Merge branch 'release/v3.8.51' into fix/claude-oauth-sticky-refresh

2026-09-19 13:23:50 +03:00 · 2026-09-16 16:48:09 +02:00
parent 70b3777fdd 1cb4c82071
commit 3c623f1645
469 changed files with 152513 additions and 139729 deletions
--- a/.env.example
+++ b/.env.example
@@ -2748,6 +2748,13 @@ APP_LOG_TO_FILE=true
 # tokens (accessToken / refreshToken / providerSpecificData). Default OFF —
 # only non-credential metadata is synced. See docs/security/SOCKET_DEV_FINDINGS.md §5.
 # OMNIROUTE_CLOUD_SYNC_SECRETS=false
+#
+# Set to "true" to reject an UNSIGNED Cloud sync response when no local secret
+# is configured (#13679). Default OFF keeps v3.8.x back-compat for peers that
+# have not rotated in a shared secret yet; v3.9 flips the default to enforced.
+# A signature that IS present is always verified, and always rejected when
+# OMNIROUTE_CLOUD_SYNC_SECRET is unset, regardless of this flag.
+# OMNIROUTE_CLOUD_SYNC_ENFORCE_SIGNATURE=false

 # ─── Zed import legacy compat (v3.8.6) ──────────────────────────────────────
 # Set to "true" to fall back to the v3.8.5 one-step "import everything from
@@ -3043,6 +3050,13 @@ QUOTA_STORE_DRIVER=sqlite
 # CHATGPT_WEB_CODEX_CHROME_PATH=/usr/bin/chromium
 # CHROME_PATH=/usr/bin/chromium
 # CHATGPT_WEB_CODEX_CDP_URL=http://chatgpt-web-codex-browser:9223
+# CDP_PROXY_TOKEN required by docker/chatgpt-web-codex-browser/cdp-proxy.mjs (#13679):
+# when set, every request to the CDP proxy sidecar must present it as an
+# `X-Omni-Cdp-Token` header. Left unset, the proxy keeps forwarding requests
+# unauthenticated (network isolation via docker-compose.yml's dedicated
+# `chatgpt-web-codex-net` is the default mitigation). Generate with:
+# `openssl rand -hex 32`
+# CDP_PROXY_TOKEN=
 # CHATGPT_WEB_CODEX_TUNNEL_ID=tunnel_0123456789abcdef0123456789abcdef
 # CHATGPT_WEB_CODEX_RUNTIME_KEY=
 # CHATGPT_WEB_CODEX_CONNECTOR_NAME=OmniRoute Codex v2
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -511,9 +511,11 @@ jobs:
      - run: node scripts/i18n/check-ui-keys-coverage.mjs --threshold=65
      # Real-translation ratchet: a leaf copied verbatim from en.json passes key
      # parity above but is still English to the user (es shipped 55% English).
-      # Advisory in PR-0; flipped to blocking once the backlog is retranslated (PR-4).
-      - name: i18n real-translation ratio (advisory)
-        run: node scripts/i18n/check-translation-ratio.mjs --warn
+      # Blocking since PR-4 retranslated the verbatim-English backlog: the share of
+      # untranslated leaves per locale may only fall (ratchet baseline in
+      # config/quality/i18n-translation-baseline.json; `npm run i18n:check-ratio:update`).
+      - name: i18n real-translation ratio
+        run: node scripts/i18n/check-translation-ratio.mjs
      # #8463: a rewritten English value used to leave its 39 translations behind
      # silently (googleOAuthWarning shipped wrong copy in 39 locales for months).
      # Key parity above cannot see it — a stale translation counts as covered.
@@ -531,6 +533,15 @@ jobs:
        env:
          BASE_REF: ${{ github.base_ref && format('origin/{0}', github.base_ref) || '' }}
        run: node scripts/i18n/check-new-key-coverage.mjs
+      # Absolute complement of the two gates above: every locale must carry exactly the key
+      # set of en.json, whatever the age of the key. A locale batch is generated from the
+      # en.json of the day the branch is cut and translates for days while the base keeps
+      # adding keys — the batch PR adds no key itself, so the new-key gate stays silent and
+      # 43 absent keys out of ~13,000 still read 99.7 % coverage. Incident 2026-09-15:
+      # batch 1 (#13044) landed 43 keys short in nine locales, batch 2 (#13660) 10 keys short
+      # in eight. Fix is `sync-ui-keys --locale=<codes> --translate-markers`.
+      - name: i18n key completeness (every locale carries every en.json key)
+        run: node scripts/i18n/check-key-completeness.mjs

