mirror of
https://github.com/diegosouzapw/OmniRoute.git
synced 2026-08-20 14:12:59 +03:00
bb07a8e5ffdcdfa8bdb59683fd8e4e49f84bf272
185 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
Diego Rodrigues de Sa e Souza
|
670e8314cc |
fix(translator): thread model through normalizeResponsesReasoningEffort in promotion path (#8997)
Closes #8997 Refs: base-red #9737 fix/8997-gpt56-max-reasoning-rewritte |
||
|
小妍儿 ✨
|
9dc8cdbcb1 |
fix(kiro): keep interleaved tool results grouped without dropping assistant text (#8903) (#8931)
Merge-train validated (tip 6ce4effef8). Vitest failures confirmed as base-red (#9679). |
||
|
Diego Rodrigues de Sa e Souza
|
ff679ab86e |
fix(sse): move Antigravity client system content to first user message to avoid upstream 429 (#9030)
Closes #9030 |
||
|
Diego Rodrigues de Sa e Souza
|
c9debe92bd |
fix(translator): restore original tool name casing in Gemini response translators (#9568)
Closes #9568 |
||
|
diegosouzapw
|
0afbe3295b | Merge branch 'release/v3.8.50' into pr-8755-head | ||
|
Diego Rodrigues de Sa e Souza
|
e7f6b1d130 |
feat(radar): flag-gated signed free-model catalog overlay (#9515)
* feat(dashboard): add RADAR_ENABLED flag (default off) * feat(db): radar feed cache + settings with encrypted supporter key * feat(radar): signed feed sync with pinned key and version floor - feedSchema.ts: Zod v4 schema mirroring the server feed format (discriminated union on budget.kind, enum constraints, etc.) - pinnedKeys.ts: Ed25519 SPKI-DER pinned key + env override for forks - verify.ts: signature verification over exact wire bytes, never throws - sync.ts: full download/verify/validate/cache pipeline with injectable deps, feature-flag gate, opt-in gate, version floor (numeric compare), and sanitized error reasons (no stack traces) - 40 tests covering: contract hash, key handling, sig verification, schema validation, version compare, all sync paths (disabled, opt_out, invalid_signature, invalid_schema, stale, updated, error), auth header injection, and cache-untouched assertions for every failure mode * feat(radar): read-time overlay merge rules over the free catalog Pure function applyFeed() merges the cached Radar feed over the static baseline catalog at read time, honoring 4 rules: 1. Feed never overwrites a local override field. 2. enabled:false disables the entry with disabledBy:"radar" provenance. 3. User-added entry NOT in the feed survives untouched. 4. User deletion tombstone prevents feed resurrection. getRadarCatalog() accessor in index.ts: flag off / no cache / corrupt payload all fall back to baseline. Valid cache applies the overlay and returns feed metadata (version, tier, fetchedAt). TDD: 19 tests (4 rules + dedup + origin + accessor flag/cache/corrupt/ valid/bad-feed + baselineToMergedEntries converter). * feat(dashboard): radar catalog and guided setup screens - API routes: GET /api/radar/catalog, POST /api/radar/sync, POST /api/radar/settings - All gated on RADAR_ENABLED flag (404 when off) - Error responses via buildErrorBody(), never raw stack/message - Settings never echoes clear supporter key (masked omr_****<last4>) - Sync delegates to syncRadar() server-side, never proxies feed URL - Dashboard pages: - /dashboard/radar: 4 states (flag off, opt-in pending, empty, populated) - /dashboard/radar/setup?provider=X: guided setup with steps, key URL, test connection - Uses existing Card component and next-intl patterns - Sidebar: radar entry in costs group with icon - i18n: pt-BR and en keys for radarPage and radarSetupPage namespaces - Tests: - radar-api-routes.test.ts: 11 tests (flag-off 404, flag-on shape, error sanitization) - radar-page-state.test.ts: 5 tests (pure state logic) - All 90 radar tests pass (including prior 74) * docs(radar): module doc and flag-off inertia test Add docs/frameworks/RADAR.md covering the flag gate, the separate data-sync opt-in and privacy promise, the Ed25519 signature/pinned-key security model, tiers, the read-time overlay merge rules, and the self-hosting env vars — plus index entries in CLAUDE.md/AGENTS.md/docs/README.md/REPOSITORY_MAP.md. Document RADAR_FEED_URL and RADAR_FEED_PUBKEY in .env.example and docs/reference/ENVIRONMENT.md to satisfy check:env-doc-sync, which was failing on this branch since the sync.ts commit added the reads. Add tests/unit/radar-inertia.test.ts as the single canonical place asserting the "RADAR_ENABLED off => zero behavioral delta" claim end to end: the three /api/radar/* routes 404, the flag resolves to the definition default with no override, getRadarCatalog() returns exactly the baseline without touching the cache, and computeFreeModelTotals() keeps its pinned values with the Radar module imported alongside it. * fix(db): renumber radar migration to 135 after collision with 134 The base branch introduced 134_proxy_logs_egress_ip while this branch carried 134_radar_cache_settings; the migration runner rejects duplicate numeric prefixes. This migration has never been applied to a real database (the PR is unmerged), so no retroactive isSchemaAlreadyApplied guard is needed. * i18n(radar): translate radar catalog and setup strings to all locales The UI-coverage ratchet measures (present - placeholder) / total_en, so the __MISSING__ sentinels that i18n:sync-ui writes do not count as covered — only real translations restore the metric. Scoped to this PR's namespaces (radarPage, radarSetupPage, sidebar.radar*) instead of a bulk sync, which would have pulled ~978 unrelated pending keys into this diff. Placeholders and code identifiers verified preserved across all 1682 strings. * fix(radar): trust the served-tier header instead of the signed body field The signed feed body always carries tier:"live" by design (one signed artifact per version — rewriting the field server-side per request would break the exact-bytes Ed25519 signature). The server now returns the tier ACTUALLY served via the x-omniroute-feed-tier response header, so free users on a delayed community snapshot no longer see "Ao vivo (tempo real)" in the UI. sync.ts now reads and validates that header (falling back to the body's tier only when the header is absent or holds an unrecognized value) and stores the served tier in the cache; index.ts already surfaces cache.tier to the UI unchanged. * test(combo): shorten an assert message that exceeded the line limit The assertion added by #9507 was 104 chars, so prettier reformatted it into five lines on the next commit that touched the file, pushing it past its frozen size (3449) and failing check:file-size. The message is shortened (the issue reference stays in the comment directly above); the assertion itself is unchanged, and the file is back to 3448 lines and prettier-clean. * i18n(radar): use the canonical zh-TW glossary terms The machine translation produced retired renderings the glossary gate blocks: 供應商 for provider (canonical 提供者) and 文檔 for documentation (canonical 文件). Fixed across the 11 affected radar strings; tests/unit/i18n-glossary-consistency-check.test.ts is back to 17/17. * fix(radar): point the default feed URL at the domain that exists radar.omniroute.dev was a placeholder for a domain that was never registered, so an out-of-the-box sync would fail DNS resolution for every user. The live feed is served from radar.omniroute.online (the subdomain the design always specified), now behind Cloudflare TLS. Forks still override it via RADAR_FEED_URL. --------- Co-authored-by: diegosouzapw <diegosouzapw@users.noreply.github.com> |
||
|
Diego Rodrigues de Sa e Souza
|
e0759f6485 |
fix(claude): reconcile compacted tool results (#9308)
Validated in local merge-train (devbox-vm-06-dev002) @ combined-tip (FAST gates — only pre-existing audit.test.ts flake). |
||
|
Prudhvi Vuda
|
c485c5e0a5 |
fix(sse): preserve Gemini thought_signature on Claude Desktop tool turns (#9015)
Validated in local merge-train (devbox-vm-06-dev002) @ combined-tip (FAST gates: typecheck/complexity/cognitive/changelog/vitest — only pre-existing audit.test.ts flake). Evidence: /home/diegosouzapw/dev/proxys/OmniRoute/.claude/worktrees/merge-train-20260805-222248-suite.log |
||
|
VXNCXNX
|
a5a78fd1d8 |
fix(kiro): preserve GPT-5.6 Max reasoning via Responses (#9163)
Validated in local merge-train (devbox-vm-06-dev002) @ combined-tip (FAST gates green: static + changed tests + vitest — only pre-existing audit.test.ts flake). Evidence: /home/diegosouzapw/dev/proxys/OmniRoute/.claude/worktrees/merge-train-20260805-213228-suite.log |
||
|
VXNCXNX
|
4533dd245f |
feat(providers): native xAI Agent Tools passthrough for /v1/responses (#9111)
Validated in local merge-train (devbox-vm-06-dev002) @ combined-tip (FAST gates green: static + changed tests + vitest — only pre-existing audit.test.ts flake). Evidence: /home/diegosouzapw/dev/proxys/OmniRoute/.claude/worktrees/merge-train-20260805-213228-suite.log |
||
|
Dohyun Jung
|
5f3f25e541 |
fix(kiro): keep relocated tool documentation on multi-turn requests (#9036)
Validated in local merge-train (devbox-vm-06-dev002) @ combined-tip (FAST gates green: static + changed tests + vitest — only pre-existing audit.test.ts flake). Evidence: /home/diegosouzapw/dev/proxys/OmniRoute/.claude/worktrees/merge-train-20260805-213228-suite.log |
||
|
ikelvingo
|
c790b57af8 |
fix(translator): pass output_config.effort=max through verbatim (#9053)
The claude->openai translator was unconditionally rewriting max to xhigh, which broke any OpenAI-shape upstream that accepts max literally (e.g. ollama-cloud, opencode-go deepseek, moonshot k3, native Claude). Provider-aware effort policy is owned by sanitizeReasoningEffortForProvider in the executor; the translator should only do form conversion. Regression guard: tests/unit/base-executor-sanitize-effort.test.ts end-to-end case (claude -> ollama-cloud preserves max). |
||
|
csoftware-arigpt
|
c30724742d |
fix(sse): replay Gemini thought_signature on direct Claude→Gemini path (#2504)
Direct Claude↔Gemini translator was missing the thought_signature round-trip that the OpenAI-hub path already had (#2504). Gemini 3+ thinking models strictly validate thought_signature on every functionCall part in a multi-turn tool-call batch and return HTTP 400 ("Function call is missing a thought_signature in functionCall parts") when it is absent — breaking all agentic workflows through Claude Code. Three fixes across the direct path: 1. translator/index.ts — thread the per-connection signature namespace (connectionId) into the direct-path credentials, mirroring the hub path. Without this, claudeToGeminiRequest never receives _signatureNamespace and cannot look up stored signatures. 2. translator/response/gemini-to-claude.ts — capture thoughtSignature from Gemini response parts (functionCall, thought, or standalone signature parts) via state.pendingThoughtSignature, and persist it keyed by buildGeminiThoughtSignatureKey(connectionId, toolId). The signature frequently lands on a preceding thought part rather than the functionCall itself, so pending tracking across stream chunks is required. 3. translator/request/claude-to-gemini.ts — for each tool_use block, resolve the stored signature and attach it as thoughtSignature on the functionCall part. When no signature is available (historical tool calls predating the store, or cold-start), omit the functionCall part entirely and convert the matching tool_result to plain text — mirrors openai→gemini context mode, avoiding the bare-functionCall 400 while preserving conversation context. |
||
|
nguyenha935
|
d4b9ce6016 |
fix(kiro): harden auth flows, quota lookup, and model discovery (#8565)
* fix(kiro): fetch builder id quota without profile arn * fix(kiro): harden auth imports polling and model discovery * fix(kiro): preserve auth identity and OAuth polling semantics * docs(changelog): add Kiro auth and model discovery fix --------- Co-authored-by: Nguyễn Thanh Hà <nguyenha@Mac-mini-M4.local> Co-authored-by: nguyenha935 <208228297+nguyenha935@users.noreply.github.com> |
||
|
Hernan Javier Ardila Sanchez
|
3be0a5d290 |
