Retargetado para release/v3.8.51 (release/v3.8.50 está congelada — freeze issue #11439). Resolvido um conflito de merge não-relacionado em src/shared/utils/wsPath.ts (originado de um refactor já mergeado nessa branch depois do fork deste PR; o diff real deste PR — scripts/dev/sync-env.mjs + tests/unit/sync-env.test.ts — ficou intacto) e revalidado: typecheck:core limpo, 13/13 testes focados passando.
Segue o precedente correto do #1622 (STORAGE_ENCRYPTION_KEY) para os dois secrets restantes que a postinstall preenchia por engano, defeituando o mecanismo de ensureSecrets(). Obrigado pela contribuição!
Retargetado para release/v3.8.51 (release/v3.8.50 está congelada — freeze issue #11439). Validado em lote combinado (batch-0824h2) contra o tip de release/v3.8.51: typecheck:core limpo, gates estáticos + migration-numbering OK, 127/127 testes focados passando (8/8 do PR entre migration-163 e radar-feed-cache-generated-at).
Migração limpa (ADD COLUMN nullable, sem backfill necessário), aditiva na API, mantém "unknown" honesto para linhas antigas. Obrigado pela contribuição!
Retargetado para release/v3.8.51 (release/v3.8.50 está congelada — freeze issue #11439). Validado em lote combinado (batch-0824h2, junto de #11435/#11436/#11437) contra o tip de release/v3.8.51: typecheck:core limpo, gates estáticos OK, 127/127 testes focados passando.
Investigação sólida com repro real via container isolado, três causas independentes identificadas e corrigidas com testes de regressão dedicados para cada uma. Obrigado pela contribuição!
Validated on a 4-PR combined board: memory-cache-safe-injection + memory-glm-injection + chatcore-memory-skills-injection 39/39 (node:test), injection.test.ts 23/23 (vitest), typecheck:core clean, check:cycles clean, gates within baseline. Narrows the #3890 mid-array system-splice skip to the exact shape that 400s on Claude Opus 5 (splice point not preceded by a server tool result) instead of disabling the cache-safe optimization for every Claude request — preserves the fast path everywhere it's actually safe. Closes#11290.
Validated on a 2-PR combined board: routeGuard 36/36 (within the 68/68 focused-file total), a2a-task-owner-idor 7/7, a2a-tasks-auth, search-baseurl-ssrf-guard, cli-serve-hostname, spawn-capable-prefixes-client-safe all green, typecheck:core + dashboard-typecheck clean, gates within baseline. Five real High-severity advisories fixed with TDD (each failing-then-passing): settings export/import-json ALWAYS_PROTECTED completion, MITM route LOCAL_ONLY+SPAWN_CAPABLE gating, search baseUrl SSRF/IMDS guard, A2A REST task auth+ownership (previously none at all), and the loud boot exposure warning. GHSA-cjv9 confirmed already closed on this base (verified). Round 3 of the advisory sweep.
Validated on a 3-PR combined board: kiro-connection-identity 8/8 (written failing-first — 3 new cases red on the pristine release/v3.8.50 tip, green with this change), typecheck:core + dashboard-typecheck clean, all static gates within baseline. Root cause is exactly right: a CodeWhisperer profile ARN identifies the profile, not the account, and distinct Builder ID accounts via social login can share one — the ARN is now trusted only alongside a non-contradicting account-level identifier (email/clientId). Closes#10815. Thank you @engenhariaandrereis01-ai!
Validated on the combined 8-PR board + the branch itself: vertex-anthropic-models 4/4 (new — pushed to your branch as a fix-in-place commit per Hard Rule #8, covering the parser's global/project-scoped resource names, malformed-input handling, and the claude-* → targetFormat heuristic on vertex/vertex-partner), 88/88 across the board's focused suites, typecheck:core clean, gates within baseline. Anthropic partner models on Vertex AI now discover dynamically via the Model Garden publisher endpoint and route through the Claude translator even for models outside the static registry. Thank you @maci0!