  # #8038: cheap glossary/protected-terms consistency gate —
  # complements i18n-ui-coverage (key parity) and the ICU `i18n` job below
--- a/.github/workflows/release-acceptance.yml
+++ b/.github/workflows/release-acceptance.yml
@@ -0,0 +1,42 @@
+name: Release acceptance
+
+on:
+  push:
+    branches: ["release/v*"]
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+concurrency:
+  group: release-acceptance-${{ github.ref }}
+  cancel-in-progress: false
+
+jobs:
+  acceptance:
+    name: Release acceptance
+    if: github.event_name != 'pull_request'
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v5
+        with:
+          persist-credentials: false
+      - uses: actions/setup-node@v5
+        with:
+          node-version: "22"
+          cache: npm
+      - run: npm ci
+      - name: Emit shadow acceptance report
+        run: |
+          node scripts/quality/validate-release-acceptance.mjs \
+            --plan tests/fixtures/release-acceptance/plan-lint.json \
+            --manifests tests/fixtures/release-acceptance/shadow-manifests \
+            --out release-acceptance-report.json
+        continue-on-error: true
+      - uses: actions/upload-artifact@v4
+        if: always()
+        with:
+          name: release-acceptance-report
+          path: release-acceptance-report.json
+          if-no-files-found: ignore
+          retention-days: 30
--- a/.gitleaks.toml
+++ b/.gitleaks.toml
@@ -97,4 +97,11 @@
      # credential; the generic-api-key rule flags the long hyphenated string.
      '''omniroute-cheaperinference-sponsor-banner-dismissed-v\d+''',
      '''SunbreakWebUI1''',
+      # Uzbek dashboard catalog (#13727, src/i18n/messages/uz.json `outputTokenDesc`):
+      # "Yakunlash/javob tokenlari" = "completion/response tokens". The rule reads the
+      # `...TokenDesc` key as a token assignment and the translated words as its value.
+      '''Yakunlash/javob''',
+      # Feature-flag id from #13439 (src/shared/constants/featureFlagDefinitions.ts):
+      # `key: "PROTECTED_PRIORITY_INFRA_502_ENABLED"` is a flag name, not a credential.
+      '''PROTECTED_PRIORITY_INFRA_502_ENABLED''',
    ]
--- a/@omniroute/opencode-plugin/package.json
+++ b/@omniroute/opencode-plugin/package.json
@@ -23,7 +23,7 @@
  "scripts": {
    "build": "tsup",
    "clean": "rm -rf dist",
-    "test": "node --import tsx/esm --test tests/scaffold.test.ts tests/auth.test.ts tests/options-schema.test.ts tests/multi-instance.test.ts tests/fetch-interceptor.test.ts tests/telemetry.test.ts tests/provider.test.ts tests/gemini-sanitize.test.ts tests/combos.test.ts tests/config-shim.test.ts tests/features.test.ts tests/feature-defaults.test.ts tests/usable-combo.test.ts tests/disk-snapshot-perms.test.ts tests/fork-features.test.ts tests/auto-combo-context.test.ts tests/provider-id-routing.test.ts tests/management-read-token.test.ts tests/auto-sync.test.ts tests/model-allowlist.test.ts tests/log-level.test.ts tests/effort-tier-variants.test.ts tests/naming.test.ts tests/free-budget-magnitude.test.ts tests/models-fetcher.test.ts",
+    "test": "node --import tsx/esm --test tests/scaffold.test.ts tests/auth.test.ts tests/options-schema.test.ts tests/multi-instance.test.ts tests/fetch-interceptor.test.ts tests/telemetry.test.ts tests/provider.test.ts tests/gemini-sanitize.test.ts tests/combos.test.ts tests/config-shim.test.ts tests/features.test.ts tests/feature-defaults.test.ts tests/usable-combo.test.ts tests/disk-snapshot-perms.test.ts tests/fork-features.test.ts tests/auto-combo-context.test.ts tests/provider-id-routing.test.ts tests/management-read-token.test.ts tests/auto-sync.test.ts tests/model-allowlist.test.ts tests/log-level.test.ts tests/effort-tier-variants.test.ts tests/naming.test.ts tests/free-budget-magnitude.test.ts tests/models-fetcher.test.ts tests/issue-13000-cold-start-combo-limit.test.ts",
    "prepublishOnly": "npm run clean && npm run build && npm test"
  },
  "keywords": [
--- a/@omniroute/opencode-plugin/src/index.ts
+++ b/@omniroute/opencode-plugin/src/index.ts
@@ -4631,9 +4631,21 @@ export function buildStaticProviderEntry(
          .map((m) => m.max_output_tokens)
          .filter((v): v is number => typeof v === "number" && v > 0);