fix: combo input-bound, Responses->Chat image strip, qwen-web toolCalling, empty-response exhaustion (#8476)
* test(tail): retire stale i18n __MISSING__ repro + fix qianfan website URL
Base-red slice 6, rebased onto the advanced release/v3.8.49 (
|
||
|
Diego Rodrigues de Sa e Souza
|
e0aef4deb9 |
fix(translator): set status:completed on Responses input items to satisfy strict upstream validators (#8083) (#8507)
Co-authored-by: ikelvingo <im.kelvinwong@gmail.com> |
||
|
backryun
|
8eebda13ca |
refactor(sse): resolve open-sse utils/translator type diagnostics for TS 7 (#8483)
First slice of the TypeScript 7 migration split requested on #7697: resolve the type diagnostics under `open-sse/tsconfig.json` in the lowest-risk modules, with no toolchain change. 12 diagnostics across 8 files, all outside the hot path — `chatCore.ts` and `stream.ts` are deliberately left for a later, standalone slice. Fixes, by cause: * `Transformer.cancel` (progressTracker, sseHeartbeat, and stream.ts's existing handler) — the WHATWG Streams standard defines `transformer.cancel(reason)` and Node implements it (verified on v24: cancelling the readable side invokes it), but `lib.dom.d.ts` still omits it from `Transformer`, so every such handler was TS2353. These handlers clear the heartbeat/progress intervals when an SSE client disconnects, so deleting them to satisfy the checker would leak a timer per abandoned stream. The interface is patched in `open-sse/types.d.ts` instead. * `earlyStreamKeepalive` — `SettledHandler` was discriminated by `ok: true | false`. This workspace compiles with `strictNullChecks: false`, where a boolean-literal discriminant narrows the positive branch but not the negative one, so reading `.error` off the rejected arm did not type-check (the two `.response` reads elsewhere in the file did, which is why only one site errored). Retagged with a string discriminant, which narrows both branches under the same settings. * `toolCallShim` / `openai-responses` — assigning back to a property declared `unknown` resets the `typeof` narrowing, so the following comparison no longer saw a number/array. Both now read through a local. The `Read` limit clamp is behavior-identical: its two branches are mutually exclusive at READ_MAX_LIMIT 2000. * `sanitizeToolResultId` — takes `unknown` but forwards to a `string` parameter; a non-string id previously reached `.replace()` and threw. Coerced instead. * `openaiHelper` — `opts = {}` inferred `{}`; typed as `FilterToOpenAIFormatOptions`. * `cursorAgentProtobuf` — `Buffer.alloc(0)` infers `Buffer<ArrayBuffer>` under @types/node 26 while the decoded field is `Buffer<ArrayBufferLike>`; the locals now use bare `Buffer`, matching `requestMetadata` a few lines above. Validation: 335 -> 321 diagnostics with zero new errors (full tsc error-set diff against the base config). typecheck:core clean, lint clean, check:type-coverage 92.17% -> 94.17%. All 114 existing test files that import a touched module pass; `plan3-p0.test.ts` fails identically with and without this change (it reads the developer's real ~/.omniroute DB instead of a test-scoped DATA_DIR). The new test covers the three behavioral surfaces rather than the refactors the existing keepalive/heartbeat suites already hold: that `transformer.cancel()` really fires and can clear an interval, the id coercion, and the limit-clamp bounds. |
||
|
Bob.Hou
|
406f41de30 |
fix(translator): cap thinking budget on explicit budget_tokens path (#8312)
* fix(translator): cap thinking budget on explicit budget_tokens path * fix(translator): stop dropping thinkingConfig on cap-0 reasoning_effort path The thinking-budget-cap guard added in this branch skipped thinkingConfig entirely whenever a model's thinkingBudgetCap was 0 (e.g. gemini-3-flash), including on the reasoning_effort/budgetMap path. That regressed the pre-#6943 native-defaults contract (thinkingBudget 0 / includeThoughts false must still be present) and crashed callers that read `.thinkingConfig.thinkingBudget` unconditionally (translator-openai-to-gemini-defaults.test.ts). Also restore includeThoughts:true on the Claude-format explicit thinking.budget_tokens path (openai-to-gemini.ts's Claude-format field and claude-to-gemini.ts's native thinking field): budget_tokens:0 there is the client's dynamic-thinking sentinel (#6813), not an off-switch, and must stay true even after the new capping — the cap must only clamp positive explicit values, never flip the zero sentinel's semantics. Updates two tests this branch added that encoded the incorrect "omit thinkingConfig / includeThoughts:false for the 0 sentinel" behavior, to match the pre-existing, still-required contracts above. Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com> * fix(providers): revert scope-creep flip of Gemini 3.5/3.6 Flash supportsThinking The thinking-budget-cap fix accidentally expanded 5 shorthand modelSpecs entries (gemini-3.5-flash, gemini-3.5-flash-low, gemini-3.6-flash-high/ medium/low) into explicit objects setting supportsThinking:true and thinkingBudgetCap:24576. That flip was unrelated to the two proven test regressions (translator-openai-to-gemini-defaults.test.ts and claude-to-gemini-budget-tokens-zero-6813.test.ts, which only exercise gemini-3-flash-preview, gemini-3.1-pro and gemini-2.5-pro) and reopens a deliberately closed path from #8013: Antigravity still rejects client-supplied thinking params for these Gemini 3.5/3.6 Flash tier ids, so supportsThinking must stay false (inherited from GEMINI_35_FLASH_MODEL_SPEC). Reverted all 5 entries back to the release shorthand `{ ...GEMINI_35_FLASH_MODEL_SPEC }`. gemini-3-flash, gemini-3.1-pro and gemini-2.5-pro (the models the regression tests actually exercise) were already correctly specced in the release baseline and are untouched. Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com> --------- Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com> |
||
|
Diego Rodrigues de Sa e Souza
|
ff3d3762af | fix(api): fold namespace into the flattened Chat tool name so cross-namespace leaves do not collide (#8322) | ||
|
Markus Hartung
|
6e1e5c9a45 |
fix(sse): tool-incapable provider handling (AI Horde + Responses content-collapse scoping) (#8212)
* fix(sse): collapse single-text-part Responses-API content to a plain string
Every /v1/responses request — even the simplest single-string input —
got 500'd by AI Horde's Aphrodite-backed facade. Root cause:
normalizeResponsesInputForChat() always wraps a plain string input as
`content: [{ type: "input_text", text: value }]` (a one-element array),
and openaiResponsesToOpenAIRequest() mapped that straight through to
`content: [{ type: "text", text: value }]` on the Chat Completions side
— an array. That's spec-valid (OpenAI's own API accepts both shapes),
but strict/naive OpenAI-compatible backends like AI Horde's only
implement the plain-string form and reject the array form outright.
A single-text-part array and a plain string are semantically
identical, so collapse is safe. Real multi-part messages (text+image,
text+file) are left untouched.
Regression test: tests/unit/openai-responses-single-text-content-string.test.ts
(RED before the fix — every collapsed-content assertion failed with an
object instead of a string; GREEN after).
Also adds a deeper AI Horde load-test suite (sequential/concurrent/
cross-model/sustained-throughput/new-capable-model-candidates) that
surfaced this bug via real live traffic after Behemoth-X-123B was
temporarily added to the "default" combo for evaluation.
Co-Authored-By: Markus Hartung <markus.hartung@gmail.com>
* fix(sse): unsupportedParams provider-level fallback for aihorde's live-discovered models
Real OpenClaw traffic against the newly-added Behemoth-X-123B combo
target kept 500ing on every attempt even after the Responses-API
content-array fix landed. The pipeline artifact showed why: `tools`
was still present, unstripped, in the request actually sent to AI
Horde's Aphrodite backend.
Root cause: `unsupportedParams: ["tools", "tool_choice",
"parallel_tool_calls"]` was only declared on the 3 models statically
listed in the aihorde registry entry (Cydonia-24B, Skyfall-31B,
google/gemma-4-31b). AI Horde uses `passthroughModels: true` — its
live worker roster changes constantly — so Behemoth-X-123B, like every
other dynamically-discovered aihorde model, had no model-specific
unsupportedParams entry, and getUnsupportedParams() returned [] for
it. But "the workers run raw text-completion backends" (no tool
calling) is true of every model AI Horde serves, not just the 3
catalogued ones.
Adds a provider-level `unsupportedParams` fallback on RegistryEntry,
checked by getUnsupportedParams() after the per-model lookup misses.
Set on the aihorde entry so it covers its entire live-discovered
roster, present and future, without needing a static per-model catalog
entry for each one.
Regression test: tests/unit/aihorde-tools-unsupported-provider-fallback.test.ts
(RED before the fix — Behemoth-X and deepseek-v4-flash both returned
[] instead of the stripped param list; GREEN after, with a control
case confirming the fallback doesn't leak to unrelated providers).
Co-Authored-By: Markus Hartung <markus.hartung@gmail.com>
* fix(sse): flatten leftover tool-call history when stripping unsupported tools
Third bug in the same AI Horde/Behemoth-X saga: even after tools/
tool_choice were correctly stripped from the live request (previous
fix), real combo traffic still 500'd. The conversation history itself
carried a prior turn's role:"assistant" tool_calls and role:"tool"
result messages, left over from before the combo failed over from a
tool-capable model (Gemini) to a non-tool-capable one (AI Horde). Its
raw completion backend doesn't understand those message shapes at all,
independent of whether live `tools` is present — confirmed by
reproducing with a role:"tool" message and NO tools param at all.
flattenToolHistory() (open-sse/utils/flattenToolHistory.ts) already
existed for exactly this, fully unit-tested — it just had zero call
sites anywhere in the request pipeline. Extracts the unsupported-params
strip into a small testable module
(open-sse/handlers/chatCore/unsupportedParamsStrip.ts, following the
existing chatCore god-file decomposition pattern e.g.
executorClientHeaders.ts) that now also flattens tool-call history
whenever "tools" was among the stripped params.
Regression test: tests/unit/chatcore-unsupported-params-strip.test.ts
(RED before the fix — the flattening test failed with the raw
tool_calls array still present; GREEN after). All 434 existing
chatcore-*.test.ts tests still pass.
Co-Authored-By: Markus Hartung <markus.hartung@gmail.com>
* fix(sse): gate tool-history flattening on unsupported, not on stripped-this-request
The previous commit's flattening only fired when "tools" was actually
present-and-stripped on THIS request. A second live reproduction
against AI Horde had no live `tools` param at all — only stale
tool_calls/tool-result messages inherited from before a combo
failover — and still 500'd, because that condition never triggered.
A model that can't do tool calling can't do it whether or not the
current request happens to carry a `tools` array. Gate on the
unsupported-params list itself (unsupported.includes("tools")) instead
of the subset that was actually present-and-deleted this time.
Regression test added to the same file (RED before — the no-live-tools
case left tool_calls/role:"tool" untouched; GREEN after). All 435
chatcore-*.test.ts still pass.
Co-Authored-By: Markus Hartung <markus.hartung@gmail.com>
* fix(sse): skip tool-incapable combo targets, error clearly on direct requests
Two complementary fixes for a model that structurally can't do tool
calling at all (e.g. AI Horde's raw completion backends) rather than
silently degrading — following up on the earlier strip/flatten fix,
which stopped the crashes but let a tool-incapable target still get
selected and return a 200 that narrates a fake tool call in prose
instead of erroring or being skipped.
