Validado no worktree combinado: typecheck:core, changelog-integrity, complexity, cognitive-complexity, file-size, lint e teste focado (colocate-standalone-esm-scope) todos verdes. Fix real, correção de regressão introduzida por #10836 (server.js CommonJS quebrando com type:module reintroduzido). CI vermelho é o base-red já rastreado em #9985. Obrigado!
Validado no worktree combinado: typecheck:core, changelog-integrity, complexity, cognitive-complexity, file-size, lint e testes focados (freeProviderRankings-filters) todos verdes. Feature aditiva bem documentada (campo reliability nos rankings). CI vermelho é o base-red já rastreado em #9985. Obrigado!
Validado no worktree combinado: typecheck:core, changelog-integrity, complexity, cognitive-complexity, file-size, lint todos verdes. Fix real bem documentado (loopback readiness gate memorizava falha permanentemente + log-spam por caller). CI vermelho é o base-red já rastreado em #9985. Obrigado!
Validado no worktree combinado: typecheck:core, changelog-integrity, complexity, cognitive-complexity, file-size, lint e testes focados (auth-login-route, login-bootstrap-route, feature-flags-settings — corrigi EXPECTED_FEATURE_FLAG_COUNT 51→52 fix-in-place, novo flag adicionado sem atualizar a própria contagem) todos verdes. CI vermelho é o base-red já rastreado em #9985. Obrigado!
Reconciliado com #10935 (já mergeada) — mesclado o guard inline recém-mergeado com a extração para `privateHostname.ts` deste PR, mantendo a intenção original: os 3 workers de relay agora usam a MESMA função compartilhada. Validado: lint limpo, 49/49 testes focados passando (incluindo verificação de que nenhum worker mantém cópia inline). Hardening de segurança real e bem documentado (4 gaps de bypass: `::`, `localhost.`, `::127.0.0.1`, `feb0::1`). CI vermelho é o base-red já rastreado em #9985. Obrigado!
Validado no worktree combinado do lote: typecheck:core, changelog-integrity, complexity, cognitive-complexity, file-size e 163 testes focados (incluindo cloudflare-relay-path-ssrf) todos verdes. Fix de segurança real e bem documentado (SSRF via concatenação pós-validação no Cloudflare relay worker). CI vermelho é o base-red já rastreado em #9985. Obrigado!
Reconciliado com a release (mesmo drift dos PRs irmãos em typecheck-baseline/glm.ts/fetchTimeout.ts/stryker.conf.json). Validado: lint limpo, teste focado passando. Fix real (schema Zod não incluía customSystemPromptEnabled/customSystemPrompt, causando perda silenciosa da configuração). CI vermelho é o base-red já rastreado em #9985. Obrigado!
Reconciliado com a release (mesmo drift dos PRs irmãos) e corrigi 2 problemas de lint reais: import restrito `@/lib/localDb` → `@/lib/db/settings`, e `no-explicit-any` no teste (tipo explícito no callback do map). Validado: lint limpo, 2/2 testes focados passando. Fix real de segurança — endpoints de busca agora respeitam `blockedProviders`. CI vermelho é o base-red já rastreado em #9985. Obrigado!
Reconciliado com a release (mesmo drift de typecheck-baseline/glm.ts/fetchTimeout.ts/stryker.conf.json dos PRs irmãos) e corrigi o `no-explicit-any` no teste novo (cast tipado, mesmo padrão do repo). Validado: lint limpo, teste focado passando. Fix real (busca de credenciais opencode-zen/opencode via PROVIDER_SEARCH_PAIRS). CI vermelho é o base-red já rastreado em #9985. Obrigado!
Reconciliado com a release (mesmo drift de typecheck-baseline/glm.ts/fetchTimeout.ts/stryker.conf.json que os PRs irmãos) e corrigi o `no-explicit-any` no teste novo (o tipo `ModelCompatOverride` já expõe apiFormat/targetFormat/supportsVision — o cast era desnecessário). Validado: lint limpo, 2/2 testes focados passando. Fix real e bem documentado (persistência de overrides de protocolo por modelo). CI vermelho é o base-red já rastreado em #9985. Obrigado!