-        if (contextValues.length > 0 && outputValues.length > 0) {
+        // Prefer the server-computed aggregate (accounts for explicit
+        // context_length overrides and members outside memberEntries, e.g.
+        // not yet resolved in /v1/models) over the raw Math.min(member)
+        // lower bound. Mirrors mapComboToModelV2's limit.context logic
+        // (#13000) so the static catalog and the dynamic hook agree.
+        const preferredContext =
+          typeof combo.computed_context_length === "number" && combo.computed_context_length > 0
+            ? combo.computed_context_length
+            : contextValues.length > 0
+              ? Math.min(...contextValues)
+              : undefined;
+
+        if (preferredContext !== undefined && outputValues.length > 0) {
          entry.limit = {
-            context: Math.min(...contextValues),
+            context: preferredContext,
            output: Math.min(...outputValues),
          };
        }
@@ -5511,6 +5523,32 @@ export function createOmniRouteConfigHook(

        const modelsFetchOk = !modelsFetchThrew && localRawModels.length > 0;

+        // Snapshot backfill for computed_context_length: a live /api/combos
+        // response can come back without this field (server hasn't finished
+        // recomputing it yet, e.g. just after a restart) even though the
+        // combo's members and identity are otherwise unchanged. When that
+        // happens, prefer the last-known-good value from the warm disk
+        // snapshot over the Math.min(member) fallback in
+        // mapComboToModelV2() — never overwrite any other combo field
+        // (models/name/etc.) with stale data, only this one derived number.
+        if (warmSnapshot) {
+          const snapshotComboById = new Map(warmSnapshot.rawCombos.map((c) => [c.id, c]));
+          for (const combo of localRawCombos) {
+            const hasLive =
+              typeof combo.computed_context_length === "number" &&
+              combo.computed_context_length > 0;
+            if (hasLive) continue;
+            const stale = snapshotComboById.get(combo.id);
+            if (
+              stale &&
+              typeof stale.computed_context_length === "number" &&
+              stale.computed_context_length > 0
+            ) {
+              combo.computed_context_length = stale.computed_context_length;
+            }
+          }
+        }
+
        // Disk-cache fallback (cold first run, no warm snapshot): when the
        // live fetch returned no models AND features.diskCache !== false,
        // hydrate from the last-known-good snapshot so OC still surfaces a
--- a/@omniroute/opencode-plugin/tests/issue-13000-cold-start-combo-limit.test.ts
+++ b/@omniroute/opencode-plugin/tests/issue-13000-cold-start-combo-limit.test.ts
@@ -0,0 +1,221 @@
+/**
+ * Repro for #13000: combo context limits fall back to Math.min(member)
+ * instead of using computed_context_length after cold start — no disk
+ * snapshot fallback.
+ *
+ * Scenario (mirrors the report): a warm disk snapshot holds the combo with
+ * its correct server-computed `computed_context_length` (245000, from all 6
+ * members). After a restart, the live refresh's combos fetch returns the
+ * SAME combo but without `computed_context_length` (e.g. the value hasn't
+ * propagated yet), and the live models fetch only resolves 2 of the 6
+ * members (the rest not yet in /v1/models). The background refresh then
+ * republishes the provider block built from this degraded live data,
+ * downgrading a previously-known-good 245000 limit to Math.min(163840,
+ * 1_000_000) = 163840 — exactly the member-minimum described in the issue.
+ */
+
+import test from "node:test";
+import assert from "node:assert/strict";
+import type { Config } from "@opencode-ai/plugin";
+
+import {
+  createOmniRouteConfigHook,
+  _resetInflightRefresh,
+  type OmniRouteAutoCombosFetcher,
+  type OmniRouteCombosFetcher,
+  type OmniRouteCompressionMetaFetcher,
+  type OmniRouteEnrichmentFetcher,
+  type OmniRouteFetchCache,
+  type OmniRouteModelsFetcher,
+  type OmniRouteProvidersFetcher,
+  type OmniRouteRawCombo,
+  type OmniRouteRawModelEntry,
+  type OmniRouteReadAuthJson,
+  type OmniRouteStaticProviderEntry,
+  type OmniRouteDiskSnapshotReader,
+  type OmniRouteDiskSnapshotWriter,
+} from "../src/index.js";
+
+test.beforeEach(() => {
+  _resetInflightRefresh();
+});
+
+function stubReadAuthJson(value: Record<string, unknown>): OmniRouteReadAuthJson {
+  return async () => value as never;
+}
+
+function authStub() {
+  return stubReadAuthJson({
+    "opencode-omniroute": {
+      type: "api",
+      key: "sk-test",
+      baseURL: "https://or.example.com/v1",
+    },
+  });
+}
+
+function makeInput(): Config {
+  return { provider: {} } as unknown as Config;
+}
+
+// The two members resolvable in the degraded live /v1/models response.
+const MEMBER_DEEPSEEK: OmniRouteRawModelEntry = {
+  id: "deepseek-v4-pro",
+  capabilities: { tool_calling: true, reasoning: true, vision: false, thinking: false },
+  context_length: 163_840,
+  max_output_tokens: 64_000,
+  input_modalities: ["text"],
+  output_modalities: ["text"],
+};
+
+const MEMBER_GLM: OmniRouteRawModelEntry = {
+  id: "glm-5.2",
+  capabilities: { tool_calling: true, reasoning: true, vision: false, thinking: false },
+  context_length: 1_000_000,