Validated on the combined 8-PR board: ollama-local-capabilities-routing 3/3, managed-model-import 9/9 (including the integration with the carried Gemini-3.5-Flash cleanup from #11259), 88/88 across the board's focused suites, typecheck:core + dashboard-typecheck clean, gates within baseline. This brings #11088 to the release line — it had squash-merged to main by base error (mine) — AND fixes the two defects the port caught: the global filter drop that leaked image/video models into OpenAI chat selections (now scoped to self-hosted providers) and the unregistered hard-lease credential site. Exemplary port discipline: byte-identical carry + the corrections in a separate reviewable commit + the superpowers docs deliberately left out. main still needs the same two-line fix. Thank you @yourspraveen!
Validated on the resolved merge against the current tip (527da656 + the post-#11281 rebaseline): the single conflict was a comment-only collision in providers/[id]/models/route.ts (kept the tip's #10828-ordering note). Focused suites 125/125 across all 13 touched test files (build-sqlite-stub, cc-compatible, copilot-claude-messages, copilot-gemini-route, executor-github, ghe-copilot, github-copilot-discovery-token, github-copilot-model-discovery, noauth-sibling-7620, provider-header-profiles, provider-models-config, request-log-payloads, upstream-error-passthrough), typecheck:core clean, file-size/changelog-integrity OK. Merged --admin over the inherited 2026-08-23 base-red cluster (#9985) — the reds are proven tip failures (CLI catalog cluster + @testing-library allowlist, being drained by #11280), not from this diff. Note: the rebase means several items the body listed (relay x-relay-path SSRF, /v1/search blocked-providers, #10736 rotation fence, #10903, #10865, #10899, #10916) already landed upstream and are NOT in this delta — the delta is: better-sqlite3 build guard + build heap/worker caps + telemetry-off (#10060 re-derived), credential-echo passthrough refusal + OCR/moderation redaction + call-log key redaction, Copilot CLI 1.0.81-6 wire identity + Claude→/v1/messages name-matched routing + discovery token fix, CC model_not_found 400, compat overrides for no-auth aliases (#7620-pinned). The Copilot wire-identity change is the one to watch in production. Thank you @arminanton — and the ported-author credits in the commit history (@rqzbeh, yidecode, the #10899/#10916 authors) are preserved. Your config-posture finding (REQUIRE_API_KEY default vs 0.0.0.0) is noted for a maintainer decision, as you scoped it.
The structural admission gate (src/shared/middleware/chatBodyAdmission.ts —
the bounded heavyweight lease + healthy-headroom path from #10110/#10437)
returns its 503 chat_admission_busy BEFORE request logging, so a shed left
no trace: no counter, no log line, and the process-wide snapshot was never
consumed by any route. This adds pure observability — admission behavior,
defaults, and thresholds are untouched.
- chatBodyAdmission.ts: in-memory shed history (shedTotal + shedsByReason)
on ChatAdmissionController, recorded at the two capacity-driven give-up
points in acquireHeavyWithin (queue_timeout when the bounded wait
expires, queued_bytes_budget when the heap valve refuses to park). A
client abort mid-wait is deliberately not counted — capacity was never
denied. Each shed also emits exactly one structured pino warn (module
chat-admission) with reason/activeHeavy/waiting/queuedBytes and the HMAC
session fingerprint — never a raw credential. PerConnectionAdmission
Controller.snapshot() now carries the counters plus activeHealthyHeadroom.
- /api/monitoring/health: the structural snapshot is exposed under a new
additive chatAdmission key next to the existing adaptiveAdmission (the
shadow-mode layer) — allowlisted projection in observability.ts, nothing
removed from the current payload.
- Tests: tests/unit/chat-admission-visibility-11244.test.ts (RED→GREEN:
shed counting per reason, abort exclusion, snapshot shape, and the warn
log carrying the fingerprint but never the raw API key) + a chatAdmission
allowlist-projection test in observability-payloads.test.ts.
Co-authored-by: Xiangzhe <bakryun0718@proton.me>
Residuals of #11236 after #10371/#10491 landed on the tip:
- Dist-fold residuals (bugs 2+3): managedBinaryName() (binaryManager),
resolveSpawnArgs() (installers/cliproxy) and the per-OS memory probes in
getProcessInfo() (processManager) still read the process.platform literal,
which the Linux build of the published artifact constant-folds (precedent:
b43a212680 / #10244). Converted to call-time os.platform() reads, matching
the module's documented anti-fold pattern. Test-side process.platform uses
are not bundled and stay.