Reconciliado com a release (drift em typecheck-baseline.json/glm.ts/fetchTimeout.ts/stryker.conf.json — a tip já simplificou essas funções, mantida a versão da tip) e corrigi o `no-explicit-any` no teste novo (cast tipado, mesmo padrão já usado em outros testes do repo). Validado: lint limpo, teste focado passando. Fix real e bem documentado (claude-*/gemini-*/gemma-* sem provider ativo agora retorna 404 model_not_found em vez de 401 enganoso). CI vermelho é o base-red já rastreado em #9985. Obrigado!
Validado + reconciliado: 86/86 testes focados (combo-disable-session-stickiness, base-executor-sanitize-effort, command-code-executor) passando. Incluí o rebaseline do file-size (commandCode.ts 1023→1038, crescimento legítimo deste PR) diretamente no branch — evitando o erro que cometi antes (rebaseline só na worktree local, nunca chegando ao branch real). Correção real de bug com repro ao vivo documentada. CI vermelho é o base-red já rastreado em #9985. Obrigado!
Reconciliado com a release (conflito aditivo em targetTimeoutRunner.ts — combina o warning G3 já mergeado com a resolução de effectiveTimeoutMs deste PR) e revalidado: 41/41 testes focados passando (upstream-timeout-connection-tier, combo-target-timeout-runner, provider-specific-data-schema). CI vermelho é o base-red já rastreado em #9985. Obrigado!
Validado no worktree combinado do lote: typecheck:core, lint, gates de qualidade (file-size rebaselineado com justificativa — crescimento legítimo em modelCapabilities.ts/commandCode.ts) e os 97+9 testes focados (vision-bridge, command-code vision, model-select-field-catalog-vision) todos verdes. Duas correções reais (#10808/#10809) bem documentadas. CI vermelho neste PR é o base-red já rastreado em #9985. Obrigado!
Validado no worktree combinado do lote: typecheck:core, lint, gates de qualidade e os novos testes gcf-numeric-domain/gcf-count-mismatch verdes (mais os já existentes do codec GCF). Fix de losslessness bem documentado e cirúrgico. CI vermelho neste PR é o base-red já rastreado em #9985. Obrigado!
Validado no worktree combinado do lote: typecheck:core, lint, gates de qualidade e os 13 testes unitários + 1 de integração (cline-task-id-propagation) todos verdes. Correção legítima de identidade de tarefa fabricada. CI vermelho neste PR é o base-red já rastreado em #9985. Obrigado!
Validado no worktree combinado do lote: typecheck:core, lint, gates de qualidade e o novo teste OpenClawToolCard-secret-ref-apikey.test.tsx (via vitest) verdes. Correção real e bem isolada de um crash client-side (`e.apiKey.slice is not a function`). CI vermelho neste PR é o base-red já rastreado em #9985. Obrigado!
Validado no worktree combinado do lote: typecheck:core, lint, gates de qualidade e testes focados (opencode-go-catalog-alignment + opencode-go-effort-aliases-8353, incluindo os novos casos muse-spark-1.2-contributor-*) todos verdes. CI vermelho neste PR é o base-red já rastreado em #9985. Obrigado!
Validado no worktree combinado do lote (`.claude/worktrees/batch-round2-0821`): typecheck:core, lint, changelog-integrity, file-size, complexity e cognitive-complexity todos verdes; testes focados (71 casos citados na PR + suíte combo-builder-effort-variants/model-discovery-reasoning-levels) passando. CI vermelho neste PR é o base-red já rastreado em #9985. Obrigado!