+  max_output_tokens: 16_384,
+  input_modalities: ["text"],
+  output_modalities: ["text"],
+};
+
+// The other member that IS present once the server is fully warm.
+const MEMBER_GLM_53_HIGH: OmniRouteRawModelEntry = {
+  id: "GLM-5.3-high",
+  capabilities: { tool_calling: true, reasoning: true, vision: false, thinking: false },
+  context_length: 245_000,
+  max_output_tokens: 128_000,
+  input_modalities: ["text"],
+  output_modalities: ["text"],
+};
+
+const COMBO_MODELS: OmniRouteRawCombo["models"] = [
+  { kind: "model", model: "deepseek-v4-pro", weight: 25 },
+  { kind: "model", model: "glm-5.2", weight: 25 },
+  { kind: "model", model: "GLM-5.3-high", weight: 50 },
+];
+
+test("issue #13000: warm combo limit (245000) survives a degraded post-restart refresh instead of downgrading to Math.min(member)", async () => {
+  const warmSnapshot: Omit<import("../src/index.js").OmniRouteFetchCacheEntry, "expiresAt"> = {
+    rawModels: [MEMBER_DEEPSEEK, MEMBER_GLM, MEMBER_GLM_53_HIGH],
+    rawCombos: [
+      {
+        id: "orchestrator",
+        name: "orchestrator",
+        models: COMBO_MODELS,
+        computed_context_length: 245_000,
+      },
+    ],
+    rawAutoCombos: [],
+    rawEnrichment: new Map(),
+    rawCompressionCombos: [],
+    rawConnections: [],
+  };
+
+  const fetcher: OmniRouteModelsFetcher = async () => [MEMBER_DEEPSEEK, MEMBER_GLM];
+  const combosFetcher: OmniRouteCombosFetcher = async () => [
+    {
+      id: "orchestrator",
+      name: "orchestrator",
+      models: COMBO_MODELS,
+      // computed_context_length intentionally omitted.
+    },
+  ];
+  const autoCombosFetcher: OmniRouteAutoCombosFetcher = async () => [];
+  const enrichmentFetcher: OmniRouteEnrichmentFetcher = async () => new Map();
+  const compressionMetaFetcher: OmniRouteCompressionMetaFetcher = async () => [];
+  const providersFetcher: OmniRouteProvidersFetcher = async () => [];
+
+  const diskSnapshotReader: OmniRouteDiskSnapshotReader = async () => warmSnapshot;
+  const diskSnapshotWriter: OmniRouteDiskSnapshotWriter = async () => {};
+
+  const sharedCache: OmniRouteFetchCache = new Map();
+
+  const hook = createOmniRouteConfigHook(
+    { providerId: "omniroute", modelCacheTtl: 60_000 },
+    {
+      readAuthJson: authStub(),
+      fetcher,
+      combosFetcher,
+      autoCombosFetcher,
+      enrichmentFetcher,
+      compressionMetaFetcher,
+      providersFetcher,
+      diskSnapshotReader,
+      diskSnapshotWriter,
+      cache: sharedCache,
+    }
+  );
+
+  const input = makeInput();
+  await hook(input);
+
+  // Let the detached background refresh (degraded live data) complete and
+  // republish the block.
+  await new Promise((r) => setTimeout(r, 100));
+
+  const entryAfter = (input as { provider: Record<string, OmniRouteStaticProviderEntry> }).provider[
+    "opencode-omniroute"
+  ];
+  const comboModelAfter = entryAfter.models["orchestrator"];
+  assert.ok(comboModelAfter, "combo model still published after refresh");
+
+  assert.equal(
+    comboModelAfter.limit.context,
+    245_000,
+    `expected the combo limit to stay at the known-good 245000, but got ${comboModelAfter.limit.context} ` +
+      `(Math.min(member) fallback — the exact bug described in #13000)`
+  );
+});
+
+test("issue #13000 (control): no warm snapshot exists — Math.min(member) fallback is still used (expected, documented behavior)", async () => {
+  const fetcher: OmniRouteModelsFetcher = async () => [MEMBER_DEEPSEEK, MEMBER_GLM];
+  const combosFetcher: OmniRouteCombosFetcher = async () => [
+    {
+      id: "orchestrator",
+      name: "orchestrator",
+      models: COMBO_MODELS,
+      // computed_context_length intentionally omitted.
+    },
+  ];
+  const autoCombosFetcher: OmniRouteAutoCombosFetcher = async () => [];
+  const enrichmentFetcher: OmniRouteEnrichmentFetcher = async () => new Map();
+  const compressionMetaFetcher: OmniRouteCompressionMetaFetcher = async () => [];
+  const providersFetcher: OmniRouteProvidersFetcher = async () => [];
+
+  // No prior snapshot on disk.
+  const diskSnapshotReader: OmniRouteDiskSnapshotReader = async () => undefined;
+  const diskSnapshotWriter: OmniRouteDiskSnapshotWriter = async () => {};
+
+  const sharedCache: OmniRouteFetchCache = new Map();
+
+  const hook = createOmniRouteConfigHook(
+    { providerId: "omniroute", modelCacheTtl: 60_000 },
+    {
+      readAuthJson: authStub(),
+      fetcher,
+      combosFetcher,
+      autoCombosFetcher,
+      enrichmentFetcher,
+      compressionMetaFetcher,
+      providersFetcher,
+      diskSnapshotReader,
+      diskSnapshotWriter,
+      cache: sharedCache,
+    }
+  );
+
+  const input = makeInput();
+  await hook(input);
+
+  const entryAfter = (input as { provider: Record<string, OmniRouteStaticProviderEntry> }).provider[
+    "opencode-omniroute"
+  ];
+  const comboModelAfter = entryAfter.models["orchestrator"];
+  assert.ok(comboModelAfter, "combo model published on cold first run");
+
+  // No snapshot to backfill from — Math.min(163840, 1_000_000) = 163840.
+  assert.equal(
+    comboModelAfter.limit.context,
+    163_840,
+    "pure cold start with no snapshot must keep using the Math.min(member) fallback"
+  );
+});
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -46,7 +46,7 @@ Repository map and Reference Documentation sections below.