- Fold guard: new tests/unit/windows-platform-fold-guard-11236.test.ts pins
zero out-of-comment process.platform occurrences in the four artifact
runtime files, with a comment-stripping tokenizer plus mutation self-checks.
- Bug 6 (pid null on Windows): portProbe.resolvePortPid() only probed
lsof/ss/net-tools netstat. Added a netstat -ano probe with a dedicated
LISTENING-row parser (parseWindowsNetstatPid) as the last fallback; Unix
probes unchanged, and the Windows parser never matches Unix rows (LISTEN vs
LISTENING). Also converted the darwin args branch in the same array to
os.platform() (same fold class, same hunk).
- Bug 5 hardening: runOAuthStatus coerces an out-of-contract 200 payload to
an empty list with a sanitized stderr warning instead of crashing on
.filter over a non-array.
TDD: guard test, parser tests and the oauth hardening test all failed RED
before the fix and pass GREEN after; sibling suites (binaryManager,
processManager, portProbePid, cli-oauth-commands, installers,
ServiceSupervisor, version-manager) green. The 6877 spawn-args test's win32
mock moved from defineProperty(process.platform) to
mock.method(os, "platform") to match the new runtime read — assertion
unchanged.
Co-authored-by: Xiangzhe <bakryun0718@proton.me>
The dashboard stores LM Studio connections under the hyphenated provider id
"lm-studio", but the embedding registry keys the provider as "lmstudio"
with no alias. As a result, "lm-studio/<model>" embedding requests failed
with a 400 unknown-provider error, and "lmstudio/<model>" requests always
hit the hardcoded http://localhost:1234/v1/embeddings endpoint, ignoring the
baseUrl of the configured connection.
Mirror the ollama-local pattern from #2824/#9225:
- embeddingRegistry: add "lm-studio" -> "lmstudio" to
EMBEDDING_PROVIDER_ALIASES (registry key unchanged so existing
"lmstudio/<model>" clients keep working).
- embeddings service: extend the optional keyless-connection hydration to
lmstudio; getProviderCredentials("lmstudio") already resolves the
"lm-studio" connection via the provider search pool/alias, and a
selection/rate-limit failure still proceeds without credentials.
- embeddings handler: apply the same baseUrl override + normalization
(strip trailing slashes and /v1, /v1/chat/completions, /v1/embeddings
suffixes, then rebuild <host>/v1/embeddings) to lmstudio, keeping the
static localhost fallback when no connection or empty baseUrl.
TDD: tests/unit/lmstudio-connection-baseurl-11233.test.ts failed on the
alias, override and service-hydration asserts before the fix and passes
after; ollama-local (#2824) and lmstudio registry (#7601) sibling tests
remain green.
Co-authored-by: Xiangzhe <bakryun0718@proton.me>
Validated on the combined 12-PR batch board: focused node:test suites green (incl. provider-rate-limit-overrides-schema + ratelimit-admission-control-6593 isolated 11/11 + the new EditConnectionModal vitest 3/3), typecheck:core clean, all static gates within baseline. Pushed one follow-up commit to your branch with the real Vietnamese translations for the two new keys (repo convention: vi never ships __MISSING__ placeholders). Per-connection maxWaitMs override lands, folding the zai-web exception into the general mechanism. Thank you @maxmad64bis!
Merged after conflict resolution: the 5 conflicting test files were the base-red drains that #11201 already landed on the tip — kept the tip versions; the feature content is untouched. Validated on the combined batch board + this branch: codex-app-server + codex-gpt56-catalog 25/25, typecheck:core clean, docs-counts green (351 providers), provider-consistency 268/351/0. The opt-in codex-app-server transport (JSON-RPC-over-WS, turn/completed-awaited close, Responses SSE bridge) leaves the default codex path untouched. Thank you @arminanton — a 3.4k-line transport with the docs wave and tests to match!