Adds a Grok Build card to the CLI Code dashboard: configures the main model and per-subagent model overrides, surgically upserts only the [model.omniroute] section of config.toml (preserving unrelated user sections), resolves config.toml via GROK_HOME or the CLI config home, and omits API keys from settings responses. Closes#10829.
Validated in an isolated worktree boarded onto origin/release/v3.8.50 (0 conflicts, 14 files):
- 6/48 focused node:test cases in tests/integration/cli-settings-grok-build.test.ts initially failed with 422 instead of 200/409 — traced to this route being the first CLI-tool settings route to actually pass a targetPath into the shared guardCliConfigWrite/ensureCliConfigWriteAllowed container-ephemeral-write guard (every other cli-tools/*-settings route calls it with no path, which always no-ops). This devbox genuinely runs inside Docker with no bind mount for the test's tmpdir fixtures, so the guard correctly refused — a real environment-dependent gap in the PR's own test setup (missing OMNIROUTE_ALLOW_CONTAINER_CONFIG_WRITE=1, the pattern already used by tests/unit/cli-tools-apply-container-422.test.ts for this exact guard). Fixed by setting that flag in the test file's setup, pushed fix-in-place.
- 61/61 tests pass after the fix (grok-build-config, check-tool-config-status, all-statuses-route, cli-settings-grok-build, plus 13/13 UI tests via vitest for GrokBuildToolCard/ToolDetailClient).
- Verified the "omit API keys from settings responses" claim: GET applies omitApiKeys() to both config/settings before responding.
- check-file-size, check-changelog-integrity: OK.
- typecheck:core: clean.
- check-complexity / check-cognitive-complexity: OK, both under baseline.
Co-authored-by: tuandinh0801 <tuandinh0801@users.noreply.github.com>
Dashboard-installed SkillsMP and skills.sh skills now store under the canonical global skill scope and merge into every API-key-scoped lookup, so marketplace installs actually reach API keys instead of staying invisible outside the installing session. Tenant-owned skill overrides stay isolated; existing skillsmp/skillssh rows are recognized without a migration, with canonical rows preferred on identity overlap. Closes#9716.
Validated in an isolated worktree boarded onto origin/release/v3.8.50 (0 conflicts, 9 files):
- 94/94 skills-*.test.ts tests pass, including the tenant-isolation regression coverage in skills-injection.test.ts (global skills reach a different API key without leaking another tenant's skills).
- check-file-size, check-changelog-integrity: OK.
- typecheck:core: clean.
- check-complexity / check-cognitive-complexity: OK, both under baseline.
Co-authored-by: kriptoburak <kriptoburak@users.noreply.github.com>
Requests AgentUrlConfig from Cursor with each selected account token and selects the account's actual server-assigned Agent endpoint (agentUrl/agentnUrl) instead of a fixed global/us host, which fails for teams pinned to a different region. Caches validated endpoints by connection+token. Closes#10802.
Validated in an isolated worktree boarded onto origin/release/v3.8.50 (0 conflicts, 4 files):
- 18/18 focused tests pass (cursor-agent-host, cursor-apikey-provider).
- provider-translate-path-golden.test.ts initially failed — traced to a pre-existing base-red (stale golden snapshot left by an earlier freebuff merge, #10531, unrelated to this PR) and confirmed it reproduces on the pure release tip without this PR's changes. Fixed directly on release/v3.8.50 (mechanical key-ordering regen, values unchanged) rather than folding it into this PR's scope; green after merging that fix in.
- check-file-size, check-changelog-integrity: OK.
- typecheck:core: clean.
- check-complexity / check-cognitive-complexity: OK, both under baseline.
- Author additionally validated live: a real Cursor request selected agentn.us.api5.cursor.sh and returned HTTP 200/PING.