 ## Project at a Glance

-**OmniRoute** — unified AI proxy/router. One endpoint, 358 LLM providers, auto-fallback.
+**OmniRoute** — unified AI proxy/router. One endpoint, 359 LLM providers, auto-fallback.

 | Layer         | Location                | Purpose                                                                                                                                                                   |
 | ------------- | ----------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
@@ -56,7 +56,7 @@ Repository map and Reference Documentation sections below.
 | Translators   | `open-sse/translator/`  | Format conversion (OpenAI↔Claude↔Gemini)                                                                                                                                  |
 | Transformer   | `open-sse/transformer/` | Responses API ↔ Chat Completions                                                                                                                                          |
 | Services      | `open-sse/services/`    | Combo routing, rate limits, caching, etc                                                                                                                                  |
-| Database      | `src/lib/db/`           | SQLite domain modules (176 migrations)                                                                                                                                    |
+| Database      | `src/lib/db/`           | SQLite domain modules (177 migrations)                                                                                                                                    |
 | Domain/Policy | `src/domain/`           | Policy engine, cost rules, fallback logic                                                                                                                                 |
 | MCP Server    | `open-sse/mcp-server/`  | 110 tools (45 canonical + memory/skill/GitHub/pool/gamification/plugin/Notion/Obsidian/local-corpus/RTK modules), 3 transports (stdio / SSE / Streamable HTTP), 33 scopes |
 | A2A Server    | `src/lib/a2a/`          | JSON-RPC 2.0 agent protocol                                                                                                                                               |
--- a/README.md
+++ b/README.md
@@ -7,7 +7,7 @@

