fix(ci): drain the release/v3.8.51 base-reds — auggie/projectCombo typecheck + three stale catalog guards (#14496) by diegosouzapw · Pull Request #14511 · diegosouzapw/OmniRoute

diegosouzapw · 2026-09-22T13:09:55Z

Drena a metade código + guards da base-red de hoje (#14496). A contagem de migrations (AGENTS.md/README.md/65 espelhos de llm.txt) está com outra sessão — recorte combinado no comentário da issue, e nenhum arquivo é compartilhado entre as duas.

Dois defeitos de compilação — `API Route Typecheck` estava a 4 erros fora do baseline

auggie.ts (8 diagnósticos, uma causa). 10c8aaae (hoje) tipou o helper como buildAuggieSpawnOptions<S extends readonly string[]>. Nenhum overload de spawn() aceita readonly string[] como stdio → TS2769 nos 4 call-sites. E como o overload não resolve, o ChildProcess retornado perde a tipagem dos streams → os 4 TS18047 de child.stdout/child.stdin "possibly null" são consequência, não causa. Restringi o genérico a StdioOptions: a tupla literal continua preservada e o spawn() resolve. Nenhum ! e nenhum guard defensivo foi necessário.

projectCombo.ts. Importava ComboLike, ComboCollectionLike e ResolvedComboTarget de comboStructure.ts, que apenas os consome; os três são declarados e exportados no módulo de tipos do combo (TS2459 ×2, TS2724 ×1).

Verificado com o tsconfig do próprio gate (tsconfig.typecheck-api.json) na .113 ociosa: 0 ocorrências dos dois arquivos depois da correção.

Três guards que ficaram para trás de mudanças deliberadas

Guard	Mudança que o deixou para trás	Ajuste
`combo-builder-options-route`	#14143/#14135 (`f4d7d8a5`) fez `qualifiedModel` carregar o prefixo roteável do provider node em vez do id interno — `parseModel()` só resolve o alias	passa a afirmar contra o `prefix` do próprio node (`${provider.prefix}/gpt-custom`), então os dois não podem divergir de novo
`provider-node-reserved-prefix`	#14289 (`257f0a0d`) aposenta `gemini-cli`, e um id retirado conta na mesma varredura (414 → 415)	número atualizado com a procedência anotada, no padrão que o arquivo já mantém para cada mudança anterior
`translate-path` golden	#14316 (`5f67e11c`) tirou o DeepSeek da Responses API (o protocolo exigia eco de `reasoning_text` e quebrava tool-call multi-turno com 400)	snapshot regenerado; o diff é exatamente `openai-responses → openai` e as duas URLs

Nenhuma asserção foi removida ou enfraquecida — as duas de contagem viraram asserções por invariante ou ganharam a justificativa rastreável.

O que esta PR não cobre

Os demais testes da #14496 seguem vermelhos no tip e estão sendo diagnosticados um a um: chatcore-translation-paths (4 — mesma origem #14316), model-token-limit-catalog (4), chat-route-coverage (2), image-model-not-in-chat-catalog-6457 (2), domain-branch-hardening (1), opencode-executor (1), qwen38-max-bare-id-alias, specialty-model-catalog-routes. Entram em PR seguinte para esta não ficar parada — ela já devolve o API Route Typecheck ao verde, que é o gate bloqueando #14491 (fix de segurança #14486) e #14340 (contribuidor).

Refs #14496.

⚠️ base-red inherited: #14496 — o único vermelho é Docs Gates reclamando da contagem de migrations (178→179) em README.md/AGENTS.md/llm.txt, que é o recorte da outra sessão (worktree fix/release-v3.8.51-basereds-0922). Os outros 11 checks estão verdes.