Co-authored-by: tuandinh0801 <tuandinh0801@users.noreply.github.com>
Two fixes: (1) createDisconnectAwareStream now distinguishes graceful max_tokens truncation (partial content already reached the client, upstream closes without a terminal marker → clean stop, no error) from a real empty-content failure (still surfaces the 502). Fixes#7699, keeps #8649 intact. (2) liteEngine's compressToolResults now requires an explicit boolean before overriding step config, instead of letting a malformed value leak through the `??` chain.
Validated in an isolated worktree boarded onto origin/release/v3.8.50 (0 conflicts, 4 files):
- 68/68 tests pass (silent-sse-close-7699, compression/lite, empty-stream-no-content-8649, stream-handler).
- check-file-size, check-changelog-integrity: OK.
- typecheck:core: clean.
- check-complexity / check-cognitive-complexity: OK, both under baseline.
Note: the empty-content Claude error message text changed from "Upstream stream ended without a terminal marker" to "Provider returned empty content" (matches the OpenAI/Responses branch wording) — intentional, documented in the PR.
Co-authored-by: minhlongs <minhlongs@users.noreply.github.com>
Adds A2A v1.0 client compatibility: aliases the renamed v1.0 method names (SendMessage → message/send, SendStreamingMessage → message/stream) and reshapes the synchronous reply into the v1.0 SendMessageResponse shape (task.status.message.parts[].text, task.artifacts) for requests that arrive via a v1.0 method — v0.3 callers keep the exact existing response. Also serves a v1.0 Agent Card at /.well-known/agent-card.json declaring both protocol versions on the same JSON-RPC endpoint.
The PR had no tests and the author noted they couldn't build/typecheck locally — validated in an isolated worktree boarded onto origin/release/v3.8.50 (0 conflicts, 2 files):
- Wrote a TDD regression test (tests/unit/a2a-v1-compat-10839.test.ts, 4 tests) exercising the real POST handler end-to-end: v1.0 SendMessage aliasing + response reshaping, v0.3 message/send keeping its existing shape, SendStreamingMessage no longer 404ing, and the new agent-card.json route declaring both 1.0/0.3 interfaces. Pushed fix-in-place.
- 20/20 existing A2A tests still pass (a2a-auth-timing-safe, a2a-enabled-route, a2a-tasks-auth, t09-a2a-lifecycle, agent-card-route) — no regressions.
- check-file-size, check-changelog-integrity: OK.
- typecheck:core: clean (the author's local-build concern didn't reproduce).
- check-complexity / check-cognitive-complexity: OK, both under baseline.
Co-authored-by: wpec <wpec@users.noreply.github.com>
Re-runs hoistLeadingSystemMessage on the final outbound array at translateRequest's single return, instead of only pre-translation. claudeToOpenAI (and the Responses source path, which never ran the pre-translation hoist at all since `messages` doesn't exist yet there) re-introduces/normalizes a leading system message after the hoist already ran, so a strict provider (e.g. vLLM/Qwen3, xiaomi-mimo) could still receive a non-compliant array and 400 with "System message must be at the beginning."
Validated live against a vLLM/Qwen3 endpoint (documented in the PR) plus in an isolated worktree boarded onto origin/release/v3.8.50 (0 conflicts, 2 files):
- 39/39 focused tests pass (probe-7293-strict-system-hoist including the new Claude-source regression case, memory-system-first-6135, claude-system-role-cache-boundary, memory-cache-safe-injection).
- check-file-size, check-changelog-integrity: OK.
- typecheck:core: clean.
- check-complexity / check-cognitive-complexity: OK, both under baseline.
Co-authored-by: Kizuno18 <Kizuno18@users.noreply.github.com>
OpenCode Zen serves muse-spark-1.2 and muse-spark-1.2-contributor-free only on the OpenAI Responses API endpoint, not /chat/completions. Declares targetFormat: "openai-responses" for both so requests route correctly instead of returning null/empty content. Closes#10867.