Merged after conflict resolution (validated on the combined batch board + this branch): component integrated with #10489's health-race semantics (stale-check invalidation, 3-state badge, correct useCallback deps — your validation gate and tutorial are preserved on top); the two pre-gate tests updated to validate-search-before-enable (contract propagation, commented); the docs/superpowers planning files were dropped — project rule keeps planning artifacts out of docs/ (they live in the private _tasks/ repo). Base was mistakenly main; retargeted. qdrant-config-card 7/7 + qdrant-routes integration 18/18. Thank you @rafacpti23 — the pre-enable real-search gate closes a real misconfiguration footgun!
Merged after conflict resolution (validated on the combined batch board + this branch): stale drift files (package.json/package-lock/.gitignore) kept at tip; the two modals merged additively — AWS SigV4 fields coexist with the chatgpt-web-codex tunnel fields and the #11156 Enter-handler mirror (vitest 2/2 re-run). file-size gate green with the dated AddApiKeyModal 1173 entry (owner-authorized). aws-polly modal + provider-validation-specialty 127/127. Base was mistakenly main; retargeted to release/v3.8.50. Credentials stay in the protected field; nothing logged. Thank you @rafacpti23!
Validated on the combined batch board over tip 8a42aeeb: static gates clean (changelog, file-size 159 frozen, complexity 2621<=2774, cognitive 1181<=1223, dead-code 408<=416), typecheck:core clean, 107 focused tests green.
Combo precedence preserved when a requested name matches an existing combo or combo/* prefix, and hidden/disabled models are skipped during alias resolution (wildcard + mapped). model-alias-seed-fallback green. Related to #10124. Thank you @SCys!
Validated on the combined batch board + this branch: oauth-400-recovery + quota-connection-recovery 22/22; file-size gate green with the test/route.ts 1215 frozen entry (dated annotation for the +190 rebuilt-probe lines, owner-authorized). Unknown expiry + refresh token now means proactive refresh before probing, a hard 400 gets exactly one refresh+retry with fresh-token assertion, and the recovery tick clears stale error labels — the four dead-for-weeks antigravity connections from the production report would have self-healed. TDD with bug-injection proof on each leg. Thank you @HouMinXi!
Validated on the combined batch board + this branch: perf-a-b-c-d + account-fallback-service 93/93, gates + typecheck clean. Owner approved the full bundle including item A (async proxy-log batching, 1s/100-item flush with an unref'd timer — reviewed the implementation: flush helper exists for shutdown wiring, buffered logs are the accepted tradeoff). B (lazy modals), C (O(1) alias maps), D (pre-compiled regex — this is also the entire content of #11187, being closed as subsumed) ride along. Conflict with the tip was only stale provider-count docs. Thank you @rqzbeh!
Merged after conflict triage: the six base-red repair files (vi.json, opencode.ts JSDoc, context-manager test, the three webhook dispatcher tests, the uncloseai orphan-test rename) were already drained on the tip by today's #11130/#11157/#11160/#11113 — those hunks resolved to the tip shape. What lands is the production-fix set: GLM transport-aware Anthropic headers, Claude Code-compatible model-listing rejection, combo live-test single-probe, zero-cost Auto-Combo interval normalization, recovery-clearing union handling, LLMLingua real-path compare, macOS netstat PID discovery, AI Horde R2 strict public-host validation. Sweep of every touched test file: 243/243 green; typecheck + file-size clean. (guide-settings-route's 4 reds reproduce on the pure tip — pre-existing drift from #11079, not from here.) Thank you @backryun!
Merged after conflict resolution in modelMetadataRegistry.ts: the tip's effortTiers chain (declared efforts → declared tiers → undefined-if-thinking-declared → codex extension) now carries this PR's GLM guard as the final-fallback override — GLM-family models without a provider-declared contract get the authoritative empty tier list instead of generic OpenAI tiers. GLM/ZCode suites 40/40 on the resolved branch. Closes#10962. Thank you @xz-dev!
* feat(api): structured ?format=json for the self-service usage endpoint
GET /api/usage/om-usage already let any key read its own usage — personal
daily/weekly USD limits and the provider quota snapshot — but only as
text/plain, which a UI cannot parse safely. OmniCopilot issue #8 asks exactly
for this surface.