…pecheck, prefix and catalog guards - auggie.ts: buildAuggieSpawnOptions constrained its stdio generic to readonly string[], which satisfies no spawn() overload (TS2769 x4); with the overload unresolved the ChildProcess lost its stream types, producing the TS18047 'possibly null' pile. Constrained to StdioOptions instead — the literal tuple is preserved and spawn() resolves. - projectCombo.ts: ComboLike/ComboCollectionLike/ResolvedComboTarget are declared in the combo type module; comboStructure.ts only consumes them (TS2459/TS2724). - combo-builder-options-route: #14143/#14135 made qualifiedModel carry the provider node's routing prefix instead of the internal id — the guard now asserts against the node's own prefix so the pair cannot drift again. - provider-node-reserved-prefix: #14289 retires gemini-cli, adding one member to the reserved walk (414 -> 415), logged like every prior change. - translate-path golden: regenerated for #14316, which moved DeepSeek from the Responses API to Chat Completions by default. Refs #14496.

…e-v3.8.51-basereds-0922-tests

…strip #14252 added stripInternalBodyFields() at the shared pre-executor boundary so _omniroute* routing markers cannot leak upstream. The same helper also removes the four _native*Passthrough markers — but those are read INSIDE the executor (codex.ts:1259, xai.ts:130), which deletes them itself. Stripping them a layer early turned every Responses-native request into a translated one, which then lost client fields to the #2608 allowlist; 'metadata' is how it surfaced. Executor-consumed markers are now a named list the boundary keeps and the serialization strip (applyFingerprint) still removes, so the leak fix stands. Bisected to 1fb7c9d on a clean checkout. Also aligns opencode-executor's auth expectation with #14230, which routed Zen GPT-5.6 through /v1/responses where #12633's x-api-key rule applies. Refs #14496.

…ta threshold guard - chatcore-translation-paths: #14316 moved DeepSeek's default to Chat Completions; the three Responses cases now pick that path per-connection (apiType: responses), the way an operator still can, so the Chat->Responses reasoning-replay coverage stays on the provider it was written for. - domain-branch-hardening: #14276 made a window whose fraction upstream never reported UNKNOWN rather than '0% remaining' (total: 0 is exactly that), so it must not reach the exhaustion threshold. Added the counterpart case — a reported 0% still does — so the safety change cannot silently disable exhaustion detection. Refs #14496.

… contracts - domain-branch-hardening: #14276 made a window whose fraction upstream never reported UNKNOWN rather than '0% remaining' — total:0 and an Infinity percentage are both exactly that, so neither reaches the exhaustion threshold. Added the counterpart case (a reported 0% still does) so the safety change cannot silently disable exhaustion detection. - qwen38-max-bare-id-alias: #14181 stopped aliasing the GA qwen3.8-max to the preview — opencode-go serves it and the upstream rejects the preview id — so the GA model owns its spec row. The guard now asserts what still matters: neither id may fall back to the 128k window. Refs #14496.

…e-v3.8.51-basereds-0922-tests

…ker tap.testFiles Refs #14547.

…d by #14511 The translate-path snapshot and chatcore-translation-paths DeepSeek fixtures are already reconciled by #14511 (apiType: responses). Keep only the chat-route-coverage changes for the #14316 Chat default (task-aware routing + reasoning-drop).

…w the Codex image catalog ids - chatcore-translation-paths / chat-route-coverage: #14316 moved DeepSeek to Chat Completions by default and kept Responses as a registry alternate format. The five cases that assert a Responses upstream now select it the way the connection modal does — providerSpecificData.targetFormat — on every DeepSeek call (the replay cases' second turn included). The previous apiType attempt never worked: apiType only drives the openai-compatible-* family. Coverage stays on DeepSeek's reasoning_content replay. - model-token-limit-catalog / specialty-model-catalog-routes / image-model-not-in-chat-catalog-6457: #14216 gave Codex GPT-5.6 image rows their own catalog ids (gpt-5.6-sol-image), so /v1/models no longer lists one id twice with two types. parseImageModel() maps both the new and the bare id back to gpt-5.6-sol, so image callers of either keep working; the guards now assert each entry under its own id, and #6457's "neither entry is dropped". Refs #14547.

