM through one endpoint (Bun Native)"\
<img src="./docs/diagrams/readme-hero.svg" width="100%" alt="OmniRoute — Never stop coding. Every AI tool → 351 providers — 90+ free — through one endpoint. Claude Code, Codex, Cursor, Cline, Copilot & Antigravity into FREE Claude / GPT / Gemini with auto-fallback. RTK + Caveman stacked compression saves 15–95% tokens (~89% avg) — never hit limits. 351 AI providers · 90+ free tiers · ~1.51B free tokens/mo · 19 routing strategies · $0 to start."/>
<img src="./docs/diagrams/readme-hero.svg" width="100%" alt="OmniRoute — Never stop coding. Every AI tool → 355 providers — 150+ free — through one endpoint. Claude Code, Codex, Cursor, Cline, Copilot & Antigravity into FREE Claude / GPT / Gemini with auto-fallback. RTK + Caveman stacked compression saves 15–95% tokens (~89% avg) — never hit limits. 355 AI providers · 150+ free tiers · ~1.51B free tokens/mo · 19 routing strategies · $0 to start."/>
</div>
@@ -17,9 +17,9 @@
</div>
> Stacking free tiers by hand is painful — dozens of SDKs, dozens of rate limits, and no idea how much you actually have. OmniRoute catalogs **445 free-tier entries across 39 recurring pool keys** and computes the token headline from the **20 pools with a published positive monthly budget**, deduplicated by shared pool. The result stays visible on the dashboard (`/dashboard/free-tiers`).
> Stacking free tiers by hand is painful — dozens of SDKs, dozens of rate limits, and no idea how much you actually have. OmniRoute catalogs **446 free-tier entries across 38 recurring pool keys** and computes the token headline from the **20 pools with a published positive monthly budget**, deduplicated by shared pool. The result stays visible on the dashboard (`/dashboard/free-tiers`).
<img src="./docs/diagrams/free-tier-budget.svg" width="100%" alt="OmniRoute free-tier budget card: ~1.51B free tokens per month steady, up to ~2.13B in the first month with signup credits, from 39 documented recurring pool keys covering 445 cataloged free-tier entries behind one endpoint. Honest pool-deduped math — each shared pool counted once, including 20 recurring pools with a published positive monthly token budget; 13 providers are marked avoid in the terms-risk catalog so you decide. Budget bar includes Mistral 1B, LLM7 150M, Nara 150M, Gemini 60M and smaller pools, plus first-month signup credits and permanently-free no-token-cap providers surfaced separately so they never inflate the headline. Live used/remaining on /dashboard/free-tiers."/>
<img src="./docs/diagrams/free-tier-budget.svg" width="100%" alt="OmniRoute free-tier budget card: ~1.51B free tokens per month steady, up to ~2.13B in the first month with signup credits, from 38 documented recurring pool keys covering 446 cataloged free-tier entries behind one endpoint. Honest pool-deduped math — each shared pool counted once, including 20 recurring pools with a published positive monthly token budget; 13 providers are marked avoid in the terms-risk catalog so you decide. Budget bar includes Mistral 1B, LLM7 150M, Nara 150M, Gemini 60M and smaller pools, plus first-month signup credits and permanently-free no-token-cap providers surfaced separately so they never inflate the headline. Live used/remaining on /dashboard/free-tiers."/>
> Animated summary of the live `/dashboard/free-tiers` page. Full methodology (pool dedupe, credit tiers, provider terms): **[docs/reference/FREE_TIERS.md](docs/reference/FREE_TIERS.md)**.
<img src="./docs/diagrams/promise-pillars.svg" width="100%" alt="The Promise — One endpoint and 351 providers. Automatic fallback keeps routing while another healthy target is available. Six pillars: resilient fallback across 351 providers · up to 95% token savings on eligible workloads · $0 to start with 90+ free tiers and 56 recurring/keyless free-forever providers · 35 CLI/agent integrations through one config · OpenAI, Claude, Gemini and Responses API compatibility at /v1 · production controls including circuit breakers, TLS stealth, MCP 110 tools, A2A, memory, guardrails, evals and 39,000+ static test declarations across 5,100+ tracked test files."/>
<img src="./docs/diagrams/promise-pillars.svg" width="100%" alt="The Promise — One endpoint and 355 providers. Automatic fallback keeps routing while another healthy target is available. Six pillars: resilient fallback across 355 providers · up to 95% token savings on eligible workloads · $0 to start with 150+ free tiers and 53 recurring/keyless free-forever providers · 36 CLI/agent integrations through one config · OpenAI, Claude, Gemini and Responses API compatibility at /v1 · production controls including circuit breakers, TLS stealth, MCP 110 tools, A2A, memory, guardrails, evals and 39,000+ static test declarations across 5,100+ tracked test files."/>
<br/>
<br/>