Adds ?format=json, returning the ApiKeyUsageLimitStatus + UsageSnapshot the
text is rendered from. Text and JSON share the same collectors
(collectUsageSnapshots, getApiKeyUsageLimitStatus), so the two can never
disagree about a number. The response is a discriminated union: a key without
allowUsageCommand (403) or an invalid key (401) returns
{ allowed:false, error:{message} }, distinct from allowed:true with empty
sections — the state a panel must render as "nothing learned yet", not a
refusal. Text form unchanged; without ?format the contract is untouched.
The endpoint was previously missing from API_REFERENCE.md; it now has a
section documenting both forms, the allowUsageCommand gate, and the
self-service auth model (caller's own key, not requireManagementAuth).
Regression guards in tests/unit/usage-command-json-format.test.ts (4 tests:
json shape, text default preserved, structured 403, sanitized 401 with no
stack trace). Existing internal-usage-command suite still 12/12.
* chore(changelog): correct the fragment to the real PR number (#11190)
* feat(api): return every connection's snapshot under providers[] in om-usage json
Closes#11191. buildUsageCommandJson picked a single snapshot via selectUsageSnapshot, so a panel could only ever show one provider. The collector already had them all — the single-pick is a presentation choice for a terminal. The JSON form now also returns the full UsageSnapshot[] alongside the selected provider, so a UI can render Codex / Claude / OpenCode side by side. The text form is untouched.
---------
Co-authored-by: Xiangzhe <bakryun0718@proton.me>
* feat(api): structured ?format=json for the self-service usage endpoint
GET /api/usage/om-usage already let any key read its own usage — personal
daily/weekly USD limits and the provider quota snapshot — but only as
text/plain, which a UI cannot parse safely. OmniCopilot issue #8 asks exactly
for this surface.
Adds ?format=json, returning the ApiKeyUsageLimitStatus + UsageSnapshot the
text is rendered from. Text and JSON share the same collectors
(collectUsageSnapshots, getApiKeyUsageLimitStatus), so the two can never
disagree about a number. The response is a discriminated union: a key without
allowUsageCommand (403) or an invalid key (401) returns
{ allowed:false, error:{message} }, distinct from allowed:true with empty
sections — the state a panel must render as "nothing learned yet", not a
refusal. Text form unchanged; without ?format the contract is untouched.
The endpoint was previously missing from API_REFERENCE.md; it now has a
section documenting both forms, the allowUsageCommand gate, and the
self-service auth model (caller's own key, not requireManagementAuth).
Regression guards in tests/unit/usage-command-json-format.test.ts (4 tests:
json shape, text default preserved, structured 403, sanitized 401 with no
stack trace). Existing internal-usage-command suite still 12/12.
* chore(changelog): correct the fragment to the real PR number (#11190)
---------
Co-authored-by: Xiangzhe <bakryun0718@proton.me>
Validated on the combined batch board over tip 92ef3c71: static gates clean (changelog, file-size, complexity 2624<=2774, cognitive 1182<=1223, dead-code 411<=416), typecheck:core clean, focused tests green.
Companion to #11113 (consumer side): tokenrouter joins BUILTIN_PROVIDERS_SYSTEM_MUST_BE_FIRST — memory-system-first-6135 suite green. Live-confirmed 400 class documented in the body. Thank you @ggdayup!
Cherry-picked onto the current tip (authorship preserved), noise stripped. Validated on BOTH runtimes: bun test tests/unit/db-adapters/ 44/44 under the pinned Bun 1.3.14 (native bun:sqlite path), and node --test on the same suite 52 pass / 0 fail / 1 skip (Bun-only adapter skips under Node, as designed). Dockerfile.bun entrypoint now matches the standalone runner shape. Follow-up to #11039. Thank you @rqzbeh!
Cherry-picked the JSDoc commit onto the current tip (authorship preserved), stripping the stale generated-count noise files. Focused: opencode-v2-config-11070 2/2. Comments now match the 128k fallback shipped in #11035/#11054. Thank you @rqzbeh!