Validated in an isolated worktree boarded onto origin/release/v3.8.50 (0 conflicts, 1 file):
- Added a TDD regression test (tests/unit/opencode-muse-spark-responses-10867.test.ts) since the PR had none — confirmed RED against origin/release/v3.8.50 (entries absent) and GREEN on this branch, pushed fix-in-place.
- check-file-size, check-changelog-integrity: OK.
- typecheck:core: clean.
- check-complexity / check-cognitive-complexity: OK, both under baseline.
Co-authored-by: zoser69 <zoser69@users.noreply.github.com>
Adds a snapshot-generation button to each Auto-Combo catalog card: computeSnapshotWeights() scores candidates (taskFit/stability/tierPriority/costInv) at combo-creation time instead of the previous hardcoded weight:1, and the new POST /api/combos/duplicate endpoint materializes any auto/* template into a persistent, editable static combo with normalized weights. Closes#10231.
Validated in an isolated worktree boarded onto origin/release/v3.8.50 (0 conflicts, 51 files):
- 19/19 focused tests pass (snapshot-weights, combos-duplicate-route, combos-duplicate-resolution-audit) — covers auth gate (401/403), input validation (400/422), success shape, weight normalization, naming/dedup, and error-response sanitization (no stack traces).
- check-file-size, check-changelog-integrity: OK.
- typecheck:core: clean.
- check-complexity / check-cognitive-complexity: OK, both under baseline.
Co-authored-by: swingtempo <swingtempo@users.noreply.github.com>
Adds native support for Freebuff (Codebuff CLI free-tier gateway): executor with upstream session acquisition, agent-run lifecycle (START/FINISH), canonical system-prompt injection, and model→agent mapping for 9 free models; registry entry, dashboard branding/icons, and API-key validation. Closes#6793.
Validated in an isolated worktree boarded onto origin/release/v3.8.50 (0 conflicts after resolving generated provider-count drift, 15 files):
- 9/9 focused tests pass (freebuff-provider, providers-constants-split).
- Dropped one out-of-scope, unrelated hunk in scripts/build/build-next-isolated.mjs (build-memory heap tuning) that had nothing to do with the Freebuff provider — kept the branch scoped to its stated purpose.
- check-changelog-integrity: OK.
- file-size: gateways.ts and AddApiKeyModal.tsx crossed the frozen cap by +15/+5 lines (irreducible catalog-entry + credential-hint additions) — rebaselined with justification, pushed fix-in-place to the PR branch.
- typecheck:core: clean.
- check-complexity / check-cognitive-complexity: OK, both under baseline.
Co-authored-by: adrianaryaputra <adrianaryaputra@users.noreply.github.com>
Aligns provider-test CLI paths with the server's connection-owned management API: `omniroute test` now resolves a connection and calls `POST /api/providers/{id}/test` instead of the missing `/api/v1/providers/test` route; `--all-providers` carries exact connection ids into both non-interactive and TUI runs. Fixes#10570.
Validated in an isolated worktree boarded onto origin/release/v3.8.50 (0 conflicts, 4 files):
- 42/42 focused tests pass (cli-provider-test-routes-10570, cli-providers-command, cli-providers-rotate, cli-route-unavailable-fallback-10081, cli-expanded-commands).
- One pre-existing test in cli-expanded-commands.test.ts (not touched by the PR) mocked the old route and the old `success` response field, exposed only after merging with the current release tip — fixed the mock to match the new per-connection route and the `valid` field the real route actually returns, pushed fix-in-place to the PR branch (owner-authorized rule: fix-in-place over reimplementation, credit preserved).
- check-file-size, check-changelog-integrity: OK.
- typecheck:core: clean.
- check-complexity / check-cognitive-complexity: OK, both under baseline.
Co-authored-by: hydraxman <hydraxman@users.noreply.github.com>
Mirrors the existing prompt_tokens_details cache-field mapping from the message_delta finish path into the message_stop fallback, so OpenAI-compatible clients see cache_read/cache_creation counters when the finish signal arrives without usage on the same event. Fixes#10535.