Merge-batch 2026-09-23 (PRs do mantenedor, 99 triadas por ⭐; esta entrou pela Trilha A). Validação única na árvore combinada das 49 PRs da trilha sobre `release/v3.8.51` @18bbb101: - `typecheck:core`: só o erro herdado `cliproxyAccountHealth.ts:157` (TS2322, não desta leva) - `check:open-sse-typecheck`: 0 erros (o auggie.ts da base é corrigido pela #14511 do mesmo lote) - eslint nos arquivos alterados: limpo - 89 arquivos de teste node focados (710 testes) + 11 suítes vitest de UI: todo vermelho restante reproduz também no tip puro (base-red #14547/#14496) e a árvore combinada fica com um subconjunto estrito dos vermelhos do tip, ou seja, nenhum introduzido pelo lote - teto de file-size acertado na PR de acompanhamento da leva

) Merge-batch 2026-09-23 (PRs do mantenedor, 99 triadas por ⭐; esta entrou pela Trilha A). Validação única na árvore combinada das 49 PRs da trilha sobre `release/v3.8.51` @18bbb101: - `typecheck:core`: só o erro herdado `cliproxyAccountHealth.ts:157` (TS2322, não desta leva) - `check:open-sse-typecheck`: 0 erros (o auggie.ts da base é corrigido pela #14511 do mesmo lote) - eslint nos arquivos alterados: limpo - 89 arquivos de teste node focados (710 testes) + 11 suítes vitest de UI: todo vermelho restante reproduz também no tip puro (base-red #14547/#14496) e a árvore combinada fica com um subconjunto estrito dos vermelhos do tip, ou seja, nenhum introduzido pelo lote - teto de file-size acertado na PR de acompanhamento da leva

…#14410) Merge-batch 2026-09-23 (PRs do mantenedor, 99 triadas por ⭐; esta entrou pela Trilha A). Validação única na árvore combinada das 49 PRs da trilha sobre `release/v3.8.51` @18bbb101: - `typecheck:core`: só o erro herdado `cliproxyAccountHealth.ts:157` (TS2322, não desta leva) - `check:open-sse-typecheck`: 0 erros (o auggie.ts da base é corrigido pela #14511 do mesmo lote) - eslint nos arquivos alterados: limpo - 89 arquivos de teste node focados (710 testes) + 11 suítes vitest de UI: todo vermelho restante reproduz também no tip puro (base-red #14547/#14496) e a árvore combinada fica com um subconjunto estrito dos vermelhos do tip, ou seja, nenhum introduzido pelo lote - teto de file-size acertado na PR de acompanhamento da leva

Merge-batch 2026-09-23 (PRs do mantenedor, 99 triadas por ⭐; esta entrou pela Trilha A). Validação única na árvore combinada das 49 PRs da trilha sobre `release/v3.8.51` @18bbb101: - `typecheck:core`: só o erro herdado `cliproxyAccountHealth.ts:157` (TS2322, não desta leva) - `check:open-sse-typecheck`: 0 erros (o auggie.ts da base é corrigido pela #14511 do mesmo lote) - eslint nos arquivos alterados: limpo - 89 arquivos de teste node focados (710 testes) + 11 suítes vitest de UI: todo vermelho restante reproduz também no tip puro (base-red #14547/#14496) e a árvore combinada fica com um subconjunto estrito dos vermelhos do tip, ou seja, nenhum introduzido pelo lote - teto de file-size acertado na PR de acompanhamento da leva