 # 🚀 OmniRoute — The Free AI Gateway

-<img src="./docs/diagrams/readme-hero.svg" width="100%" alt="OmniRoute — Never stop coding. Every AI tool → 358 providers — 150+ free — through one endpoint. Claude Code, Codex, Cursor, Cline, Copilot & Antigravity into FREE Claude / GPT / Gemini with auto-fallback. RTK + Caveman stacked compression saves 15–95% tokens (~89% avg) — never hit limits. 358 AI providers · 150+ free tiers · ~1.47B free tokens/mo · 19 routing strategies · $0 to start."/>
+<img src="./docs/diagrams/readme-hero.svg" width="100%" alt="OmniRoute — Never stop coding. Every AI tool → 359 providers — 150+ free — through one endpoint. Claude Code, Codex, Cursor, Cline, Copilot & Antigravity into FREE Claude / GPT / Gemini with auto-fallback. RTK + Caveman stacked compression saves 15–95% tokens (~89% avg) — never hit limits. 359 AI providers · 150+ free tiers · ~1.47B free tokens/mo · 19 routing strategies · $0 to start."/>

 </div>

@@ -17,9 +17,9 @@

 </div>

-> Stacking free tiers by hand is painful — dozens of SDKs, dozens of rate limits, and no idea how much you actually have. OmniRoute catalogs **443 free-tier entries across 34 recurring pool keys** and computes the token headline from the **16 pools with a published positive monthly budget plus five per-model Groq caps**, deduplicated by shared pool. Quotas that only open after a regional identity check (today: ModelScope) are shown apart, +~6M behind regional identity verification, and never summed into the headline. The result stays visible on the dashboard (`/dashboard/free-tiers`).
+> Stacking free tiers by hand is painful — dozens of SDKs, dozens of rate limits, and no idea how much you actually have. OmniRoute catalogs **446 free-tier entries across 34 recurring pool keys** and computes the token headline from the **16 pools with a published positive monthly budget plus five per-model Groq caps**, deduplicated by shared pool. Quotas that only open after a regional identity check (today: ModelScope) are shown apart, +~6M behind regional identity verification, and never summed into the headline. The result stays visible on the dashboard (`/dashboard/free-tiers`).

-<img src="./docs/diagrams/free-tier-budget.svg" width="100%" alt="OmniRoute free-tier budget card: ~1.47B free tokens per month steady, up to ~2.07B in the first month with signup credits, from 34 documented recurring pool keys covering 443 cataloged free-tier entries behind one endpoint. Honest pool-deduped math — each shared pool counted once, including 16 recurring pools with a published positive monthly token budget plus five per-model Groq caps; 13 providers are marked avoid in the terms-risk catalog so you decide. Budget bar includes Mistral 1B, Nara 210M, LLM7 150M, Groq 30M (five per-model caps) and smaller pools, plus first-month signup credits and permanently-free no-token-cap providers surfaced separately so they never inflate the headline. Live used/remaining on /dashboard/free-tiers."/>
+<img src="./docs/diagrams/free-tier-budget.svg" width="100%" alt="OmniRoute free-tier budget card: ~1.47B free tokens per month steady, up to ~2.07B in the first month with signup credits, from 34 documented recurring pool keys covering 446 cataloged free-tier entries behind one endpoint. Honest pool-deduped math — each shared pool counted once, including 16 recurring pools with a published positive monthly token budget plus five per-model Groq caps; 13 providers are marked avoid in the terms-risk catalog so you decide. Budget bar includes Mistral 1B, Nara 210M, LLM7 150M, Groq 30M (five per-model caps) and smaller pools, plus first-month signup credits and permanently-free no-token-cap providers surfaced separately so they never inflate the headline. Live used/remaining on /dashboard/free-tiers."/>