@@ -332,6 +332,8 @@ No combo to create. Set your model to `auto` (or a variant) and OmniRoute builds
<tr><td align="left" nowrap><code>auto/cheap</code></td><td align="left">💰 Cheapest per token first</td></tr>
<tr><td align="left" nowrap><code>auto/offline</code></td><td align="left">🔋 Most quota / rate-limit headroom first</td></tr>
@@ -429,7 +431,7 @@ All **19** strategies — mix & match per combo step:
<tr>
<td align="center">17</td>
<td nowrap><code>auto</code></td>
<td>15-factor live scoring across every connection 🤖</td>
<td>16-factor live scoring across every connection 🤖</td>
</tr>
<tr>
<td align="center">18</td>
@@ -443,13 +445,13 @@ All **19** strategies — mix & match per combo step:
</tr>
</table>
<sub>The Auto-Combo engine scores every candidate on **15 factors** (health, quota, cost, latency, task fit, quality, session availability…) — see [`docs/routing/AUTO-COMBO.md`](docs/routing/AUTO-COMBO.md).</sub>
<sub>The Auto-Combo engine scores every candidate on **16 factors** (health, quota, cost, latency, task fit, quality, session availability…) — see [`docs/routing/AUTO-COMBO.md`](docs/routing/AUTO-COMBO.md).</sub>
##
### 🧱 Resilience is built in (3 independent layers)
<img src="./docs/diagrams/resilience-layers.svg" width="100%" alt="OmniRoute resilience — 3 independent self-healing layers, the right layer for the right failure. Layer 1 provider circuit breaker (whole provider): trips only on 408/5xx, thresholds OAuth 10× / API-key 15× / local 2×, resets 60s/30s/15s into a HALF-OPEN probe, lazy recovery; while OPEN the combo reroutes to the next provider. Layer 2 connection cooldown (one key/account): base 5s OAuth / 3s API-key, exponential ×2 backoff with anti-thundering-herd guard, 429 honors Retry-After, success clears all error state; one cooling key is skipped while sibling keys keep serving. Layer 3 model lockout (one model): per-model 429, local 404 or mode denials lock just that model — never the whole connection. Terminal states (banned, expired, credits exhausted) are for the operator, not cooldowns."/>
<img src="./docs/diagrams/resilience-layers.svg" width="100%" alt="OmniRoute resilience — 3 independent self-healing layers, the right layer for the right failure. Layer 1 provider circuit breaker (whole provider): trips only on 408/5xx, thresholds OAuth 8× / API-key 12× / local 2×, resets 60s/30s/15s into a HALF-OPEN probe, lazy recovery; while OPEN the combo reroutes to the next provider. Layer 2 connection cooldown (one key/account): base 5s OAuth / 3s API-key, exponential ×2 backoff with anti-thundering-herd guard, 429 honors Retry-After, success clears all error state; one cooling key is skipped while sibling keys keep serving. Layer 3 model lockout (one model): per-model 429, local 404 or mode denials lock just that model — never the whole connection. Terminal states (banned, expired, credits exhausted) are for the operator, not cooldowns."/>
@@ -461,7 +463,7 @@ All **19** strategies — mix & match per combo step:
</div>
<img src="./docs/diagrams/comparison-table.svg" width="100%" alt="What sets OmniRoute apart — a dated feature snapshot vs 9router, OpenRouter, CLIProxyAPI and LiteLLM across 13 capabilities. OmniRoute: 351 providers, 90+ free tiers built in, 19 routing strategies, 12-engine token compression, built-in MCP server with 110 tools, A2A agent protocol, persistent memory, guardrails, cloud agents, TLS fingerprint stealth, Desktop/Termux/PWA and 43 i18n UI locales. OmniRoute is MIT-licensed and self-hostable. Competitor capabilities and counts may change; see the linked methodology."/>
<img src="./docs/diagrams/comparison-table.svg" width="100%" alt="What sets OmniRoute apart — a dated feature snapshot vs 9router, OpenRouter, CLIProxyAPI and LiteLLM across 13 capabilities. OmniRoute: 355 providers, 150+ free tiers built in, 19 routing strategies, 12-engine token compression, built-in MCP server with 110 tools, A2A agent protocol, persistent memory, guardrails, cloud agents, TLS fingerprint stealth, Desktop/Termux/PWA and 43 i18n UI locales. OmniRoute is MIT-licensed and self-hostable. Competitor capabilities and counts may change; see the linked methodology."/>
<sub>📊 Full methodology & per-feature detail vs 9router, OpenRouter, CLIProxyAPI & LiteLLM → [`docs/comparison/OMNIROUTE_VS_ALTERNATIVES.md`](docs/comparison/OMNIROUTE_VS_ALTERNATIVES.md)</sub>
@@ -548,7 +550,7 @@ the current catalog at **[radar.omniroute.online/planos](https://radar.omniroute
- **🗜️ Compression hardening** — default-on inflation guard, Caveman packs for DE / FR / JA + Chinese (wényán), RTK filters for Gradle & .NET. → [Compression](docs/compression/COMPRESSION_ENGINES.md)
- **⚖️ Quota-Share routing** — split a shared account's quota fairly across pooled keys, work-conserving so idle slices are lent out. → [Resilience Guide](docs/architecture/RESILIENCE_GUIDE.md)