Validated on the combined board over tip 80d931ae: quota-redis-store (incl. the KEY_PREFIX derivation test), local-redis-status and rate-limiter-redis-optional green, typecheck:core clean. One pre-merge fix pushed to the branch: docs/reference/ENVIRONMENT.md gained the REDIS_KEY_PREFIX row (env-doc-sync gate requires every .env.example var documented). Board note: the redis tests leave an ioredis retry handle open and hang the runner exit locally — assertions all pass; pre-existing pattern, not from this PR. Thank you @MeRezaRezaei!
Validated on the combined board over tip 80d931ae: kimi suites green (executor-kimi-web, kimi-partner-aff-links, token-health-check-kimi 33 assertions), vitest providerPageHeaderKimiPartnerLink green, typecheck:core clean. Audited the diff: only kimi-web switches to the international www.kimi.ai (Connect-RPC base, website, auth hints); kimi-coding / kimi-coding-apikey affiliate links stay on kimi.com as intended — asserted by the updated tests. Owner approved merge without the VPS smoke. Thank you @MeRezaRezaei!
Cherry-picked the author's cleaned single-commit head onto the current tip. One fix applied pre-merge: the removal migration collided at number 161 with today's config_audit_log migration (#11103) — renumbered to 162_remove_hackclub_provider.sql. Focused tests: remove-hackclub-11118 + provider-metrics-deleted-provider green; check:provider-consistency OK (266/348/0). Provider removal requested by Hack Club maintainers (#11118). Thank you @rqzbeh!
Validated on the combined batch board over release/v3.8.50 tip d91238b7: static gates clean, typecheck:core clean, focused tests green.
Operator-declared per-provider error rules via settings, consulted before the built-ins; the allowlist bypass is correct — declaring a rule is itself the opt-in, and no provider-specific rule is hardcoded. Thank you @maxmad64bis!
Validated on the combined batch board over release/v3.8.50 tip d91238b7: static gates clean, typecheck:core clean, focused tests green.
Config-audit survives restarts with bounded growth (migration 161 + OR IGNORE seed, retention wired into runAutoCleanup); route cabling deliberately out of scope. Thank you @maxmad64bis!
Validated on the combined batch board over release/v3.8.50 tip d91238b7: static gates clean, typecheck:core clean, focused tests green.
Strict schema + {sanitized, rejected} DB boundary — silent validation degradation now answers 400 with the offending keys. Caller audit done: only the providers write path consumes the sanitizers. Thank you @maxmad64bis!
Validated on the combined batch board over release/v3.8.50 tip d91238b7: static gates clean, typecheck:core clean, focused tests green.
Real suggestionCount replaces the conflated link count (deprecated alias kept), dashboard deep-link fixed. Thank you @maxmad64bis!
⭐5 — Fecha #10265: chat do command-code migra do endpoint CLI-only /alpha/generate (version-gated + proxy-blocked para callers externos) para o documentado /provider/v1/chat/completions (OpenAI format). Removido o envelope CLI reverse-engineered (config/memory/taste/skills + headers CLI-impersonation), substituído por passthrough OpenAI plano com normalização de model id vendor-prefixed (#10809), clamp de max_tokens (#5166), sanitização de reasoning_effort. commandCode.ts 1037→171 linhas.
Validado no worktree board sobre tip: typecheck:core limpo; 175/175 testes focados (command-code executor/vision/usage/maxtokens/user-array/validation-specialty/responses-handler/provider-models-scoping); changelog-integrity/file-size/complexity/cognitive todos OK. TDD RED→GREEN documentado.
⭐4 — Suporte de backend nativo Bun + Dockerfile.bun multi-stage + fallback dinâmico de driver SQLite (better-sqlite3 prioritário sob Bun, bun:sqlite fallback; Node preservado) + correção de estabilidade do DAST CI smoke.
Validado a fundo (worktree board sobre tip): bun-support 4/4, typecheck:core limpo, dashboard-typecheck OK (220 dentro do baseline), open-sse-typecheck OK (5 pré-existentes), gate de runtime OK sob Node, changelog-integrity OK, file-size/complexity/cognitive/dead-code OK. Verificado que o driver preserva a cadeia Node/falback conforme AGENTS.md; teste bun-support presente. Baselines de typecheck removidos são ratchet honesto (erros não existem mais).