Validated in an isolated worktree boarded onto origin/release/v3.8.50 (0 conflicts, 2 files):
- 16/16 focused tests pass (translator-resp-claude-to-openai.test.ts), including the new regression test for this exact fallback path.
- check-file-size, check-changelog-integrity: OK.
- typecheck:core: clean.
- check-complexity / check-cognitive-complexity: OK, both under baseline.
Co-authored-by: NahuSaruf <NahuSaruf@users.noreply.github.com>
Canonical provider id renamed freepik → magnific (Magnific Mystic official API), with a permanent redirect + runtime alias so old freepik/<model> traffic and /dashboard/providers/freepik URLs keep working. Existing provider=freepik connection rows are rewritten to magnific by migration 160.
Closes#10604.
Validated in an isolated worktree boarded onto origin/release/v3.8.50 (0 conflicts, 138 files):
- Focused suite: 54/54 tests pass (magnific-image-handler, provider-validation-image-only, provider-alias-uniqueness, redirects-cli-renames).
- check-file-size, check-changelog-integrity: OK.
- typecheck:core: clean.
- check-complexity / check-cognitive-complexity: OK, both under baseline.
Co-authored-by: RaviTharuma <RaviTharuma@users.noreply.github.com>
Rescoped from #10606 — see PR body for the full rationale (combo-routing half made moot by #10162's advisory-only architecture, chatCore.ts hard-reject bypass retains real value).
Validated in an isolated worktree boarded onto origin/release/v3.8.50:
- 65/65 focused unit tests pass (chatcore-model-output-cap-wiring + feature-flags-settings).
- check-file-size, check-changelog-integrity: OK.
- typecheck:core: clean.
- check-complexity / check-cognitive-complexity: OK, both under baseline.
Co-authored-by: JxnLexn <10897478+JxnLexn@users.noreply.github.com>
Obrigado — bug real: MUSIC_PROVIDERS.minimax declara format "minimax-music" e seus modelos são publicados pelo catálogo, mas handleMusicGeneration nunca teve um branch para esse format — todo request minimax/* caía no guard final com "Unsupported music format", modelos anunciados mas inalcançáveis. Handler completo cobrindo os dois output formats (url/hex), envelope base_resp, endpoint regional, e guarda local de credencial ausente.
Validação (worktree própria a partir de origin/release/v3.8.50, merge limpo, 0 conflitos):
- typecheck:core limpo, complexity/cognitive-complexity dentro do baseline
- tests/unit/minimax-music-generation.test.ts — 9/9 passando
Live SSE frames for a phase:"commentary" message were already dropped
per #6199, but the terminal response.completed.response.output array was
forwarded verbatim whenever the upstream echoed the same item back
non-empty, since backfillResponsesCompletedOutput only fills an empty
array. Reuse the existing isResponsesCommentaryMessageItem predicate to
filter the terminal snapshot's output array (and, defensively, the
backfill buffer it can be seeded from) so both representations agree.
Regression test added to tests/unit/responses-commentary-passthrough-6199.test.ts
reproducing the exact upstream shape from the issue.
v1SearchSchema.provider was a hard-coded z.enum that rejected any id outside
its list before the route's own resolveSearchProvider() check ever ran,
so unknown/short-alias provider ids (grok, brave, serper, ...) always
surfaced a generic "Invalid request" instead of the informative
"Unknown search provider: <id>" message. Relax the schema to a free-form
string and let resolveSearchProvider() own runtime validation (as it
already did for ids that passed the enum). Also extend
SEARCH_PROVIDER_ALIASES with short-form aliases mirroring the existing
jina/jina-ai pattern (brave, serper, perplexity, exa, tavily, google-pse,
linkup, ollama, searchapi, youcom, searxng, zai, duckduckgo), and surface
the first Zod validation issue's field name instead of the generic
message for other still-invalid fields (e.g. search_type).