@@ -559,9 +561,10 @@ the current catalog at **[radar.omniroute.online/planos](https://radar.omniroute
- **🖼️ New endpoints** — `/v1/ocr` (Mistral OCR) and `/v1/audio/translations` (Whisper-style) round out the media surface. → [API Reference](docs/reference/API_REFERENCE.md)
- **🎨 Image / video / audio generation** — one API for media: xAI Grok Imagine & Novita AI video, ComfyUI, Magnific, Adobe Firefly, Segmind, and speech providers such as ElevenLabs. → [API Reference](docs/reference/API_REFERENCE.md)
- **🤝 More providers & agents** — Cursor Cloud Agent, Grok Build (xAI) with browser + OAuth login, Ollama first-class card, Claude Opus 5 & Sonnet 5, Kimi official partnership (Code/Web/Moonshot), Zed, Requesty, SenseNova, Yuanbao, Agnes AI… and a refreshed **351-provider catalog**. → [Providers](docs/reference/PROVIDER_REFERENCE.md)
- **🤝 More providers & agents** — cloud agents (Codex Cloud, Cursor, Devin, Jules), Grok Build (xAI) with browser + OAuth login, Ollama first-class card, Claude Opus 5 & Sonnet 5, Kimi official partnership (Code/Web/Moonshot), Zed, Requesty, SenseNova, Yuanbao, Agnes AI… and a refreshed **352-provider catalog**. → [Providers](docs/reference/PROVIDER_REFERENCE.md)
- **📡 Routing transparency** — every response carries an `X-OmniRoute-Decision` header naming the strategy/provider/latency that served it, a new `cache-optimized` combo strategy + Auto-Combo `cacheAffinity` factor route repeat requests back to the connection holding the cached prefix, and a read-only `/v1/auto-combo/{channel}/candidates` endpoint exposes an `auto/*` channel's live candidate pool. → [Auto-Combo](docs/routing/AUTO-COMBO.md)
- **⚡ Local performance & infra** — one-click local Redis, Cloudflare Workers / Deno Deploy relay deployers, Bifrost & Mux as supervised embedded services. → [Embedded Services](docs/frameworks/EMBEDDED-SERVICES.md)
- **🧩 Also in the box** — plugin framework + marketplace, Omni/Agent/GitHub skills frameworks, Obsidian vault integration (22 MCP tools), OpenAI-compatible Batch & Files APIs, semantic response cache, gamification with leaderboards, ACP agent discovery (15 built-in agents), scheduled log export to BigQuery, `auto/chaos` fault injection, a Telegram bot bridge, an in-app version manager and LMArena-ELO free-provider rankings. → [Docs](docs/README.md)
<br/>
@@ -612,7 +615,7 @@ the current catalog at **[radar.omniroute.online/planos](https://radar.omniroute
<b>＋ also works with</b> · Kiro · Command Code · Antigravity · Windsurf · AMP · <b>any OpenAI-compatible tool</b>
omniroute run gemini --model glm/glm-5.2 -- --skip-trust -p "reply OK"
# Or pick provider+model interactively and write the tool's own config:
omniroute configure codex # also: claude opencode qwen aider goose cline continue kilo
omniroute configure codex # also: claude opencode qwen aider goose gemini cline continue kilo
```
Every command honors the active remote context (`omniroute connect <host>`), `--dry-run`
@@ -642,11 +645,11 @@ of your shell history. → [CLI Integrations](docs/guides/CLI-INTEGRATIONS.md)
<div align="center">
## 🌐 351 AI Providers — 154 Catalog-Marked Free
## 🌐 352 AI Providers — 152 Catalog-Marked Free
</div>
> **351 registered providers** across the canonical chat, media, search, local, cloud-agent and system collections, including **154 carrying `hasFree: true` discovery metadata**. The chat model registry covers **268 providers / 2,566 distinct provider-model pairs / 1,312 raw model IDs**; the separate free-budget catalog has **455 per-model rows**, **40 recurring pools** and **56 recurring/keyless free-forever providers**. These are different denominators by design; definitions and pool-deduped calculations live in the [Provider Reference](docs/reference/PROVIDER_REFERENCE.md) and [Free Tiers](docs/reference/FREE_TIERS.md).
> **352 registered providers** across the canonical chat, media, search, local, cloud-agent and system collections, including **152 carrying `hasFree: true` discovery metadata**. The chat model registry covers **229 providers / 2,554 distinct provider-model pairs / 1,283 raw model IDs**; the separate free-budget catalog has **446 per-model rows**, **38 recurring pools** and **53 recurring/keyless free-forever providers**. These are different denominators by design; definitions and pool-deduped calculations live in the [Provider Reference](docs/reference/PROVIDER_REFERENCE.md) and [Free Tiers](docs/reference/FREE_TIERS.md).
- **🎯 Adaptive context-budget** _(the dial)_ — instead of one on/off token threshold, escalate the cheapest, most-lossless engines only as far as needed to **fit the model's context window**. Policy: `reserve-output` (default, model-aware) · `percentage` · `absolute`. Mode: `floor` (guarantee fit) · `replace-autotrigger` (your explicit choice wins) · `off` (legacy threshold).