…14423) Merge-batch 2026-09-23 (PRs do mantenedor, 99 triadas por ⭐; esta entrou pela Trilha A). Validação única na árvore combinada das 49 PRs da trilha sobre `release/v3.8.51` @18bbb101: - `typecheck:core`: só o erro herdado `cliproxyAccountHealth.ts:157` (TS2322, não desta leva) - `check:open-sse-typecheck`: 0 erros (o auggie.ts da base é corrigido pela #14511 do mesmo lote) - eslint nos arquivos alterados: limpo - 89 arquivos de teste node focados (710 testes) + 11 suítes vitest de UI: todo vermelho restante reproduz também no tip puro (base-red #14547/#14496) e a árvore combinada fica com um subconjunto estrito dos vermelhos do tip, ou seja, nenhum introduzido pelo lote - teto de file-size acertado na PR de acompanhamento da leva

…14429) Merge-batch 2026-09-23 (PRs do mantenedor, 99 triadas por ⭐; esta entrou pela Trilha A). Validação única na árvore combinada das 49 PRs da trilha sobre `release/v3.8.51` @18bbb101: - `typecheck:core`: só o erro herdado `cliproxyAccountHealth.ts:157` (TS2322, não desta leva) - `check:open-sse-typecheck`: 0 erros (o auggie.ts da base é corrigido pela #14511 do mesmo lote) - eslint nos arquivos alterados: limpo - 89 arquivos de teste node focados (710 testes) + 11 suítes vitest de UI: todo vermelho restante reproduz também no tip puro (base-red #14547/#14496) e a árvore combinada fica com um subconjunto estrito dos vermelhos do tip, ou seja, nenhum introduzido pelo lote - teto de file-size acertado na PR de acompanhamento da leva

Merge-batch 2026-09-23 (PRs do mantenedor, 99 triadas por ⭐; esta entrou pela Trilha A). Validação única na árvore combinada das 49 PRs da trilha sobre `release/v3.8.51` @18bbb101: - `typecheck:core`: só o erro herdado `cliproxyAccountHealth.ts:157` (TS2322, não desta leva) - `check:open-sse-typecheck`: 0 erros (o auggie.ts da base é corrigido pela #14511 do mesmo lote) - eslint nos arquivos alterados: limpo - 89 arquivos de teste node focados (710 testes) + 11 suítes vitest de UI: todo vermelho restante reproduz também no tip puro (base-red #14547/#14496) e a árvore combinada fica com um subconjunto estrito dos vermelhos do tip, ou seja, nenhum introduzido pelo lote - teto de file-size acertado na PR de acompanhamento da leva

Merge-batch 2026-09-23 (PRs do mantenedor, Trilha B, retrabalhada). Um worker corrigiu o defeito apontado na revisão direto no branch da PR; o detalhe está na seção "Rework (merge-batch 2026-09-23)" do corpo, com a prova de red→green. Revalidado numa árvore combinada das PRs da 2ª leva da Trilha B (#14415 #14472 #14468 #14493 #14403) sobre o tip atual de `release/v3.8.51`: - 104/104 testes node focados + 46/46 vitest (as 6 suítes de UI que saíram da quarentena); `check:provider-consistency` OK (359) - `typecheck:core`: só o herdado `cliproxyAccountHealth.ts:157`; `check:open-sse-typecheck`: só o herdado `auggie.ts` (base-red #14547, corrigido pela #14511) - eslint nos arquivos alterados limpo; file-size OK

…s lossy policy; drain 12 base-red files #14529 moved lossy engines off the header-less path in the runtime but not in deriveEffectivePreviewPlan(), so the dashboard preview showed rtk -> caveman while requests ran session-dedup -> lite (#12063 again). The preview now applies downgradeUnrequestedLossy(); the #12063 test pins preview == runtime for the profile, engines-map and safe-profile cases (fails on the old code). Test guards realigned to merged design changes, each traced to its commit: - #14529: compression suites opt in with allow-lossy where they test the operator plan, and assert the header-less downgrade; default seed is session-dedup + lite; the #10503 probe ignores the loopback dashboard telemetry POST that stacked compression emits (the context gate still rejects locally — verified with logging). - #14530: breaker profile asserted on the connection breaker; the provider-wide cooldown case drives the network-error path. - #14370: @opencode/plugin (stable OpenCode 2.x contract, npm maintainer thdxr) added to the dependency allowlist. - #14213: the opt-in empty-turn retry credential site is inventoried as class A (dispatch goes through assertManagedLeaseFence). - #14223 / #14329: flag count 77, quota registry 13 (muse-code). Refs #14496