1. Root cause, combo routing: getResolvedModelCapabilities()'s
`supportsTools` resolution only checked per-model registry entries,
synced capabilities, and static specs — none of which exist for a
dynamically-discovered model (AI Horde's passthroughModels roster
changes as workers come and go). It fell through to
heuristicToolCalling(), which optimistically defaults to `true` for
any unrecognized model (TOOL_CALLING_UNSUPPORTED_PATTERNS is empty).
Added a provider-level fallback reusing the same unsupportedParams
signal the request-time strip already relies on. This makes the
EXISTING filterTargetsByRequestCompatibility (comboStructure.ts) —
which already correctly excludes non-tool-capable targets when a
request requires tools — actually work for these models; no combo.ts
changes were needed, it was only ever fed bad capability data.
2. Direct/pinned requests: filterTargetsByRequestCompatibility only
protects combo routing. A direct request naming an exact
tool-incapable model has no other target to fail over to — added
checkToolCallingRequiredButUnsupported (chatCore/toolCallingRequiredCheck.ts),
gated on isCombo: false, returning a clear 400 instead of a 200 that
silently can't do what was asked.
Regression tests (both RED before, GREEN after):
- tests/unit/model-capabilities-provider-unsupported-tools.test.ts
- tests/unit/chatcore-tool-calling-required-check.test.ts
All 463 chatcore-*/model-capabilities-*.test.ts and 31 combo
compatibility-filter tests still pass.
Co-Authored-By: Markus Hartung <markus.hartung@gmail.com>
* fix(sse): correct handleChatCore return shape for the tool-calling-blocked error
handleChatCore's documented contract is `{ success, response, status,
error }`, not a raw Response — returning `new Response(...)` directly
(copied from a different early-return whose surrounding context turned
out not to share this function's top-level contract) produced "No
response is returned from route handler ... Expected a Response object
but received 'undefined'" and a bare 500 with an empty body, caught
immediately when verifying the previous commit live.
Uses createErrorResult() (already used by the adjacent
translation-failure branch a few lines up) instead of hand-building the
Response, matching the same pattern already established in this
function for early error returns.
Co-Authored-By: Markus Hartung <markus.hartung@gmail.com>
* fix(sse): scope Responses single-text-content collapse to providers that need it
The single-text-part content array -> plain string collapse (added for AI
Horde's Aphrodite facade, which 500s on the array form) was applied
unconditionally to every provider, silently breaking the standard OpenAI
array-shaped content contract that other providers and existing tests
depend on. Added RegistryEntry.requiresPlainStringContent, gated the
collapse on it (true only for aihorde), and threaded modelInfo.provider
through responsesHandler -> responsesApiHelper -> the translator so the
real /v1/responses call site can identify the provider.
Co-Authored-By: Markus Hartung <markus.hartream@gmail.com>
---------
Co-authored-by: Markus Hartung <markus.hartung@gmail.com>
Co-authored-by: Markus Hartung <markus.hartream@gmail.com>
|
||
|
Markus Hartung
|
4ea08f520b |
fix(sse): Gemini malformed function-call handling + tool_choice translation (#8211)
* fix(sse): synthesize tool_calls for Gemini's malformed function-call abort reasons Live incident (dashboard log id 1784489701456-d8c0e9): Gemini terminates a stream with finishReason MALFORMED_FUNCTION_CALL/UNEXPECTED_TOOL_CALL when its own parser rejects an attempted tool call — there's no real functionCall part, only a human-readable finishMessage. gemini-to-openai.ts passed this through raw as finish_reason (9router#2462's fix, correctly keeping it off a clean "stop"/Claude end_turn), but a raw "malformed_function_call" isn't one of OpenAI's 5 documented finish_reason values, so a real OpenAI-format client (OpenClaw) has no handling for it at all and silently never notices the turn failed — confirmed live via tests/integration/live-gemini-workload.test.ts's [28] streaming case after the Gemini TPM/rebase work on this branch. Fix: synthesize a tool_calls entry (arguments carry the error code + Gemini's finishMessage, valid JSON) and finish_reason: "tool_calls" instead, routing the failure into the ordinary "tool call arguments didn't parse" path every OpenAI-compatible agent loop already handles. Defers to a real tool call if one already completed earlier in the same turn — the real call wins, no synthetic entry piles on top of it. Tests (TDD, each confirmed red-before-green): - 5 new unit tests in the existing 9router#2462 regression file, covering the synthesis itself, UNEXPECTED_TOOL_CALL, the real-tool-call-wins edge case, and no-regression on a clean STOP. - New fixture (tests/fixtures/translation/gemini-malformed-function-call-stream.json): the real 6-chunk event series from the live incident, sanitized (personal paths/URLs replaced with generic placeholders, structure preserved exactly). - New integration test chains the real translator into the real Responses API transformer using that same fixture, proving correct behavior on BOTH /v1/chat/completions and /v1/responses from one shared ground-truth event series. Co-authored-by: Markus Hartung <markus.hartung@gmail.com> * fix(sse): don't drop a malformed tool-call failure when it lands beside a real one Live incident (dashboard log id 1784589106014-2a42f8), analyzing why the prior malformed-function-call fix (3568c7259) still wasn't reaching the client in this case: Gemini can emit a REAL, valid functionCall AND finish the SAME candidate with MALFORMED_FUNCTION_CALL — the model attempted multiple tool calls in one turn (here: a real status-check call plus a malformed "exec"+"cron" multi-call attempt), one parsed cleanly and the other didn't. The first fix version skipped synthesizing a failure signal whenever a real tool call already existed (state.toolCalls.size > 0), on the assumption that meant the model was retrying a LATER, separate attempt after an earlier one already succeeded. That's indistinguishable, from the translator's state, from this same-turn case — so it silently discarded the malformed attempt's information entirely: the client saw the real call succeed and never learned the other tool calls were attempted and rejected. Fix: always synthesize the failure entry when a malformed abort reason is seen, appending it alongside any real tool call rather than skipping it. Multiple tool_calls in one response is normal, well-supported OpenAI behavior (parallel tool calls), so this adds the failure as an additional entry instead of replacing or hiding the real one. Tests (TDD, confirmed red-before-green): - Rewrote the unit test that encoded the old (wrong) assumption to assert both the real and synthesized calls are present. - New fixture (gemini-malformed-function-call-parallel-real-call-stream.json): the real event series from this incident, sanitized. - New integration tests (same file as 3568c7259's) prove both /v1/chat/completions and /v1/responses surface both tool calls correctly from this fixture. Co-authored-by: Markus Hartung <markus.hartung@gmail.com> * feat(sse): honor tool_choice when translating OpenAI requests to Gemini Investigating a live report that gemini-3.1-flash-lite frequently narrates an intended tool call in plain text instead of actually emitting one (dashboard log id 1784591483850-49c408 — 9 raw provider chunks, all plain text, zero functionCall parts, clean finishReason STOP): body.tool_choice was never read anywhere in the OpenAI->Gemini request translator. result.toolConfig was unconditionally hardcoded to { functionCallingConfig: { mode: "VALIDATED" } } whenever tools were present, regardless of what the caller sent. VALIDATED lets the model respond with plain text OR a schema-validated function call at its own discretion — it never forces a call the way OpenAI's tool_choice: "required" (Gemini's ANY mode) does, so a caller had no way to compel a tool call even when explicitly requesting one. Added convertOpenAIToolChoiceToGemini(), mirroring the existing convertOpenAIToolChoice() in openai-to-claude.ts for the same OpenAI tool_choice shapes (string "auto"/"none"/"required", or {type:"function",function:{name}} to force one specific tool): - unset/"auto" -> VALIDATED (unchanged default, no regression) - "required"/"any" -> ANY (forces a call) - "none" -> NONE (disables function calling) - {type:"function",...} -> ANY + allowedFunctionNames: [name] Wired into both Gemini request paths: the direct/base translator (openaiToGeminiBase) and the Antigravity/Cloud Code envelope (wrapInCloudCodeEnvelope), which now reuses the base translator's already- computed toolConfig instead of re-deriving its own hardcoded VALIDATED. This unblocks (but does not itself resolve) the live question — a tool_choice: "required" A/B test against gemini-3.1-flash-lite follows to confirm ANY mode actually changes the narrate-vs-act behavior in practice. Also updates the T11 any-budget allowlist for this file: the "any" string comparisons (tool_choice value "any", not a TypeScript type) are the same documented false-positive pattern already carved out for executors/base.ts. Tests (TDD, confirmed red-before-green): 9 new unit tests covering all tool_choice shapes on both the direct and Antigravity/Cloud Code paths, plus the no-tools and unset-default no-regression cases. Co-authored-by: Markus Hartung <markus.hartung@gmail.com> * chore(quality): file-size baseline for own-growth (#8211) Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com> --------- Co-authored-by: Markus Hartung <markus.hartung@gmail.com> Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com> |
||
|
Diego Rodrigues de Sa e Souza
|
917314156e |
fix(gemini): drop HARM_CATEGORY_CIVIC_INTEGRITY from the default Gemini safety settings (#8231) (#8238)
Co-authored-by: Probe Test <probe@example.com> |
||
|
backryun
|
888c872459 |
refactor(antigravity): align official clients and callable catalog (#8013)
* fix(antigravity): preserve protocol fidelity and fail closed * chore: add PR-numbered changelog fragment * test: split oversized Antigravity suites * refactor(antigravity): align official IDE and CLI identities * fix(antigravity): align catalog with callable models * test(antigravity): update 2 test files to renamed version-cache API (#8013 fix) Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com> --------- Co-authored-by: nguyenha935 <208228297+nguyenha935@users.noreply.github.com> Co-authored-by: backryun <backryun@users.noreply.github.com> Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com> Co-authored-by: nguyenha935 <nguyenha935@users.noreply.github.com> Co-authored-by: Probe Test <probe@example.com> |
||
|
RCrushMe
|
89bad0fa52 |
fix(responses): close namespace round-trip for Responses-Chat translation (#7936) (#8151)