- **🎛️ Where compression is decided** _(precedence, high → low)_ — per-request `x-omniroute-compression` header › routing-combo override › active named profile › adaptive / auto-trigger › panel default › off. The applied plan echoes back in the `X-OmniRoute-Compression: <mode>; source=<source>` response header.
Standard `bun install` and global installation (`bun install -g omniroute`) are supported via Bun runtime detection:
- **Built-in `bun:sqlite`**: OmniRoute uses Bun's built-in `bun:sqlite` driver when running under Bun, falling back to `better-sqlite3` on Node.js or `sql.js`.
- **Automatic Webpack bundler selection**: Development (`bun run dev`) and production builds (`bun run build`) automatically detect Bun and disable Turbopack in favor of Webpack to prevent native V8 binding incompatibilities.
- **Automatic Webpack bundler selection in dev**: Development (`bun run dev`) automatically detects Bun and disables Turbopack in favor of Webpack to prevent native V8 binding incompatibilities. Production builds (`bun run build`) follow `OMNIROUTE_USE_TURBOPACK` exactly as on Node: Turbopack by default, `OMNIROUTE_USE_TURBOPACK=0` to build with Webpack (`Dockerfile.bun` exposes it as a `--build-arg`).
- **Dedicated Bun Dockerfile**: Multi-stage `Dockerfile.bun` for native Bun production deployments (`docker build -f Dockerfile.bun -t omniroute:bun .`).
```bash
@@ -1178,9 +1183,10 @@ Métricas canônicas em 2026-08-24: **1.029 vídeos únicos** · **11.132.922 vi
<tr><td nowrap><b>Language</b></td><td>TypeScript 6.0 — <b>100% TypeScript</b> across <code>src/</code> and <code>open-sse/</code> (zero <code>any</code> in core since v2.0)</td></tr>
@@ -1263,9 +1269,9 @@ Métricas canônicas em 2026-08-24: **1.029 vídeos únicos** · **11.132.922 vi
<tr><td nowrap><b><a href="docs/compression/COMPRESSION_RULES_FORMAT.md">Compression Rules Format</a></b></td><td>JSON rule-pack schemas for Caveman and RTK filters</td></tr>
<tr><td nowrap><b><a href="docs/compression/COMPRESSION_LANGUAGE_PACKS.md">Compression Language Packs</a></b></td><td>Language detection and Caveman rule-pack authoring</td></tr>
- feat(api): add an opt-in `modelVisibilityAllowlist`/`modelVisibilityDenylist` settings pair to curate exactly which models `/v1/models` advertises, mirrored into every `auto/*` combo candidate pool so a denied model cannot be routed to via combo selection either (#11481)
- **feat(guardrails):** enforce a bounded, deterministic contract for Video Bridge transcripts — 256 cues, 4096 input code units and 4 KiB UTF-8 per cue, 64 KiB total text, malformed-Unicode rejection, focus-window scoping, cross-source reconciliation with contributing-source metadata, and a structural provenance trust boundary so caller JSON can never self-assert `embedded`/`audio-bridge` provenance ([#11652](https://github.com/diegosouzapw/OmniRoute/issues/11652))
- **feat(video):** orchestrate optional Video Bridge audio extraction and Audio Bridge STT behind a dual opt-in (operator setting AND per-request signal) — a new loopback-only broker `mode=audio` operation shares the frame path's exact process queue, deadline, AbortSignal, and byte budgets to extract a bounded mono 16 kHz PCM WAV from the same already-downloaded video, then reuses the existing Audio Bridge transcription boundary; provider segment timing is preserved when available and marked coarse otherwise, and every failure degrades to a visual-only-safe partial instead of throwing (#11654).
- **test(video):** Add the Video Bridge FU-07/FU-09 promotion-evidence harness (#11656) — a frozen Zod manifest schema covering the 8 required scenario kinds (static scenes, rapid cuts, late facts, fades, blur, small text, close events, visual prompt injection) with a minimum of 3 repetitions per case, deterministic declarative fixture recipes (`videoBridgePromotionFixtures.ts`), a pure medians/p95 metrics aggregator, a pure FU-07/FU-09 promotion-verdict evaluator applying the ticket's exact thresholds (missing token usage always holds), a digest-only persistence layer that never retains raw media or raw model responses, and a versioned per-model promotion allowlist shipped empty with every model defaulting to `hold`. The FU-07/FU-09 promotion verdicts themselves remain HOLD — they require a real evidence run against real models on VPS 192.168.0.15.
- **feat(video bridge):** "embedded" transcript provenance can now be legitimately earned instead of merely asserted — a bounded, allowlisted (`mov_text`/`subrip`/`webvtt`) subtitle probe runs through the loopback-only Video Bridge broker (at most 2 streams, 10s subdeadline bounded by the request deadline, 256 KiB output, 4096-code-unit lines), normalized through a bounded, ReDoS-safe WebVTT parser and Zod-validated end to end. The adapter always resolves to an explicit `success`/`absent`/`transient_failure` outcome — a subtitle failure never breaks the visual description path, and only a fingerprint-verified broker response (never a caller-declared label) can produce embedded cues (#11659).