…e-v3.8.51-basereds-0922-tests # Conflicts: # tests/unit/chat-route-coverage.test.ts

…n only the management key is set #14544 built the probe URL from CLIPROXYAPI_HOST with no fallback, so a key-only (local) setup probed http://undefined:8317 and showed the account as unreachable; it was also the typecheck:core TS2322 on the release tip. Test fails on the old code (actual http://undefined:8317/...). Refs #14496

Refs #14496

…irement, extracted failure projection - docs: migration count 182 -> 183 in README, AGENTS.md and llm.txt (+ the 66 llm.txt mirrors), same shape as #14559; migration 185 landed with #14544. - #14468 retired suno: APIKEY_PROVIDERS 242 -> 241 and the translate-path golden loses the suno entry (regenerated; that is the only diff). - #14527 moved the retained provider-failure projection into chatCore/providerFailureRetention.ts; the public-error guard now follows the call and still requires sanitizeErrorMessage() at the projection. Refs #14496

…e-v3.8.51-basereds-0922-tests # Conflicts: # tests/unit/provider-node-reserved-prefix.test.ts

…ched #14493 restored every quarantined component suite, leaving excluded: [] — the target state — but the inventory test still demanded a non-empty list. An exclusion added without an inventory entry stays caught by the live-config case. Refs #14496

…temp dir Three cases wrote fixed /tmp/omniroute-subtitle-runtime-*.vtt paths. On the shared self-hosted runners a leftover owned by another user makes that path unwritable (EACCES under /tmp's sticky bit, reproduced on .113 even as root), and parallel shards race on the same name. Refs #14496

…en breaker comments Keeps the #14316 realignment from growing chatcore-translation-paths by 29 lines (now +12) and trims the #14530 comments in account-fallback-service. Refs #14496

…s loader; rebaseline two test files - responses-handler: the synthesized in_progress frame now carries sequence_number 1 and a response object (#14572/#14330) — pin that prefix. - perf-waterfall A1: Home batches loadHomeSettings() (#14421/#14060) with getMachineId(); assert the loader still reads getSettings() by default. - file-size: documented rebaseline for chatcore-translation-paths (+12) and account-fallback-service (+14, 11 of them Prettier reflow of tip lines). Refs #14496

…e-v3.8.51-basereds-0922-tests # Conflicts: # config/quality/file-size-baseline.json # docs/i18n/fr/README.md # docs/i18n/km/README.md # docs/i18n/my/README.md # docs/i18n/or/README.md # docs/i18n/pa/README.md # docs/i18n/vi/README.md

… ceilings and stryker coverage The release tip crossed eight frozen file-size ceilings and one new test file (#14551) passed the 1200 test cap; four new covering tests were missing from stryker tap.testFiles. Documented rebaseline entries carry the measured sizes and the PRs that grew each file; nothing in this commit changes behavior. Refs #14496

diegosouzapw mentioned this pull request Sep 22, 2026

🔴 base-red: release/v3.8.51 voltou a vermelho no mesmo dia — 12 testes + api-typecheck + docs-counts (leva de providers de 09-22) #14496

Closed

diegosouzapw added 5 commits September 22, 2026 10:31

Merge remote-tracking branch 'origin/release/v3.8.51' into fix/releas…

dedf1df

…e-v3.8.51-basereds-0922-tests

Merge remote-tracking branch 'origin/release/v3.8.51' into fix/releas…

a968955

…e-v3.8.51-basereds-0922-tests

This was referenced Sep 23, 2026

test: realign golden + count pins left stale by the merge wave #14693

Closed

fix(compression): make the settings preview match what a request actually runs #14700