* fix(responses): close namespace round-trip for Responses-Chat translation (#7936) The #7905 custom-tool-call path landed in release/v3.8.49 but left #7936 open: Responses namespace sub-tools were flattened to a bare leaf on the Chat wire with no response-side closure, so Codex's adjudicator rejected every namespace sub-tool call with `unsupported call` -- it only has a dispatch entry for the header bits (namespace+name), no entry for the bare leaf. This patch closes the round-trip without mutating the Chat wire name (alignment with #7905's bare-leaf contract and with the issue author's proposed fix): * request side (openai-responses.ts): keep tool.function.name as the bare leaf, populate a side-band namespaceToolIdentityMap keyed on that leaf, and thread it through translatedBody._toolNameMap. * request -> response seam (chatCore.ts): extract the identity map before dispatch and pass it through to the non-stream completion path and to all three stream pipelines (translate openai-responses, translate other, passthrough). * response translator (response/openai-responses.ts): in emitToolCall (response.output_item.added) and closeToolCall (custom_tool_call / function_call output_item.done), call resolveRequestToolIdentity() to rewrite the bare leaf back to {namespace,name} and emit codex-compatible independent fields. * passthrough (utils/stream.ts): add a response passthrough rewriter restoreResponsesPassthroughFunctionCallIdentity that intercepts response.output_item.added, response.output_item.done, and response.completed and stamps the same {namespace,name} tuple. * helper (requestToolIdentity.ts): a 20-line stateless resolver; never parses a name. The wire-visible Chat tool.function.name stays the bare leaf -- non-OpenAI providers (NVIDIA, GLM, Kimi, Gemini, ...) frequently truncate or rewrite long __-dotted names; bare leaves avoid that failure mode entirely. The codex ResponseItem::FunctionCall schema (models.rs) declares an independent namespace: Option<String> field and has a function_call_deserializes_optional_namespace round-trip test, so emitting it separately matches the codex adjudicator dispatch. Includes 19 new test cases across 4 files: - request-side bare-leaf wire + side-band ledger construction - response-side tuple emit + unmapped passthrough + apply_patch exclusion - ambiguous-leaf collision safety (entry dropped, leaf emits verbatim) - per-request isolation between concurrent streams - a precompiled Atlassian-style nested namespace override * fix(responses): skip tool_search_call input items instead of 400 (#7936 addendum) Codex 0.42+ emits `tool_search_call` (and later `tool_search_result`) input items when the model uses the dynamic tool-search optimization. They are metadata-only: they record that the model queried a subset of the available tools, and carry nothing that OpenAI Chat Completions can represent. Without an explicit skip in openai-responses.ts, the input loop threw Unsupported Responses API feature: input item type 'tool_search_call' cannot be represented in Chat Completions -- and because these items stay in the Responses API `input` for every follow-up turn, the whole server returned 400 on EVERY subsequent /v1/responses in the same session until the user cleared history. Observed in the wild: /v1/responses 400 "Unsupported Responses API feature: input item type 'tool_search_call' cannot be represented in Chat Completions [longcat/LongCat-2.0 (400), longcat/LongCat-2.0 (400)]" The meituan combo (longcat fallback) was the most visible victim, but the underlying throw is source-format-side and hits any Responses-API consumer whose upstream does not natively support Responses. Fix: stop on the item type the same way `reasoning` is skipped -- display-only metadata, no chat side-effect. Covers both `tool_search_call` and the follow-up `tool_search_result` shapes. Adds 3 unit tests: - tool_search_call is silently skipped (no 400) - tool_search_result is silently skipped - tool_search_call items interspersed with real messages are skipped in order; real messages survive * chore(quality): rebaseline openai-responses.ts + stream.ts own-growth (#7936 namespace round-trip) --------- Co-authored-by: TonPro <hello@tonpro.fu> Co-authored-by: RCrushMe <RCrushMe@users.noreply.github.com> |
||
|
Jan Leon
|
6b59e814da |
Restore Responses API custom tool calls (#7905)
* fix: restore Responses custom tool calls * fix: preserve nested Responses custom tool calls * fix: reset superseded tool call state * fix: preserve tool precedence and buffered Responses tool arguments * test: verify top-level tool descriptions take precedence * fix: preserve custom Responses tool streaming semantics * test: cover declared custom tool streaming round trips * test: cover Responses custom tool metadata collection * test: cover active Responses custom tool stream * test: isolate Responses active stream regression * fix: complete Responses custom tool round trips |
||
|
Jan Leon
|
4f52e36082 |
Preserve supported Responses behavior in Chat translation (#7894)
* fix(responses): preserve additional_tools when downgrading to Chat Completions * fix: preserve Responses structured output in Chat translation * fix: translate Responses allowed tools to Chat * fix: reject unsupported Responses input items * fix: normalize Responses refusal history for Chat * fix: strip Responses-only fields from Chat requests * fix: preserve namespace tools with colliding function names * fix: merge same-named namespaces during Chat translation |
||
|
Diego Rodrigues de Sa e Souza
|
8a4a363bb9 | fix(translator): sanitize tool_result.tool_use_id symmetrically with tool_use.id (#7705) (#7823) | ||
|
Diego Rodrigues de Sa e Souza
|
8245de78a9 |
fix(sse): strip orphaned tool_use before antigravity/Vertex Claude dispatch (#7752) (#7822)
AntigravityExecutor.execute() overrides BaseExecutor.execute() and never calls super.execute(), so the shared orphan-tool_use guard (fixToolPairs, #2382/#4714) never ran on the Antigravity/Vertex Claude dispatch path. A client history with a genuinely orphaned tool_use (no matching tool_result anywhere — e.g. left behind by OpenCode's known abort/cancel bug) sailed straight through openaiToAntigravityRequest into Google's Cloud Code envelope as an unpaired functionCall, which Vertex's Claude backend rejects with HTTP 400. Mirror-image gap of #6026 (incoming direction). Fix: run fixToolPairs on body.messages in openaiToGeminiBase before building the tool_call_id map and the functionCall/functionResponse translation loop. |
||
|
backryun
|
9544fb6353 |
feat(kimi): sync Code, Web, and Moonshot providers (#7531)
* feat(kimi): sync Code, Web, and Moonshot providers * chore(quality): trim frozen file-size overflow in Kimi sync The Kimi/Moonshot provider sync added a net +1 line to both src/sse/services/auth.ts and ProviderDetailPageClient.tsx, pushing each 1 line past its frozen cap in file-size-baseline.json. Drop one optional blank line in each (prettier-neutral, no behavior change) to land back at/under the frozen baseline. Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com> --------- Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com> |
||
|
Xiangzhe
|
e5b240479c |
fix(codex): preserve GPT-5.6 reasoning contract (#7012)
* fix(codex): preserve GPT-5.6 reasoning contract * fix(vscode): expose Responses text models * fix(codex): keep GPT-5.6 limits through discovery * fix(ci): extract isUsableChatModel helpers to satisfy complexity ratchet Splitting the supported_endpoints/output_modalities guard clauses into excludesChatAndResponsesEndpoints() / excludesTextOutputModality() drops isUsableChatModel's cyclomatic complexity from 16 to under the ratchet's max of 15 (complexity-ratchets gate: 2057 -> 2056, back at baseline). Behavior is unchanged; existing vscode/codex route tests cover it. Co-authored-by: diegosouzapw <diegosouza.pw@gmail.com> * fix(codex): merge capacity limits conservatively (smaller of live vs pinned wins) Resolve the #7012 catalog-merge policy collision: instead of the pinned GPT-5.6 contract always winning for a fixed set of model ids, capacity limits (inputTokenLimit/outputTokenLimit) now merge via mergeCapacityLimitConservatively — Math.min(pinned, live) when both are present, so OmniRoute never promises more context than the account can actually serve. All other overlapping fields still take the live value unconditionally. Guard tests cover both directions (pinned smaller wins / pinned larger loses) at the route level and via an isolated helper-level unit test. Co-authored-by: diegosouzapw <diegosouza.pw@gmail.com> --------- Co-authored-by: Xiangzhe <xz-dev@users.noreply.github.com> Co-authored-by: diegosouzapw <diegosouza.pw@gmail.com> |
||
|
Adam
|
40e097cc7a |
fix(translator): preserve thinking.budget_tokens: 0 in Claude->Gemini (#6813) (#7061)
* fix(translator): preserve thinking.budget_tokens: 0 in Claude->Gemini (#6813) * test: regression guard for budget_tokens: 0 in Claude->Gemini (#6813) |
||
|
Adam
|
ff89a3d6ee |
fix(responses): map mid-conversation system turns to developer role (#6954) (#7056)
* fix(responses): map mid-conversation system turns to developer role (#6954) * test(responses): add #6954 mid-conversation system -> developer regression * fix(6954): keep bare-string content parts in buildResponsesTextParts * test(6954): cover array-form system content with bare string |
||
|
Diego Rodrigues de Sa e Souza
|
280c27bf2d |
fix(sse): stop dropping tool_search and leaking OpenAI-only params in Responses->Chat translation (#7571)
* fix(sse): stop dropping tool_search and stop leaking OpenAI-only params in Responses->Chat translation (#7532, #7533) #7532: `openai-responses.ts` unconditionally dropped `tool_search` when downgrading a Responses-shaped request to Chat Completions, hiding the tool from the model and breaking Codex's deferred/lazy tool-discovery protocol for any provider that gets downgraded (e.g. built-in providers like opencode-go). tool_search carries `execution: "client"` — the client resolves the call locally regardless of wire shape — so it is now mapped to a normal Chat function tool, mirroring the existing local_shell -> shell pattern in the same file, instead of being silently discarded. #7533: the same translator unconditionally copied two GPT-5/OpenAI-only fields (`verbosity`, `prompt_cache_key`) into the translated Chat body regardless of destination provider. A strict-protocol non-OpenAI upstream (NVIDIA confirmed by the reporter) 400s on unrecognized top-level parameters. Both fields are now gated on `credentials.provider === "openai"`, stripped otherwise; the existing OpenAI-destined behavior (needed for #517's prompt-caching fix) is preserved byte-identical via a dedicated sanity test. Regression tests: tests/unit/tool-search-filtered-responses-to-chat-7532.test.ts, tests/unit/verbosity-prompt-cache-key-provider-gate-7533.test.ts. Two existing tests that encoded the old buggy contract (unconditional tool_search drop / unconditional field leak with no credentials) were aligned to the corrected contract: tests/unit/translator-openai-responses-req.test.ts, tests/unit/openai-responses-verbosity.test.ts. Gates run green: file-size, complexity, cognitive-complexity, typecheck:core, lint (scoped to changed files), and the full touched-area unit test suite (329 tests, 0 failures). * fix(sse): keep prompt_cache_key/verbosity for the codex destination (#7533) The #7533 provider gate allowlisted only "openai", but /v1/responses routes EVERY request through this downgrade (handleResponsesCore -> convertResponsesApiFormat) regardless of provider, and codex is an OpenAI-operated upstream (chatgpt.com/backend-api/codex). Gating it out stripped prompt_cache_key for Codex and silently re-broke the prompt-cache affinity #517 exists to protect — with no test covering it. Allowlist is now {openai, codex} and carries two #517 regression guards. Non-OpenAI upstreams (NVIDIA) still get both fields stripped, per #7533. |
||
|
Diego Rodrigues de Sa e Souza
|
ca03d619a4 |
feat(mitm): add Antigravity reasoning-effort overrides (#7228)
* feat(mitm): add Antigravity reasoning-effort overrides
The Antigravity MITM alias mapping only ever swapped the destination model;
there was no way to override the reasoning effort Antigravity's own
thinkingConfig requested. Alias entries are now `{ model?, reasoningEffort? }`
(a legacy plain-string mapping still normalizes to `{ model }`, so no DB
migration is required). The standalone proxy (server.cjs) forwards the chosen
tier as a top-level `reasoningEffortOverride` on the intercepted request; the
antigravity->openai translator honors it ahead of its thinkingConfig-derived
guess, and an explicit "none" suppresses reasoning_effort entirely even when
Antigravity's own request asked for thinking. Reuses the existing canonical
5-tier reasoning vocabulary (`@/shared/reasoning/effortStandardization.ts`,
with max/extra aliasing to xhigh) instead of introducing a new one. The API
route validates the reasoning-effort value at the boundary and the Antigravity
tool card UI now exposes a per-model reasoning-effort selector alongside the
existing model-mapping input.