- **feat(zai):** add GLM-5.3-Flash Coding Plan support (1M context, 128K output, vision, `low|high|max` reasoning) and route `zai` GLM-5.3-family API-key traffic through the OpenAI-compatible Coding Plan endpoint with native thinking defaults ([#11801](https://github.com/diegosouzapw/OmniRoute/pull/11801)) — thanks @Neuron-Mr-White
- **feat(nodejs):** add `5dive` as a `configure` target — `omniroute configure 5dive` / `omniroute setup-5dive` write a 5dive auth profile that points an agent fleet's `claude` seats at OmniRoute, with the root-only write, the loopback-vs-`https` endpoint rule and the per-seat model pin handled explicitly ([#11852](https://github.com/diegosouzapw/OmniRoute/pull/11852))
- **feat(sse):** treat `max` as a first-class reasoning-effort tier and clamp per model family (GLM 5.1+/DeepSeek V4+/Kimi K3+ keep native `max`; o1/MiniMax/Grok/Muse Spark clamp to their upstream ceiling) ([#11875](https://github.com/diegosouzapw/OmniRoute/pull/11875)) — thanks @Chewji9875
- **feat(providers):** the provider plugin manifest now advertises a `usage-fetch` capability for the 40 providers that have a wired usage/quota fetcher, so external dashboards can read it from `GET /api/v1/provider-plugin-manifest` instead of parsing `open-sse/services/usage.ts` after every release. Discovery only — no new fetcher, no quota change, and the Dashboard quota widget stays gated by `USAGE_SUPPORTED_PROVIDERS`. `USAGE_FETCHER_PROVIDERS` moved to a zero-dependency leaf (`open-sse/services/usage/fetcherProviders.ts`) and is re-exported from `services/usage.ts`, keeping the manifest module a light leaf instead of pulling the ~490-module usage dispatcher into the manifest route. ([#11903](https://github.com/diegosouzapw/OmniRoute/pull/11903)) — thanks @maxmad64bis
- **feat(plugins):** `OMNIROUTE_PLUGINS_DIR` sets the directory the runtime plugin scanner reads — and the root the plugin manager installs into — overriding the `HOME`-derived default, so a Docker/K8s deployment can point straight at its bind-mounted plugin tree instead of moving `HOME` just to relocate the scan path. An image that exports no home no longer scans `/tmp/.omniroute/plugins` in silence: the resolved directory is logged once at startup as `scanner.dir_resolved`, naming the input that won. Unset, behaviour is unchanged. Distinct from the CLI-only `OMNIROUTE_PLUGIN_PATH`, which finds `omniroute-cmd-*` command packages and never reached this scanner ([#11906](https://github.com/diegosouzapw/OmniRoute/pull/11906)) — thanks @amaleta
- **feat(leases):** add an explicit owner-authenticated status action that returns only the active lease's privacy-safe configured connection and provider labels, with generation fencing and no credential or internal-id disclosure ([#11910](https://github.com/diegosouzapw/OmniRoute/pull/11910)) — thanks @KaspaPulse
- **feat(providers):** the provider plugin manifest now also advertises a `usage-supported` capability for the 46 providers whose usage API is accepted by the server and Dashboard routes, so integrators can distinguish "the server will serve quota for this provider" from "a fetcher is wired" without reading TypeScript. Discovery only — no fetcher or quota change. `usage-fetch` resolves on id or alias (the usage dispatcher accepts both); `usage-supported` resolves on id alone, matching the runtime guard `USAGE_SUPPORTED_PROVIDERS.includes(providerId)`. `USAGE_SUPPORTED_PROVIDERS` moved to a zero-dependency leaf (`open-sse/services/usage/supportedProviders.ts`) and is re-exported from `providers.ts`, mirroring the `fetcherProviders` leaf from #11903 and keeping the manifest a light module. ([#12214](https://github.com/diegosouzapw/OmniRoute/pull/12214)) — thanks @maxmad64bis
- **feat(rankings):** order Free Provider Rankings by what each provider actually served — `GET /api/free-provider-rankings?sortBy=reliability` and a "Most reliable first" toggle on the page. Providers with too few calls to state a success rate keep their score order below the measured ones; the default order is unchanged ([#12218](https://github.com/diegosouzapw/OmniRoute/pull/12218)).