Merged

diegosouzapw mentioned this pull request Sep 23, 2026

fix(security): create the data dir 0700 and .env 0600, and repair existing installs (GHSA-2pg2-xm9r-8544) #14712

Merged

diegosouzapw added 2 commits September 23, 2026 21:05

Merge remote-tracking branch 'origin/release/v3.8.51' into fix/releas…

62901a1

…e-v3.8.51-basereds-0922-tests

chore(quality): register nine covering tests the tip left out of stry…

a016d7f

…ker tap.testFiles Refs #14547.

diegosouzapw mentioned this pull request Sep 24, 2026

test(deepseek): reconcile Chat default and explicit Responses fixtures #14516

Merged

diegosouzapw mentioned this pull request Sep 24, 2026

fix(providers): alternateFormats selection only changes the format — baseUrl, authHeader and headers are never applied (#14316 half-landed) #14536

Closed

diegosouzapw added 3 commits September 23, 2026 22:23

Merge remote-tracking branch 'origin/release/v3.8.51' into fix/releas…

4dd84c6

…e-v3.8.51-basereds-0922-tests # Conflicts: # tests/unit/chat-route-coverage.test.ts

diegosouzapw mentioned this pull request Sep 24, 2026

fix(sse): empty-turn retry (#14213) leaves credentials pointing at the retry connection when it falls back to the original response #14715

Closed

diegosouzapw added 2 commits September 23, 2026 22:47

chore(test): keep the tip's stryker order for the #14486 test entry

9b68b80

Refs #14496

diegosouzapw mentioned this pull request Sep 24, 2026

test(batches): boot the batch E2E suite through the peer-stamped server (base-red #14547) #14718

Merged

diegosouzapw added 5 commits September 23, 2026 23:54

Merge remote-tracking branch 'origin/release/v3.8.51' into fix/releas…

0cd0dc0

…e-v3.8.51-basereds-0922-tests # Conflicts: # tests/unit/provider-node-reserved-prefix.test.ts

test: fold the DeepSeek Responses credentials into one fixture; short…

5dd0f25

…en breaker comments Keeps the #14316 realignment from growing chatcore-translation-paths by 29 lines (now +12) and trims the #14530 comments in account-fallback-service. Refs #14496

diegosouzapw mentioned this pull request Sep 24, 2026

[dup] fix(quality): restore release typechecks and env-doc sync #14683

Closed

5 tasks done

L4XB mentioned this pull request Sep 24, 2026

fix(sse): empty-turn retry keeps credentials on the connection that served the turn #14729

Merged

5 tasks done

diegosouzapw mentioned this pull request Sep 24, 2026

fix(quality): drain the 2026-09-22 release-tip base-reds #14520

Closed

diegosouzapw merged commit 23c01f3 into release/v3.8.51 Sep 24, 2026
6 of 9 checks passed

woodsonl mentioned this pull request Sep 30, 2026

fix(db): merge partial cavemanOutputMode writes with the stored row #15206

Merged

diegosouzapw mentioned this pull request Oct 8, 2026

🔴 Release branch not green: release/v3.8.52 #15306

Open

Navigation Menu

Repository navigation

fix(ci): drain the release/v3.8.51 base-reds — auggie/projectCombo typecheck + three stale catalog guards (#14496) - #14511

fix(ci): drain the release/v3.8.51 base-reds — auggie/projectCombo typecheck + three stale catalog guards (#14496)#14511
diegosouzapw merged 23 commits into
release/v3.8.51from
fix/release-v3.8.51-basereds-0922-tests

diegosouzapw commented Sep 22, 2026 •

edited

Loading

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

diegosouzapw commented Sep 22, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Dois defeitos de compilação — API Route Typecheck estava a 4 erros fora do baseline

Três guards que ficaram para trás de mudanças deliberadas

O que esta PR não cobre

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

diegosouzapw commented Sep 22, 2026 •

edited

Loading

Dois defeitos de compilação — `API Route Typecheck` estava a 4 erros fora do baseline