Co-authored-by: Truong Fiu <gnourtf@gmail.com>
Inspired-by: https://github.com/decolua/9router/pull/2584
* chore(changelog): fragment for #7228
---------
Co-authored-by: Truong Fiu <gnourtf@gmail.com>
|
||
|
Diego Rodrigues de Sa e Souza
|
4f97297793 |
fix(translator): preserve Gemini thought parts as reasoning_content on the OpenAI bridge (#7206)
* fix(translator): preserve Gemini thought parts as reasoning_content on the OpenAI request bridge Gemini thinking-mode output marks internal reasoning with `part.thought === true` inside a content's `parts` array. geminiToOpenAIRequest() ran every part (thought or not) through the same text-part branch, so a thought part was merged straight into the message's visible `content` — leaking private reasoning into whatever the OpenAI pivot forwarded downstream, and hiding it from Reasoning Replay Cache (which only ever inspects `reasoning_content`). Add convertGeminiContentWithReasoning(): split out `thought: true` parts before delegating to the existing convertGeminiContent(), then re-attach the joined thought text as `reasoning_content` on the resulting message (skipping tool/ functionResponse messages, whose schema has no such field). Non-strict-provider stripping and reasoning-replay injection in translator/index.ts are untouched — this only fixes what reasoning_content gets populated with on this one inbound hop. Co-authored-by: W ARELIK <warelik@users.noreply.github.com> Inspired-by: https://github.com/decolua/9router/pull/2401 * chore(changelog): fragment for #7206 --------- Co-authored-by: W ARELIK <warelik@users.noreply.github.com> |
||
|
Rafael Dias Zendron
|
df9808c0e4 |
fix(6954,6953): preserve system role + strip empty-signature thinking blocks (#6982)
* fix(6954,6953): preserve system role + strip empty-signature thinking blocks #6954 — System turns misattributed as assistant (claude-to-openai.ts:352) The ternary `msg.role === 'user' || msg.role === 'tool' ? 'user' : 'assistant'` mapped any non-user/non-tool role (including 'system') to 'assistant'. Mid-conversation system turns (Claude format) lost their role on translation to OpenAI format, causing them to be treated as assistant output. Fix: add explicit 'system' branch to the ternary. #6953 — Empty-signature thinking blocks poison Anthropic leg (openai-to-claude.ts) Non-Anthropic providers (codex/gpt-5.x) synthesize thinking blocks with signature:''\. On replay, the old code fabricated a DEFAULT_THINKING_CLAUDE_SIGNATURE to fill the empty signature — but Anthropic rejects foreign signatures with HTTP 400, permanently degrading combo/blend routes to codex-only. Fix: strip thinking blocks with empty/missing signatures and redacted_thinking blocks with empty/missing data entirely. They carry no replayable value. Tests: 8 new tests (4 per bug), all passing. Existing #5312 and #5945 regression tests still pass — no interference. * fix(6953): strip only signature:"" thinking blocks, preserve undefined signature CI caught a regression: translator-helper-branches test had a Claude-format thinking block without signature field (undefined) that was being stripped by the original fix. The fix was too aggressive — it stripped both signature:"" (non-Anthropic synthesized) and signature: undefined (legitimate Claude-format). Correct behavior: - signature === "" (empty string): strip — hallmark of codex/gpt-5.x block - signature === undefined: preserve with DEFAULT_THINKING_CLAUDE_SIGNATURE fallback - redacted_thinking data === "": strip - redacted_thinking data === undefined: preserve with fallback Added regression test for undefined-signature preservation. --------- Co-authored-by: Diego Rodrigues de Sa e Souza <8016841+diegosouzapw@users.noreply.github.com> |
||
|
Diego Rodrigues de Sa e Souza
|
bf293a9794 | chore(release): v3.8.47 pre-flight fixes — orphan test relocation (#6943), eslint suppression match, file-size/zizmor rebaselines | ||
|
Rafael Dias Zendron
|
38d6cd9955 |
fix(6813): fix thinking budget zero drop and default thinkingConfig injection (#6943)
* fix(electron): bump electron 42→43 + build better-sqlite3 from source (ABI 148) (#6605) fix(electron): bump electron 42→43 + rebuild better-sqlite3 from source against the Electron ABI (148). Electron 43 raises NODE_MODULE_VERSION to 148; better-sqlite3@12.11.1 has no electron-v148 prebuild, so the packaged app died with 'Nenhum driver SQLite disponível'. prepare-electron-standalone now compiles better-sqlite3 from source against the electron headers into build/Release (where 'bindings' resolves it). Validated by Electron Package Smoke (green) + local (node_register_module_v148). Supersedes #6378. (--admin: the only reds are SonarQube/SonarCloud failing on a coverage-report artifact digest-mismatch — a GitHub Actions infra flake, not this diff; Sonar is green on main and the diff touches only the electron build.) * deps: bump the development group across 1 directory with 6 updates (#6588) deps: bump the development group (6 updates). Rebased onto current main; all checks green after the electron-smoke fix (#6605). * fix(proxy): force CONNECT tunnel for HTTP proxied requests (undici 8.7) + production deps bump (#6620) fix(proxy): force CONNECT tunnel for HTTP proxied requests (undici 8.7) + production deps bump. undici 8.6+ changed ProxyAgent to forward plain-HTTP via request-proxy instead of CONNECT, breaking OAuth refresh through a connection proxy (501). proxyDispatcher now passes proxyTunnel:true. Validated: Unit Tests 3/8 (the OAuth-proxy test) green, new regression test green (fails without the fix on undici 8.7), SonarQube green. Supersedes #6380. (--admin: the only red is Electron Package Smoke failing on a next-build artifact 'digest-mismatch' — a GitHub Actions infra flake corrupting the asar ('file data stream has unexpected number of bytes'); the better-sqlite3 rebuild itself succeeded (gyp ok) and the electron path is unchanged from #6605 which passed the smoke. Not this diff.) * fix(6813): fix thinking budget zero drop and default thinkingConfig injection - Fix truthy check for budget_tokens to allow 0 - Stop injecting default thinkingConfig when no knobs present - Add tests covering all scenarios Related: #6813 --------- Co-authored-by: Diego Rodrigues de Sa e Souza <8016841+diegosouzapw@users.noreply.github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> |
||
|
Jade Guo
|
10807972db |
fix(sse): default reasoning summary for effort-only Responses requests (#6807)
A Chat-Completions client can only express reasoning via the top-level reasoning_effort hint and has no way to request a reasoning summary. When that hint is promoted to the Responses API's reasoning.effort, the upstream returns an empty summary and downstream chat clients see no thinking stream (encrypted reasoning only). Default reasoning.summary "auto" plus include ["reasoning.encrypted_content"] on the effort-only path so the summary actually streams back to the chat client, mirroring the Codex executor's ensureCodexReasoningSummary. An explicit reasoning object from a Responses-shaped client is preserved untouched, and reasoning_effort "none" is left without a summary. Adds regression tests for the effort-only default, the none case, and keeps the existing explicit-reasoning-object behavior unchanged. Co-authored-by: Diego Rodrigues de Sa e Souza <8016841+diegosouzapw@users.noreply.github.com> |
||
|
Diego Rodrigues de Sa e Souza
|
112b1499df |
fix(antigravity): sanitize Cloud Code safety settings (#6839)
Co-authored-by: kfiramar <83420275+kfiramar@users.noreply.github.com> |
||
|
Diego Rodrigues de Sa e Souza
|
039acabad9 |
fix(providers): Kiro adaptive-thinking allowlist excludes sonnet-4.5/haiku-4.5 (#6576) (#6726)
* fix(providers): Kiro adaptive-thinking allowlist excludes sonnet-4.5/haiku-4.5 (#6576) * chore(6726): re-sync onto release tip; CHANGELOG entry → changelog.d fragment (fragments-first) * test(kiro): migrate selector-strip test to claude-sonnet-5 (only Kiro adaptive-thinking model, #6576) Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com> |
||
|
Diego Rodrigues de Sa e Souza
|
4a9e36616b |
fix(sse): skip thinkingConfig for gemma models in openai→gemini translation (#6708)
open-sse/translator/request/claude-to-gemini.ts already guards against
sending thinkingConfig for gemma-4-* models (Gemma doesn't support it —
Vertex returns 400: "Thinking budget is not supported for this model"),
but the OpenAI-shape path (openai-to-gemini.ts) lacked the same guard, so
OpenAI-shape clients hitting a vertex gemma-4-* model still got a 400.
Mirrors the existing claude-to-gemini.ts guard: wrap the reasoning_effort
and Claude-shape thinking.budget_tokens branches with a model.startsWith
("gemma-4") check. Branch 3 (default includeThoughts for modern Gemini
models) already excludes non-"gemini" model ids and needed no change.
Inspired-by: https://github.com/decolua/9router/pull/2480
Co-authored-by: chy1211 <31048289+chy1211@users.noreply.github.com>
|
||
|
Diego Rodrigues de Sa e Souza
|
b45d10ceea |
fix(sse): unwrap bare {function:{…}} tools in openai→claude translation (#6704)
* fix(sse): unwrap bare {function:{…}} tools in openai→claude translation
Some OpenAI-shape clients send a tool as a bare `{ function: {...} }`
object, omitting the spec-required `type: "function"` parent wrapper.
The tools-mapping in openai-to-claude.ts (~line 366) only unwrapped
`tool.function` when `tool.type === "function"` was ALSO true, so a
bare-function tool fell through to `toolData = tool` (the wrapper
itself, with no `.name`), producing an empty `originalName` and
silently dropping the tool from the translated request — worse than
a 400, since the caller has no signal the tool never made it
upstream. Unwrap `tool.function` whenever present, independent of
the parent `type` field. Regression guard:
tests/unit/openai-to-claude-bare-tool.test.ts.
Co-authored-by: Samir Abis <me@samirabis.com>
Inspired-by: https://github.com/decolua/9router/pull/2473
* chore(6704): re-sync onto release tip; CHANGELOG → changelog.d fragment (fragments-first)
---------
Co-authored-by: Samir Abis <me@samirabis.com>
|
||
|
WITALO ROCHA
|
92d9870507 |
fix(translator): read PDF/video file attachments for Gemini/Antigravity and Claude (#6790)
Reconstructed onto release/v3.8.47 to drop unrelated main-drift (deps/electron/proxy files belong to #6620, not this PR); keeps only the author's translator + test changes. Co-authored-by: Wital <witalorocha216@gmail.com> Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com> |
||
|
Ronaldo Davi
|
c837de6c98 |
fix(providers): honor explicit thinking.budget_tokens 0 in openai->gemini transform (#6813) (#6821)
The transform forwarded the Claude-style thinking.budget_tokens into generationConfig.thinkingConfig.thinkingBudget, but the presence check was truthy (&& thinking.budget_tokens). An explicit budget_tokens: 0 — the natural way to disable thinking — is falsy, so it was dropped and the request fell through to the default thinkingConfig injection, making the model think despite an explicit request for zero. Use an explicit numeric check so 0 is honored as thinkingBudget 0; includeThoughts is only set for a non-zero budget. |
||
|
Thiago Reis
|
3a28b3b5e8 |
feat: add Kiro API key authentication (#6587)
* feat(oauth): add Kiro long-lived API key auth (#6587) New /api/oauth/kiro/api-key route + KiroService.validateApiKey let a Kiro account be linked with a long-lived AWS CodeWhisperer/Kiro API key instead of the interactive OAuth device flow, with live per-account model discovery (ListAvailableModels, 5-minute cache) layered over the existing static registry fallback. Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com> * fix(changelog): re-restore #6587 bullet after release sync Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com> * fix(merge): restore #6126 clinepass files reverted by release auto-resolve + baseline re-merge The release sync's auto-resolve reverted sibling PR #6126's clinepass work (registry, catalog, oauth constants, clineAuth.ts, token-refresh case, tests) and the file-size baseline — all outside this PR's scope. Restored to the release versions, re-applied only this PR's own baseline entries, restored the #6126 CHANGELOG bullet (re-inserting only this PR's own). Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com> * chore(quality): freeze public-creds FP — AWS region default in validateApiKey signature Same class as the existing minimax fn-param FPs: CRED_KEY_RE matches the apiKey: param annotation and captures the region default "us-east-1", which is not a credential. Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com> * fix(kiro): keep hard-failure reject semantics + kill public-creds fn-param FP at the source - getKiroUsage: exhausted non-auth attempts now REJECT with the last HTTP-status failure in the pre-#6587 format (usage-service-hardening relies on it); auth failures keep the soft social-auth message. - validateApiKey: region default moved out of the parameter list (the check-public-creds CRED_KEY_RE matches the apiKey: annotation and flags any literal in the signature); drops the brittle line-keyed allowlist entry. Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com> --------- Co-authored-by: strangersp <strangersp@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> |
||
|
Diego Rodrigues de Sa e Souza
|
2ecaae7c40 |
fix(translator): preserve co-located functionResponse parts in gemini→openai (#6376)
preserve co-located functionResponse parts (port PR #2394) (net +1/-0, test OK). Integrated into release/v3.8.46. |
||
|
VXNCXNX
|
dc7eeba717 |
feat(sse): surface Kiro adaptive-thinking reasoning as reasoning_content (#6213)
Kiro/CodeWhisperer streams Claude's reasoning as native `reasoningContentEvent`
frames when adaptive thinking is enabled, but the Kiro executor had no handler
for them, so `reasoning_effort` requests returned no reasoning. Wire it end to
end:
- translator (openai-to-kiro): enable Kiro thinking when the request carries
`reasoning_effort`, Anthropic `output_config.effort`, or a `thinking` block
(`{type:"enabled",budget_tokens}` mapped to a level; `{type:"adaptive"}`
defaults to `high`, matching Anthropic's documented default). Prepends the
Kiro `<thinking_mode>`/`<max_thinking_length>` prompt directive and sets
top-level `additionalModelRequestFields` ({output_config.effort,
thinking:{type:"adaptive"}, max_tokens}). Gated on `supportsReasoning`; drops
non-default temperature/top_p (rejected by adaptive-only Claude models).