- **perf(sse):** defer `cloneLogPayload()` in the structured SSE collector until after the `maxEvents`/`maxBytes` cap check, eliminating ~9,800 wasted `structuredClone` calls per streaming response (65–71% faster `push()`). Reducer snapshot isolation restored for OpenAI and Responses summaries ([#12241](https://github.com/diegosouzapw/OmniRoute/pull/12241)) — thanks @PauloHSOliveira
- **feat(auto-combo):** Auto-Combo scoring can now weigh how often a provider/model has actually succeeded. The engine already carried that number on every candidate — 24 hours of usage history behind a ten-sample floor, real-time metrics otherwise — and the scoring function never read it, while the weight table described `stability` as if it did. `reliability` (`1 - failureRate`, with the same field precedence and the same rate-bounding the speed ranking already uses, so a corrupt reading means "nothing observed" rather than "fails every call") is now a declared factor shipping at weight `0`, so routing is unchanged until an operator gives it one, and the `stability` description now matches what that factor computes ([#12317](https://github.com/diegosouzapw/OmniRoute/pull/12317))
- **feat(routing):** With `freeAccessPolicy: "strict"`, the read-only candidate listing (`GET /v1/auto-combo/{channel}/candidates`) no longer hides the candidates the zero-cost guard excludes — the same read-only transparency the resilience filter already honours (#9133). Each candidate now carries `freeAccessExclusion` saying why it would be kept out, and it tells an exhausted allowance apart from a quota reading that never arrived or went stale, which used to look identical from the outside. Routing is unchanged: the listing reports, it never enforces. The separate `excludeTosAvoid` guard still drops its candidates without a reason; that gap is now documented rather than closed ([#12319](https://github.com/diegosouzapw/OmniRoute/pull/12319))
- **feat(radar):** The Radar catalog table now shows two facts it was already receiving from the feed and dropping on the floor: the per-model rate limits (requests and tokens, per minute and per day) in a new column, and a badge when a provider's terms state it may train on the prompts you send. A limit of zero renders as zero rather than "rate-only" — for a ceiling those are opposite facts — and a model with no training statement gets no badge, because an absent statement is not a guarantee ([#12320](https://github.com/diegosouzapw/OmniRoute/pull/12320))
- **feat(radar):** explain Community, single-use, contributor, supporter, recovery, abuse, offers, and privacy rules before either Radar activation action, and remove the superseded fixed-PR grant promise from every UI locale ([#12342](https://github.com/diegosouzapw/OmniRoute/pull/12342))
- **feat(gamification):** the dashboard Profile page now shows the real daily streak — `/api/gamification/level` returns `streak: { current, longest }` (per key with `apiKeyId`, operator-wide maximum otherwise) and the streak card reads it instead of a hard-coded 0 (#2403)
- **feat(gamification):** the dashboard leaderboard now shows each API key's display name under the Name column instead of a truncated key id; `GET /api/gamification/leaderboard` attaches `name` per entry (name only — no key material), while the shared ranking helper and the federation leaderboard stay id-only — thanks @pacocartones
- **feat(gamification):** enforce the documented 1000 XP/min per-API-key anti-cheat rate limit on the XP award path; over-limit awards are logged and skipped instead of persisted, and the sliding window now matches the timestamp format stored in `xp_audit_log` ([#2403](https://github.com/diegosouzapw/OmniRoute/issues/2403))
- **feat(admin):** localize the gamification anomalies page — the loading state, the Status column and the Suspicious badge now come from the `common` catalog (new `common.suspicious` key propagated to every locale) — add it to the Gamification sidebar group as `gamification-admin` (`/dashboard/gamification/admin`), and expose the loading and empty states as polite `role="status"` live regions (#12401 — thanks @pacocartones)
- **feat(catalog):** add `OMNIROUTE_DISABLE_THINKING_LEVEL_VARIANTS` feature flag to optionally filter out thinking level variants from model catalog ([#PR_NUMBER](https://github.com/diegosouzapw/OmniRoute/pull/PR_NUMBER))
- **feat(providers):** add MaxAI as a signed, OpenAI-compatible provider serving its 13 paid chat models (GPT-5.6 / Luna / Thinking, Claude 5 Sonnet, Claude Haiku 4.5, Gemini 3.1 Pro / Flash-Lite, Grok 4.1-fast / 4.5, DeepSeek V3.2 / R1, Llama 3.3 70B) through OmniRoute's `/v1` endpoint, with per-request HMAC-SHA1→SM3→AES request signing, live model + context-window discovery from `/models/get_config`, and prompted tool-calling translated to OpenAI `tool_calls`
- **feat(providers):** MaxAI vision input — image_url content parts are forwarded inline in `message_content` to the 6 vision-capable models (GPT-5.6 / Luna / Thinking, Claude Haiku 4.5, Gemini 3.1 Pro / Flash-Lite)
- **feat(providers):** MaxAI document RAG — inline base64 file/document attachments are uploaded to MaxAI (content-addressed `doc_id`) and attached to the chat via `doc_list`
- **feat(providers):** browserless MaxAI onboarding — email device-pair login (`/api/providers/[id]/login`) and signed access-token refresh, so a connection can be created and kept fresh without a real browser or Google OAuth
- **feat(providers):** per-provider TLS impersonation profile (MaxAI presents a Windows Firefox-150 client fingerprint) so its bot-sensitive endpoints accept OmniRoute traffic
- New `/dashboard/orchestration` page: live unified view of everything running — Cloud Agent, A2A and Conductor as a real-time graph (Agents tab), the combo cascade (Routing tab, reusing the Combo Live Studio) and a state kanban (Overview tab), with a detail drawer (trace, cost, approve/cancel). Read-only over existing APIs — no new backend. Canvas concept credit: PR #11815 design
- **feat(providers):** add a Perplexity Agent API provider (`perplexity-agent` / `pplx-agent`) for Perplexity `/v1/responses`, including the documented Anthropic, OpenAI, Google, xAI, DeepSeek, Z.AI, Moonshot/Kimi, NVIDIA, and Perplexity model IDs plus Anthropic-model `max_output_tokens` compatibility.