OBS: destravei 2 base-reds do tip neste turno (push direto 7ffa3ef): movi o changelog fragment da #11050 da seção inválida breaking/ para fixes/, e rebaselinei AddApiKeyModal 1067->1073 (crescimento da #11056). Sem isso a #11039 e o resto da fila ficariam vermelhos.
⭐5 — LIMIT negativo = "sem limite" no SQLite: ?limit=-1 retornava a leaderboard inteira em endpoint management autenticado; ceil por Math.min só no upper. Duas camadas: route rejeita não-inteiro/fora de range com 400 (mesmo contrato de parseListLimit), getTopN clampeia como backstop defense-in-depth + exporta LEADERBOARD_MAX_LIMIT. TDD red→green, 5 casos novos, 84/84 suíte gamification. Fecha #11058.
⭐5 — freeAccessPolicy "strict" opt-in (default off): verifica candidatos de auto-combo contra estado de quota ao vivo + segurança econômica por conexão antes do dispatch (fail-closed — estado desconhecido/stale/incompleto é excluído). Zero mudança de comportamento com o default "off". 82 testes focados, eslint/prettier/typecheck limpos, docs em docs/routing/STRICT_ZERO_COST.md.
⭐5 — Variantes de modelo com sufixo de effort (ex: model-max/model-xhigh) são sintetizadas no catálogo a partir de supportedThinkingEfforts do modelo base, sem linha própria no registry — getResolvedModelCapabilities resolvia NULL para tool/vision/context, o que fazia combos com tools reordenar silenciosamente essas variantes para trás. Fix: quando todas as fontes diretas falham, resolve as capacidades a partir do modelo base após strip do sufixo. TDD, 65 testes focados verdes.
⭐5 — Provider x-search de primeira classe (SuperGrok/xAI x_search) em POST /v1/search e MCP omniroute_x_search. Fallback de credenciais xai-oauth→xao→xai; distinto de web search e do X Developer MCP.
Reconciliado com o release tip (que já incluía #10981 "skip catalog-default SearXNG" deste mesmo lote): merge trouxe 5 conflitos reais de contagem gerada (llm.txt/README.md/AGENTS.md/PROVIDER_REFERENCE.md/SVGs/46 mirrors i18n, todos verificados como bump puro 347→348, sem perda de conteúdo do HEAD) + 1 conflito real de mergeable=CONFLICTING.
Durante a validação, os 3 testes novos de SearXNG expuseram um bug real de interação com #10981: `isUnconfiguredLoopbackSearchProvider()` checava o baseUrl ESTÁTICO do catálogo em vez do baseUrl efetivo (após override de `provider_options.baseUrl` ou `providerSpecificData.baseUrl` da conexão), então QUALQUER request a searxng-search — mesmo com override customizado — era rejeitado como se fosse o default não-configurado. Corrigido em `open-sse/handlers/search.ts` (resolve o baseUrl efetivo via `resolveSearchBaseUrl()` antes do skip-check, tanto para o provider primário quanto o alternate). Um teste do próprio #10988 que assumia o comportamento pré-#10981 (default localhost:8888 sempre atendido) foi atualizado para refletir o comportamento já mesclado e intencional (503 quando não configurado).
Validação completa: typecheck limpo, 70/70 testes unit (search-route/search-registry/x-search-provider/searxng-loopback-default), 24/24 vitest MCP, 14/14 integration (search-providers-catalog), lint limpo nos arquivos tocados, docs-counts-sync OK (2 drifts soft pré-existentes, não relacionados), gates estáticos (file-size/complexity/cognitive/dead-code/changelog) todos OK.
⭐5 — getProviderCredentials() descartava conexões isActive:true com testStatus expired/banned/credits_exhausted e retornava null, virando 404 model_not_found mesmo com o modelo listado em /models. Agora retorna o sentinel allExpired quando toda conexão ativa é terminal, resultando em 401 com hint de reconexão. Testes de auth-terminal-status/sse-auth/chat-helpers verdes. Fecha #10969, follow-up de #7611.