- executor transformRequest: forward `additionalModelRequestFields` to AWS
(previously dropped by the strict top-level allowlist).
- executor stream loop: parse `reasoningContentEvent` (and reasoningText
variants) into the OpenAI reasoning_content channel.
Verified against the live CodeWhisperer stream: reasoningContentEvent frames are
returned, and larger effort/budget measurably deepens reasoning up to the model
cap. Unit tests cover the effort sources, forwarding, temp/top_p stripping, and
native reasoning-frame parsing.
|
||
|
Diego Rodrigues de Sa e Souza
|
b3a2cfe0ea |
fix(providers): correct Kiro model catalog to real upstream ids (#6170)
* fix(providers): correct Kiro model catalog to real upstream ids
Kiro's API (generateAssistantResponse) returns 400 "Invalid model. Please
select a different model" for any id it does not recognize. The registry
exposed fabricated ids (copied from OmniRoute's own Anthropic catalog) that
Kiro never serves, so every call to them 400'd. Live-verified on the VPS:
Removed (400 Invalid model):
- auto-kiro (no "auto" model id — was sent verbatim upstream)
- claude-fable-5 (Kiro offers no Fable)
- claude-opus-4.8/4.7/4.6 (Kiro offers no Opus)
Corrected:
- claude-sonnet-4.6 -> claude-sonnet-4.5 (Kiro's Sonnet is 4.5; 4.5 -> 200)
Kept:
- claude-sonnet-5 (real Kiro model, plan-gated per account)
- claude-haiku-4.5, deepseek-3.2, glm-5, minimax-m2.5/m2.1,
qwen3-coder-next (all proven 200 on the VPS)
Aligns the free-model catalog and drops the orphaned auto-kiro price key.
Regression guard: tests/unit/kiro-catalog-real-models.test.ts (3/3).
Kiro cluster #6112/#6113/#6099.
* test(providers): align stale Kiro-catalog tests to the corrected upstream ids
The fabricated Kiro ids removed in the parent commit (claude-fable-5,
claude-opus-4.8/4.7/4.6, claude-sonnet-4.6) were still asserted as present by
three pre-existing tests, which encoded the bug:
- catalog-updates-v3x: now asserts Kiro does NOT expose Fable 5 / Opus (kept the
legit cc exposure) and guards the real claude-sonnet-4.5 pricing.
- model-family-fallback-notation: the dot-notation example moves from kiro/ to
anthropic/ (which genuinely serves Opus/Fable in dot notation) — coverage kept.
- provider-models-route: the Kiro local-catalog assertion now expects the real
Sonnet 5 / Sonnet 4.5 set and negatively guards the fabricated ids.
Co-authored-by: diegosouzapw <diegosouza.pw@gmail.com>
|
||
|
Diego Rodrigues de Sa e Souza
|
1bda6c15dc |
Release v3.8.44 (#5925)
* fix(install): add pnpm-workspace.yaml allowBuilds + pnpm.json for pnpm 11+ pnpm 11 introduced ERR_PNPM_IGNORED_BUILDS for native addon packages. Without explicit allowBuilds approval, these packages silently skip build scripts and OmniRoute fails to start with missing native modules. Changes: - pnpm-workspace.yaml: Set allowBuilds=true for all 13 native addon packages (@parcel/watcher, @swc/core, better-sqlite3, core-js, esbuild, keytar, koffi, libxmljs2, onnxruntime-node, protobufjs, sharp, tls-client-node, unrs-resolver) - pnpm.json: Migrate onlyBuiltDependencies from package.json (deprecated field) to the new pnpm.json config file per pnpm 11 spec. Tested on: pnpm 11.9.0, Node 24, Windows 11. Fixes: pnpm install ERR_PNPM_IGNORED_BUILDS on fresh clone with pnpm 11. * chore(release): open v3.8.44 development cycle * test(security): parse Kimi Web URL host instead of substring match (CodeQL #689) (#5928) Alert js/incomplete-url-substring-sanitization: the Kimi Web executor test asserted result.url.includes("www.kimi.com"), which a hostile host like www.kimi.com.evil.net would also satisfy. Parse the URL and assert on the exact hostname (new URL(result.url).hostname === "www.kimi.com"), which is both a stronger check and clears the CodeQL warning. * refactor(translator): extract thinking-budget fitting from openai-to-claude (#5932) Extract the thinking-budget fitting cluster (fitThinkingToMaxTokens + private safeCapMaxOutputTokens + MIN_* constants) verbatim into the pure leaf openai-to-claude/thinkingBudget.ts. Host re-exports fitThinkingToMaxTokens so external importers keep working and imports it back for internal use. Host 822 -> 738 LOC (under the 800 cap). No behavior change: byte-identical bodies, public export set unchanged. Adds a split-guard test; all consumer tests stay green (translator-openai-to-claude, strip-empty, minimax-m3, passthrough). * chore(release): pipeline hardening — test-masking pre-flight gate + contributors/uncovered helpers (#5926) * chore(ci): add test-masking PR-context gate to release-green pre-flight Reproduce check:test-masking (vs origin/main) inside validate-release-green so non-allowlisted net-assert reductions surface in the local pre-flight instead of in a ~40-min CI layer on the release PR. run() now merges a per-gate opts.env so GITHUB_BASE_REF reaches the child. HARD gate; skipped under --quick. Context: v3.8.43 release cost 3 CI round-trips for PR-context gates (test-masking, file-size, pr-evidence) that check:release-green did not reproduce locally. * chore(release): add contributors generator + uncovered-commit reconciliation helpers - scripts/release/gen-contributors.mjs: reproducible `### 🙌 Contributors` table for a CHANGELOG version (parenthetical-group parser → accurate per-PR attribution, noise-handle denylist). v3.8.43 shipped without the section (a real miss) because it was hand-built. npm run release:contributors <version> [--inject]. - scripts/release/list-uncovered-commits.mjs: lists commits since the last tag with no CHANGELOG bullet (v3.8.43 had 123/176 uncovered at reconciliation start). Advisory, maintainer-side. npm run release:uncovered. - 20 unit tests (parenthetical attribution, noise exclusion, idempotent injection, coverage window). * chore(quality): absorb web-cookie-providers-new file-size drift from #5928 (base-red on release/v3.8.44) * refactor(translator): split openai-responses request translator into pure leaves (#5940) Extract the shared pure primitives and the chat->Responses direction out of the 894-line openai-responses.ts request translator: - openai-responses/helpers.ts: pure primitives (toRecord/toString/clampCallId/ normalizeVerbosity/etc + markers/regexes/JsonRecord), zero host imports - openai-responses/toResponses.ts: openaiToOpenAIResponsesRequest (chat->Responses), imports the helpers leaf Host keeps openaiResponsesToOpenAIRequest (Responses->chat, imported by production) plus both register() directions, and re-exports openaiToOpenAIResponsesRequest so external importers (tests) keep working. Host 894 -> 529 LOC (under the 800 cap). Verbatim bodies (multiset check: leaf A 54/54, leaf B 294 lines, fn1 intact), public export set unchanged, leaves never import the host (no cycle). Adds a split-guard test; all consumer tests stay green (responses-translation-fixes 37, verbosity 4, reasoning-effort 4, orphaned-tool-filter 8, empty-tool-name-loop 8, headroom-responses-format 3). * chore(ci): pr-evidence FAIL output tells you to push (body edit does not re-run the gate) (#5944) ci.yml ignores the 'edited' event, so adding the Evidence block to the PR body after a push does not re-run check:pr-evidence — you need another commit. The FAIL report now says so, at the exact place someone sees the red check. + 5 unit tests (classification + hint-on-fail / no-hint-on-pass). Decided against a separate edited-triggered workflow: pr-evidence is not a required check (no ruleset gates it; release PRs merge UNSTABLE, not BLOCKED), so the gap is cosmetic and the generate-release skill already puts Evidence in the body before the first push. * fix(providers): Perplexity Web emits real tool_calls in streaming mode (mirror chatgpt-web toolMode) (#5927) (#5937) Perplexity Web (Pro/Max) only converted <tool>{...}</tool> text into OpenAI tool_calls for non-streaming requests (hasTools && !stream). Streaming requests -- the default for agentic coding clients -- got the raw <tool> text as plain delta.content and never emitted a tool_calls SSE delta, so clients could not execute tools. Reuses the provider-agnostic buildToolModeResponse()/ toolCompletionToSseStream() helpers already shipped for chatgpt-web (#5240): when tools are requested, buffer the full completion and convert it into either a JSON completion or a terminal SSE replay carrying delta.tool_calls + finish_reason: tool_calls, regardless of the caller's stream flag. Extended buildToolModeResponse()'s idSeed to be caller-supplied (default 'cgpt', perplexity-web passes 'pplx') so tool_call ids stay provider-specific without duplicating the helper. Non-tool streaming is unchanged (still lives token-by-token via buildStreamingResponse). * fix(discovery): resolve duplicate /v1 paths and redirect aborts (#5904) Integrated into release/v3.8.44. Thanks @hamsa0x7 for diagnosing the doubled /v1 discovery path and the REDIRECT_BLOCKED probe-loop abort (#5899). De-scoped to the discovery fix (the #5903 session-affinity work is handled by #5943) and added Rule #18 regression guards. * docs(changelog): record #5926 + #5944 (release-pipeline hardening) under v3.8.44 Maintenance (#5952) * docs(claude): add Hard Rule #22 — cross-session safety (git stash + in-flight PRs) (#5955) Integrated into release/v3.8.44 — Hard Rule #22 (cross-session safety). * refactor(translator): extract pure helpers from response/openai-responses (#5949) Extract the 5 stateless helpers (normalizeToolName, stripEmptyOptionalToolArgs, normalizeOutputIndex, normalizeUpstreamFailure, extractResponsesReasoningSummaryText) verbatim into the pure leaf openai-responses/pureHelpers.ts (no stream state, no host import). Host imports them back and re-exports normalizeUpstreamFailure for external importers (tests). Host 1091 -> 1001 LOC. The stateful streaming core stays in the host (out of scope). Byte-identical bodies (multiset 73/73), no cycle. Adds a split-guard; consumer tests stay green (responses-translation-fixes 37, combo-param-validation-fallback-4519 5). * docs(compression): document upstream sync policy for RTK/Caveman engines (#5830) (#5948) Integrated into release/v3.8.44 — docs-only upstream sync policy for RTK/Caveman engines (closes #5830). All 7 checks green. * fix(sse): strip ANSI/VT100 codes from gemini-cli stream frames (#5934) Integrated into release/v3.8.44 — ReDoS-safe ANSI/VT100 strip for gemini-cli stream frames (port of upstream #2273, thanks @anki1kr). PR test green (5/5), file-size gate OK. * fix(translator): strict Anthropic content-block compliance in antigravity→openai request (#5935) Integrated into release/v3.8.44 — strict Anthropic content-block compliance in antigravity→openai (port upstream #2296). PR test green (9/9). UNSTABLE red is the pre-existing environmental setup-claude base-red (opencode-plugin dist not