- **feat(providers):** add RPD (Requests Per Day) limit to provider rate limit overrides across UI, schemas, DB, and i18n ([#PR_NUMBER](https://github.com/diegosouzapw/OmniRoute/pull/PR_NUMBER))
- **feat(providers): add UC Direct (uncensored.com Developer API), the metered OpenAI-compatible surface.** A standard OpenAI-compatible passthrough (default executor) for uncensored.com's official REST API at `https://api.uncensored.com/api/v1`: `X-api-key` auth (never-expiring `uai_sk_live_` key), `POST /chat/completions` with streaming SSE and native tool-calling, and the full live metered catalog (82 models across 15 providers, discovered from the public `GET /v1/models`). Registered as provider `uc-direct` (alias `ucd`). Complements the un-metered `uc` persona provider — same models, metered credits and a plain API key instead of a subscription session.
- **feat(providers): add UC (uncensored.com), the un-metered subscription "persona" chat as an OpenAI-compatible provider.** A WebSocket web-app port: a durable Clerk credential mints a short-lived session token per connect (browserless — no API key), driving UC's persona socket. Ships the browserless email-code login (request → verify → harvest), the 19 verified persona models (Claude Opus, Gemini, Grok, GLM, Kimi, DeepSeek, MiniMax, incl. the uncensored variants), prompted `<tool>` tool-calling with a per-model code-style dialect + auto-cure retry for guardrailed models, live `<think>`/reasoning split, streaming + non-streaming OpenAI responses, and full quota/auth error surfacing (paywall / message-limit / rate-limit → 429, invalid session → 401 re-login). Full multimodal parity via the persona blob-upload layer: **vision** (image input, 15 vision-capable models), **document RAG** (PDF/doc upload, server-side extraction), **image generation** (22 models), **video generation** (14 models, async signed-url → poll), and **TTS** (streaming MP3). Registered as provider `uc` (alias `ucn`). The metered OpenAI-compatible Developer API is a separate `uc-direct` provider.
- **fix(dashboard):** Keep local and theme-aware provider SVG icons at a definite layout size so Chromium does not collapse them to 0×0 after the v3.8.50 image-rendering change ([#12054](https://github.com/diegosouzapw/OmniRoute/pull/12054)) — thanks @ponkcore
- **fix(kie):** reroute `flux/kontext` off the KIE Market `createTask` flow — it is catalogued with `isMarket: true` but has no Market catalog page, so KIE rejected it with "model name not supported"; it now hits the dedicated `POST /api/v1/flux/kontext/generate` / `GET /api/v1/flux/kontext/record-info` endpoints instead (#11296).
- Fixed the v3.8.50 Costs and Analytics dashboards so flat-rate Claude Code usage can be shown as an explicitly requested token-price estimate without changing default billed-cost semantics.
- Fixed archived usage retention so each request is priced individually instead of pricing a day's summed tokens once, which understated archived cost whenever a day mixed cache-heavy and ordinary requests.
- Fixed the Costs dashboard so it discloses when displayed figures include flat-rate token-price estimates instead of labelling them as billed spend, using the flag the analytics API already returns; the month-end projection and the CSV/JSON exports carry the same marker, and billed-cost mode is unchanged.
- fix(ci): pass `--pass-on-unpruned-suppressions` in `run-eslint-json.mjs` so the CI Lint job no longer fails when a suppression is merely orphaned by a genuine fix, mirroring the identical fix already in `validate-release-green.mjs` (#11600)
- **fix(build):** `prepublish.ts` bundles the ChatGPT Web (Codex) MCP bridge through `runBuildTool()` instead of spawning `npx.cmd` raw, fixing the build crash on Node ≥ 20/Windows where `.cmd` shims cannot be spawned without a shell (EINVAL) ([#11704](https://github.com/diegosouzapw/OmniRoute/issues/11704))
- **fix(packages/browser-pool):** regenerate `package-lock.json` so the `packages/browser-pool` workspace's locked `playwright`/`@types/node` (and transitives) match its `package.json` specs, fixing cache-only/offline installs (`npm ci --offline`, Nix `buildNpmPackage`) that previously failed with `ENOTCACHED` ([#11747](https://github.com/diegosouzapw/OmniRoute/issues/11747)) — thanks @benjaminkitt
- **fix(usage):** quota and usage refresh no longer 409 when an exclusive lease reserves the connection ([#11758](https://github.com/diegosouzapw/OmniRoute/pull/11758)) — thanks @TheDemonTuan
- fix(sse): set X-OmniRoute-Selected-Connection-Id on successful combo dispatches so downstream consumers stop falling back to an empty connection id (#11810)
- **fix(providers):** Antigravity's dynamic mitmAlias table no longer routes `gemini-3.7-flash-{high,medium,low}` to a literal tier-suffixed upstream id just because one connected account's own discovery listed it directly — those display ids always resolve through the safe `gemini-3.7-flash-tiered` static alias, so one account's Google-provisioned access no longer 404s every sibling account of the provider ([#11824](https://github.com/diegosouzapw/OmniRoute/issues/11824), [#11651](https://github.com/diegosouzapw/OmniRoute/issues/11651))