built in fast-path), not a regression from this PR. * fix(mcp): auto-recover stale streamable HTTP sessions on initialize (#5957) Integrated into release/v3.8.44 — MCP stale streamable-HTTP session auto-recovery (thanks @Chewji9875). * fix(providers): validate v0 Platform API keys via chats endpoint (#5954) Integrated into release/v3.8.44 — v0-vercel Platform API key validation (thanks @vittoroliveira-dev). * fix(api): relax provider-scoped chat completion validation (#5907) Integrated into release/v3.8.44 — relaxed provider-scoped chat validation + regression test (thanks @nickwizard). * fix(providers): strip /v1 unconditionally to avoid /v1/v1/models fetch error (#5899) (#5920) Integrated into release/v3.8.44 — unconditional /v1 strip in both models-discovery paths + regression test (thanks @anki1kr). * fix(resilience): per-window is_exhausted + honor quota-exhaustion preflight for priority combos (#5923) (#5941) Integrated into release/v3.8.44. * fix(resilience): honor active codex session affinity over per-request reset-aware re-scoring (#5903) (#5943) Integrated into release/v3.8.44. * fix(thinking): only inject redacted_thinking replay block when tool_use present and thinking enabled (#5945) (#5953) Integrated into release/v3.8.44. * feat(providers): add ClinePass API-key provider (#5942) Integrated into release/v3.8.44 — ClinePass API-key (BYOK) provider (port upstream 9router#2304, co-authored @adentdk). Validated locally: 16 clinepass tests green; fixed the APIKEY count 158→159 + translate-path golden snapshot (clinepass is a genuine new provider). Remaining UNSTABLE red is the pre-existing environmental setup-claude base-red (opencode-plugin dist not built in fast-path). Supersedes stub #5541. * feat(api): add /v1/ocr endpoint (Mistral OCR) + Mistral moderation (#5950) Integrated into release/v3.8.44 — /v1/ocr endpoint (Mistral OCR) + Mistral moderation (port upstream 9router#2064, co-authored @waguriagentic). Validated locally: 14 ocr-route tests + moderation/servicekind/endpoint-category suites green (CORS→Zod→handler + no-stack-leak assertion). Reds are inherited DRIFT only: cognitive-complexity ratchet (none from OCR files — pre-existing cycle drift, rebaselined at release) + environmental setup-claude base-red. * fix(codex): convert chat json schema to responses text format (#5933) Integrated into release/v3.8.44 — converts Chat Completions json_schema response_format → Responses API text.format on the Codex path, and preserves existing text.format through verbosity normalization. Base redirected main→release; the openai-responses.ts split that landed this cycle was reconciled by re-applying the delta onto openai-responses/toResponses.ts. Validated locally: 48 translator-openai-responses-req + 8 codex-verbosity tests green. Co-authored-by: diegosouzapw <diegosouza.pw@gmail.com> * feat(providers): add Claude Sonnet 5 support across the model pipeline (#5833) Integrated into release/v3.8.44 — wires claude-sonnet-5 end-to-end (registries, modelSpecs, pricing ×3, cost, Sonnet-family fallback, 1M-ctx, static models). Reconciled the add/add overlap with the already-merged #5796 (kept the PR's superset test with the family-fallback assertion). Validated locally: kiro-sonnet-5 + catalog + pricing/modelSpecs/fallback suites all green. Thanks @ggiak! Co-authored-by: diegosouzapw <diegosouza.pw@gmail.com> * feat(relay): gate bifrost auto routing by provider manifest (#5870) Integrated into release/v3.8.44 — gates Bifrost auto-routing by the provider plugin manifest (only manifest-eligible providers reach the sidecar; ineligible/unknown fall back to the TS path with explicit reasons). Superset of #5869 (carries the full manifest + registry + docs). Resolved an integration-test conflict in favor of the release (which already subsumes this PR's readiness/removeDirWithRetry improvements). Validated locally: 4 provider-plugin-manifest + 11 relay-routing-backend tests green. Thanks @KooshaPari! Co-authored-by: diegosouzapw <diegosouza.pw@gmail.com> * refactor(translator): extract pure message helpers from openai-to-kiro (852→751) (#5947) * refactor(translator): extract pure message helpers from openai-to-kiro Extract the pure tool/message helpers (parseToolInput, normalizeKiroToolSchema, serializeToolResultContent) verbatim into the leaf openai-to-kiro/messageHelpers.ts. The host imports them back for convertMessages. They were module-private, so the public export set is unchanged (no re-export needed). Host 852 -> 751 LOC. Byte-identical bodies (multiset 99/99), leaf has zero imports (no cycle). Adds a split-guard; consumer tests stay green (translator-openai-to-kiro 33, translator-ai-sdk-image-parts 3). * chore: re-trigger CI (stuck runner on 2/2 shard) * refactor(executors): extract pure prompt + composer helpers from cursor (#5960) Extract two pure clusters from the cursor executor into sibling leaves: - cursor/prompt.ts: isRecordLike + toolChoiceDirectiveLine + buildCursorOutputConstraints - cursor/composer.ts: composer thinking-as-content decoding (isComposerModel, visibleComposerContentFromThinking, composerReasoningRemainder + markers) Host imports both back for internal use and re-exports the 3 composer helpers for external importers (tests). Host 1576 -> 1451 LOC. Byte-identical bodies (verbatim multiset prompt 65/65, composer 32/32), leaves have zero imports (no cycle). Adds a split-guard; consumer tests stay green (cursor-composer-thinking, cursor-streaming, cursor-agent-tool-calls, translator-openai-to-cursor, cursor-agent-system-prompt). * refactor(executors): extract pure SSE-collect parsing from antigravity (#5962) Extract the pure SSE-payload -> collected-stream parser (AntigravityCollectedStream, stripZeroWidth, parseAntigravityTextualToolCall, addAntigravityTextualToolCall, processAntigravitySSEPayload/Text, flushAntigravitySSEText) verbatim into the leaf antigravity/sseCollect.ts. Host imports the helpers it uses and re-exports processAntigravitySSEPayload for external importers (tests). Host 1812 -> 1671 LOC. Byte-identical bodies (verbatim multiset 135/135), leaf does not import the host (no cycle). Credit/quota state, auth, and HTTP dispatch untouched. Adds a split-guard; consumer tests stay green (executor-agy 8, executor-antigravity 26, antigravity-sse-collect-socket-release, copilot-agent-antigravity-parity 6). * refactor(executors): extract pure model maps + resolvers from chatgpt-web (#5967) Extract the static model maps (MODEL_MAP, MODEL_FORCED_EFFORT, THINKING_CAPABLE_SLUGS) and the pure thinking-effort resolvers (isThinkingCapableModel, normalizeThinkingEffort, resolveThinkingEffort, ResolvedChatGptModel, resolveChatGptModel) verbatim into the pure leaf chatgpt-web/models.ts. Host imports the two resolvers it uses back. Host 3205 -> 3076 LOC. Byte-identical bodies (verbatim multiset 120/120), leaf has zero imports (no cycle). Auth/PoW/session/HTTP dispatch and all module caches untouched. Adds a split-guard; consumer tests stay green (chatgpt-web 86, chatgpt-web-tools-5240 4, chatgpt-web-sha3-boringssl-5531 5). * refactor(executors): decompose grok-web into pure tool/markup leaves (#5994) Extract the pure OpenAI<->Grok tool-translation, native-tool mapping, markup cleanup, and NDJSON stream types out of the 1872-line grok-web executor into 4 sibling leaves: - grok-web/types.ts: GrokStreamResponse/GrokStreamEvent (stream types) - grok-web/tool-bridge.ts: OpenAI<->Grok tool translation + registry + classifiers - grok-web/native-tools.ts: native-tool selection/scoring + native->OpenAI mapping - grok-web/text-cleanup.ts: Grok markup stripping + GrokMarkupFilter Layered, acyclic: types <- tool-bridge <- native-tools; text-cleanup <- types; host imports the leaves. All symbols module-private (no host re-export). Host 1872 -> 887 LOC. Byte-identical bodies (verbatim per-leaf), no cycle, all new leaves <= 800 cap (tool-bridge split at line 753 to stay under). Auth/cookie/TLS/HTTP dispatch untouched. Adds a split-guard; consumer tests stay green (grok-web 62, grok-cli-oauth 15, grok-cli-strip-params 2). * refactor(executors): extract pure quota parsing from codex (#5999) Extract the pure Codex quota-snapshot parsing + reset/cooldown scheduling (CodexQuotaSnapshot, parseCodexQuotaHeaders, getCodexResetTime, getCodexDualWindowCooldownMs) verbatim into the leaf codex/quota.ts. Host re-exports the 4 symbols so handlers/chatCore/codexQuota.ts + tests keep resolving. Host 1539 -> 1427 LOC. Byte-identical bodies (verbatim 98/98), leaf has zero imports (only Date, no cycle). WS transport, auth, HTTP dispatch untouched. Adds a split-guard; consumer tests stay green (executor-codex 40, codex-quota-fetcher 7, chatcore-codex-quota 5). * refactor(executors): extract pure stream formatters from deepseek-web (#6000) Extract the pure content/citation formatters (isThinkingModel, isSearchModel, cleanDeepSeekToken, formatStreamContent, DeepSeekSearchResult, appendSearchCitations) verbatim into the leaf deepseek-web/stream-format.ts. Host imports the 5 it uses back into transformSSE/collectSSEContent (cleanDeepSeekToken stays internal to the leaf). Host 1147 -> 1108 LOC. Byte-identical bodies (verbatim 34/34), leaf has zero imports (no cycle), all module-private (no re-export). PoW/auth/token-cache/HTTP dispatch untouched. Adds a split-guard; consumer tests stay green (deepseek-web 35, deepseek-web-rolling-window-2942 5, deepseek-web-tools-execute 3). * refactor(api): add validatedJsonBody helper (salvage #5075) (#5931) Fuses JSON body parsing + Zod validation into a single call that returns either type-narrowed data or a ready-to-return 400 NextResponse with the standard error envelope. Salvaged as the Tier 1 portable helper from the closed refactor PR #5075; the bulk route migration is intentionally not ported. Adds a focused 6-case regression test. Co-authored-by: KooshaPari <KooshaPari@users.noreply.github.com> * feat(qoder): drive PAT auth via qodercli, add dashboard quota, fix connection display (#5816) Integrated into release/v3.8.44 — Qoder PAT auth via qodercli binary + dashboard quota + dual-auth connection fix. Thanks @AgentKiller45 (co-author @judy459)! Validated locally (release-green on its own merits): lint 0, typecheck:core 0, 104 qoder/usage/UI tests green, file-size gate OK (owner-approved qoderCli.ts baseline-freeze 666→989), env-doc-sync fixed (documented QODER_CLI_CONFIG_DIR). The 2 remaining CI reds are INHERITED base-reds, not caused by this PR: (1) LEDGER-4 minimax-m3 supportsVision (minimax-m3 base + cline-pass/minimax-m3 from the already-merged #5942); (2) mutation-test-coverage missing 3 tests in stryker.conf (#5903/#5942/#5923). Both cleaned up separately. Co-authored-by: diegosouzapw <diegosouza.pw@gmail.com> * fix(providers): minimax-m3 