- **fix(config):** Nous Research's `Hermes-4-405B` model now displays as "Hermes 4 405B (Nous Research)" in both the provider registry and the free-model catalog, instead of the mislabelled "Hermes 4 7B" ([#11861](https://github.com/diegosouzapw/OmniRoute/issues/11861)) — thanks @Karan825
- **fix(provider/nous):** inject required user= tag into Nous Research inference requests to resolve upstream 400 "missing tags" error ([#11861](https://github.com/diegosouzapw/OmniRoute/issues/11861)) — thanks @Karan825
- **fix(cli):** `omniroute update --apply` now tells you explicitly whether a running server was detected and, if so, that you must run `omniroute restart` to apply the update — it never restarted anything and previously implied the update was already live once files were installed. The dashboard's npm-mode Update flow (`/api/system/version`) now tries OmniRoute's own PID-file-managed supervisor before falling back to pm2, and reports an honest "restart required" step instead of a silent pm2-only "skipped" that read like a completed update. The server-side latest-version lookup backing the dashboard's update banner also gained `--prefer-online`, closing the same stale-npm-cache class already fixed in the CLI's own copy for #4376 ([#11885](https://github.com/diegosouzapw/OmniRoute/issues/11885)).
- **fix(resilience):** clear persisted LKGP pins when a target suffers connection/provider exhaustion or is skipped before dispatch due to cooldown/exhaustion/unavailability, preventing subsequent requests from repeatedly prioritizing known-dead providers ([#11911](https://github.com/diegosouzapw/OmniRoute/issues/11911)).
- fix(ollama): preserve multi-byte UTF-8 content split across stream chunks in the Ollama NDJSON transform, which previously corrupted CJK/emoji into U+FFFD (#11921)
- **fix(providers):** OrcaRouter chat requests now target `/v1/chat/completions` instead of the bare `/v1` API root, fixing the upstream `404 Invalid URL (POST /v1)` ([#11923](https://github.com/diegosouzapw/OmniRoute/pull/11923)).
- **fix(plugins):** deliver the `onStreamComplete` event to disk-installed plugins. The event shipped in v3.8.50 (#9669) was emitted internally but had no plugin-facing wiring, so no plugin could ever subscribe: the manifest schema silently dropped `hooks.onStreamComplete`, and the loader/manager only knew the seven legacy hooks. `onStreamComplete` is now a declarable manifest hook, wired through the loader and registered by the manager like the other hooks, and its payload carries a `requestId` so consumers can correlate the stream-completion event with the originating request ([#11934](https://github.com/diegosouzapw/OmniRoute/pull/11934)) — thanks @amaleta
- **fix(db):** the Qdrant embedding-model dropdown now lists local/self-hosted providers (Ollama, LM Studio, vLLM, etc.) — an active connection is treated as "configured" when the provider allows an optional API key, not only when it has a real key or OAuth, so a running local embedding provider is no longer hidden from the picker ([#11949](https://github.com/diegosouzapw/OmniRoute/issues/11949))
- **fix(providers):** Vertex AI Anthropic partner-model discovery now calls the Model Garden `v1beta1` publisher list (`/v1beta1/publishers/anthropic/models`, global) and parses its `publisherModels` envelope, so Claude models auto-synced from Vertex populate the active live catalog and route at request time instead of returning `Model '<id>' is not available in the active live catalog` ([#11991](https://github.com/diegosouzapw/OmniRoute/issues/11991)) — thanks @fabioluissilva
- **fix(cli):** support `--base-url` alongside `--endpoint` in `omniroute nodes add`, `update`, and `validate` subcommands to prevent global `--base-url` shadowing issues ([#11999](https://github.com/diegosouzapw/OmniRoute/issues/11999)).
- **fix(providers):** `cloudflare-ai` no longer refuses image content parts for every Workers AI model ([#12002](https://github.com/diegosouzapw/OmniRoute/pull/12002)) — the plain-string `content` requirement behind #2539 is carried by the _model_ schema, not by the `/ai/v1/chat/completions` endpoint (measured: an all-text part array returns 200 on `@cf/mistralai/mistral-small-3.1-24b-instruct`, `@cf/meta/llama-4-scout-17b-16e-instruct` and `@cf/meta/llama-3.3-70b-instruct-fp8-fast`, and 400 on the text-only `@cf/qwen/qwen2.5-coder-32b-instruct`). `transformRequest()` flattened every array and threw on the first non-text part (#6390), so image input was refused for vision-capable Cloudflare models that accept it. All-text arrays are still flattened — the one shape every model accepts — while an array carrying a non-text part is passed through untouched, so the attachment is still never silently dropped. Regression guards: `tests/unit/cloudflare-ai-image-parts-6390.test.ts`.
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