supportsVision (LEDGER-4) + stryker tap.testFiles drift (#6012) Release-green cleanup — clears LEDGER-4 minimax-m3 supportsVision + stryker tap.testFiles drift base-reds. Validated locally. * fix(registry): flag cline-pass/minimax-m3 as multimodal (supportsVision) (#6003) The cline-pass provider's minimax-m3 entry was missing supportsVision, breaking the LEDGER-4 registry-consistency test (all minimax-m3 entries must set supportsVision to match lite.ts — minimax-m3 is multimodal). Every other minimax-m3 registry entry (trae, bazaarlink, cline, ollama-cloud, ...) already sets it. This was a base-red on release/v3.8.44 inherited by every open PR. Validated by the existing failing-then-passing guard tests/unit/review-reviews-v3814-fixes.test.ts (LEDGER-4). * refactor(executors): extract pure payload construction from claude-web (#6006) Extract the pure Claude-web payload types + transforms + default tools/style (ClaudeWebRequestPayload, ClaudeWebStreamChunk, DEFAULT_CLAUDE_MODEL, generateMessageUUIDs, getDefaultTools, getDefaultPersonalizedStyle, transformToClaude, transformFromClaude) verbatim into the leaf claude-web/payload.ts. Host imports the 3 it uses back (ClaudeWebRequestPayload type + the two transforms). Host 1056 -> 835 LOC. Byte-identical bodies (verbatim 149/149), leaf imports only randomUUID (no host import, no cycle), all module-private (no re-export). Cookie/auth/ Turnstile/TLS/HTTP dispatch untouched. Adds a split-guard; consumer tests stay green (claude-web 13, claude-web-auto-refresh 6). * refactor(executors): extract pure upstream-header helpers from base (#6008) Extract the pure upstream-header helpers (mergeUpstreamExtraHeaders, getCustomUserAgent, setUserAgentHeader, applyConfiguredUserAgent, isOpenAICompatibleEndpoint, stripStainlessHeadersForOpenAICompat) verbatim into the leaf base/headers.ts. base.ts is imported by ~18 executors, so the host re-exports all 6 to keep those import paths intact; it also imports the 4 it uses internally in the BaseExecutor class. The trivial JsonRecord type alias is redefined locally in the leaf to avoid a base<->leaf cycle. Host 1539 -> 1451 LOC. Byte-identical bodies (verbatim 78/78), leaf does not import the host (no cycle). typecheck:core validates all base importers still resolve via the re-export. Adds a split-guard; consumer tests stay green (executor-base-utils 22, executor-default-base 49, executor-strip-stainless-openai-compat 6, plus executor sanity via typecheck). * refactor(executors): extract pure wire protocol from perplexity-web (#6014) Extract the pure Perplexity wire protocol (consts, SSE stream types, SSE parsing, OpenAI<->Perplexity message translation, request/query builders, content extraction, sseChunk) verbatim into the leaf perplexity-web/protocol.ts. Host imports back the 10 symbols it uses; everything module-private (no re-export). Session cache, TLS fetch, auth, and the executor class stay in the host. Host 1028 -> 534 LOC. Byte-identical bodies (verbatim), leaf imports only randomUUID (no host import, no cycle). Adds a split-guard; consumer tests stay green (perplexity-web 26, streaming-tools-5927 2, tls-client 6, key-validation-models 2). * refactor(executors): extract pure URL normalizers from default (#6015) Extract the pure per-provider chat-URL normalizers (normalizeBailianMessagesUrl, normalizeDataRobotChatUrl, normalizeAzureAiChatUrl, normalizeWatsonxChatUrl, normalizeOciChatUrl, normalizeSapChatUrl, normalizeXiaomiMimoChatUrl, normalizeOpenAIChatUrl, getOpenRouterConnectionPreset) verbatim into the leaf default/urlNormalizers.ts. Host imports them back into buildUrl/transformRequest; the now-dead build*ChatUrl/normalizeBaseUrl imports move to the leaf. All module-private (no re-export). Host 864 -> 815 LOC (shrunk below its frozen baseline). Byte-identical bodies (verbatim 45/45), leaf does not import the host (no cycle). buildHeaders/execute/auth untouched. Adds a split-guard; consumer tests stay green (executor-default-base 49, anthropic-compatible-bearer 3, strip-client-metadata 3). * feat(webfetch): support self-hosted FireCrawl instances (#5793) Integrated into release/v3.8.44 — self-hosted FireCrawl support (FIRECRAWL_BASE_URL/FIRECRAWL_TIMEOUT_MS). Re-cut clean onto the release tip (branch was fossilized from a pre-v3.8.40 snapshot). Validated: 4 firecrawl tests green, env-doc-sync + docs-sync pass. UNSTABLE red is the inherited environmental setup-claude base-red. * feat(xai): register XaiExecutor with reasoning-effort suffix parsing (#5800) Integrated into release/v3.8.44 — XaiExecutor with reasoning-effort suffix parsing. Re-cut clean onto the release tip (branch was fossilized). Validated: 6 xai-executor tests green, provider-consistency OK, typecheck:core 0 errors, env-doc-sync in sync. UNSTABLE red is the inherited environmental setup-claude base-red. * feat(discovery): Phase 2 — reporter, /api/discovery/* routes (strict loopback-only) + dashboard UI (#5939) * feat(discovery): Phase 2 reporter — discoveryResults DB module + service wiring Adds src/lib/db/discoveryResults.ts (CRUD over the discovery_results table from migration 074) and wires the opt-in discovery service to persist and read findings through it: persistDiscoveryResult / getDiscoveryResults / getDiscoveryResultById / markVerified / deleteDiscoveryResult, with (provider, method, endpoint) upsert de-duplication. Re-exported from localDb. The service stays opt-in / default-off. The /api/discovery/* routes and the dashboard UI tab are intentionally deferred to Phase 2b — they need the local-only enforcement model (Hard Rules #15/#17 territory) decided first. TDD: tests/unit/db/discovery-results.test.ts (8 cases, DB + service delegation), isolated DATA_DIR with resetDbInstance cleanup. * feat(discovery): Phase 2b — /api/discovery/* routes (strict loopback-only) Adds the discovery HTTP surface on top of the reporter DB module: GET /api/discovery/results list findings (optional ?providerId) GET /api/discovery/results/:id one finding (404 if absent) DELETE /api/discovery/results/:id delete a finding POST /api/discovery/scan scan a provider + persist findings POST /api/discovery/verify/:id mark a finding verified Authorization: strict loopback-only. "/api/discovery/" is added to LOCAL_ONLY_API_PREFIXES so the central authz pipeline (proxy.ts → runAuthzPipeline → managementPolicy) rejects non-loopback callers with a 403 LOCAL_ONLY before any handler runs. It is deliberately NOT in LOCAL_ONLY_MANAGE_SCOPE_BYPASS_PREFIXES — no remote manage-scope bypass — because POST /scan issues outbound probes to provider endpoints (SSRF-adjacent) and must never be tunnel-reachable. Handlers also call requireManagementAuth (defense in depth) and return sanitized errors via createErrorResponse. Tests: - tests/unit/authz/discovery-routes-local-only.test.ts (8) — security guard: isLocalOnlyPath true + not manage-scope-bypassable for all four paths. - tests/unit/api/discovery-routes.test.ts (6) — handler integration over an isolated DATA_DIR: list/filter, by-id 200/404/400, scan persist + 400 on empty/malformed body, verify 200/404, delete 200/404, no stack-trace leak. * feat(discovery): Phase 2c — dashboard UI tab (Tools → Discovery) Adds the /dashboard/discovery page (DiscoveryPageClient) that consumes the Phase 2b /api/discovery/* routes: scan a provider, list findings, verify or delete them. Registered in the sidebar under the Tools group (icon travel_explore) and given a "discovery" i18n namespace + sidebar keys in en.json (other locales fall back to en via next-intl until synced — the locale files are in a pre-existing coverage deficit unrelated to this change). Registers the UI test path in vitest.config.ts (advisory ui suite). Tests: src/app/(dashboard)/dashboard/discovery/__tests__/DiscoveryPageClient.test.tsx (3 cases: loads+renders results, empty state, fetches /api/discovery/results on mount; stable useTranslations mock to avoid the fetch-loop). NOTE: the ui vitest suite cannot run in this workspace — @testing-library/dom (a @testing-library/ react peer dep) is absent from node_modules, which fails ALL existing ui tests equally; the test runs in CI. Component verified locally via typecheck + lint. * test(discovery): register discovery-routes-local-only in stryker tap.testFiles The mutation-test-coverage gate (--strict) flags any unit test covering a mutated module that isn't listed in stryker.conf.json tap.testFiles. This PR's tests/unit/authz/discovery-routes-local-only.test.ts covers src/server/authz/ routeGuard.ts (a mutated module, which this PR edits by adding the /api/discovery/ local-only prefix), so it must be registered for its mutant kills to count. No behavior change. * refactor(discovery): split DiscoveryPageClient to satisfy max-lines-per-function The complexity ratchet (max-lines-per-function: 80) flagged the single 184-line DiscoveryPageClient function (+1 over baseline). Extract the data layer into two hooks (useDiscoveryResults for list/loading/feedback, useDiscoveryActions for scan/verify/delete), a shared callApi helper, and two presentational sub-components (DiscoveryScanForm, DiscoveryResultCard). Every function is now under the 80-line ceiling; complexity gate back to baseline 1995. No behavior change — same exported component, same endpoints, same props. * test(sidebar): include discovery in omni-proxy item-order snapshot Adding the Discovery item to the Tools group (this PR's sidebar entry) extends the ordered omni-proxy section list. Update the exact-match deepEqual snapshot in sidebar-visibility.test.ts to include "discovery" in its position (after traffic-inspector). The assertion stays exact — this reflects the intentional new item, it does not weaken the check. * docs(changelog): restore release bullets eaten by merge auto-resolve; re-add discovery bullet additively * chore(quality): bump testFrozen for translator-openai-responses-req.test.ts (1097 -> 1172) Base-red inherited from #5933, which grew the test file to 1171 lines (Hard Rule #18 regression tests) without adjusting the frozen cap. The release tip itself fails check:file-size; this unblocks every PR into release/v3.8.44. File untouched by this PR. * chore(quality): restore stryker tap.testFiles entries eaten by merge auto-resolve The merge of origin/release/v3.8.44 silently dropped the 3 entries added on the release side (#5903, clinepass, #5923). Took the release version verbatim and re-added only this PR's entry (discovery-routes-local-only) in alphabetical order. check:mutation-test-coverage green locally. * chore(quality): reconcile inherited v3.8.44 merge-burst drift + include discovery in tools-group order test - complexity 1995->2003 and cognitive 856->859: both measure IDENTICAL on the pristine release tip ( |