 > Animated summary of the live `/dashboard/free-tiers` page. Full methodology (pool dedupe, credit tiers, provider terms): **[docs/reference/FREE_TIERS.md](docs/reference/FREE_TIERS.md)**.
 >
@@ -233,7 +233,7 @@ curl http://localhost:20128/v1/chat/completions \

 </div>

-<img src="./docs/diagrams/promise-pillars.svg" width="100%" alt="The Promise — One endpoint and 358 providers. Automatic fallback keeps routing while another healthy target is available. Six pillars: resilient fallback across 358 providers · up to 95% token savings on eligible workloads · $0 to start with 150+ free tiers and 52 recurring/keyless free-forever providers · 36 CLI/agent integrations through one config · OpenAI, Claude, Gemini and Responses API compatibility at /v1 · production controls including circuit breakers, TLS stealth, MCP 110 tools, A2A, memory, guardrails, evals and 39,000+ static test declarations across 5,100+ tracked test files."/>
+<img src="./docs/diagrams/promise-pillars.svg" width="100%" alt="The Promise — One endpoint and 359 providers. Automatic fallback keeps routing while another healthy target is available. Six pillars: resilient fallback across 359 providers · up to 95% token savings on eligible workloads · $0 to start with 150+ free tiers and 53 recurring/keyless free-forever providers · 36 CLI/agent integrations through one config · OpenAI, Claude, Gemini and Responses API compatibility at /v1 · production controls including circuit breakers, TLS stealth, MCP 110 tools, A2A, memory, guardrails, evals and 39,000+ static test declarations across 5,100+ tracked test files."/>

 <br/>
 <br/>
@@ -486,7 +486,7 @@ All **19** strategies — mix & match per combo step:

 </div>

-<img src="./docs/diagrams/comparison-table.svg" width="100%" alt="What sets OmniRoute apart — a dated feature snapshot vs 9router, OpenRouter, CLIProxyAPI and LiteLLM across 13 capabilities. OmniRoute: 358 providers, 150+ free tiers built in, 19 routing strategies, 12-engine token compression, built-in MCP server with 110 tools, A2A agent protocol, persistent memory, guardrails, cloud agents, TLS fingerprint stealth, Desktop/Termux/PWA and 42 i18n UI locales. OmniRoute is MIT-licensed and self-hostable. Competitor capabilities and counts may change; see the linked methodology."/>
+<img src="./docs/diagrams/comparison-table.svg" width="100%" alt="What sets OmniRoute apart — a dated feature snapshot vs 9router, OpenRouter, CLIProxyAPI and LiteLLM across 13 capabilities. OmniRoute: 359 providers, 150+ free tiers built in, 19 routing strategies, 12-engine token compression, built-in MCP server with 110 tools, A2A agent protocol, persistent memory, guardrails, cloud agents, TLS fingerprint stealth, Desktop/Termux/PWA and 42 i18n UI locales. OmniRoute is MIT-licensed and self-hostable. Competitor capabilities and counts may change; see the linked methodology."/>

 <sub>📊 Full methodology &amp; per-feature detail vs 9router, OpenRouter, CLIProxyAPI &amp; LiteLLM → [`docs/comparison/OMNIROUTE_VS_ALTERNATIVES.md`](docs/comparison/OMNIROUTE_VS_ALTERNATIVES.md)</sub>

@@ -672,7 +672,7 @@ of your shell history. → [CLI Integrations](docs/guides/CLI-INTEGRATIONS.md)

 </div>

-> **352 registered providers** across the canonical chat, media, search, local, cloud-agent and system collections, including **152 carrying `hasFree: true` discovery metadata**. The chat model registry covers **229 providers / 2,554 distinct provider-model pairs / 1,283 raw model IDs**; the separate free-budget catalog has **443 per-model rows**, **34 recurring pools** and **52 recurring/keyless free-forever providers**. These are different denominators by design; definitions and pool-deduped calculations live in the [Provider Reference](docs/reference/PROVIDER_REFERENCE.md) and [Free Tiers](docs/reference/FREE_TIERS.md).
+> **352 registered providers** across the canonical chat, media, search, local, cloud-agent and system collections, including **152 carrying `hasFree: true` discovery metadata**. The chat model registry covers **229 providers / 2,554 distinct provider-model pairs / 1,283 raw model IDs**; the separate free-budget catalog has **443 per-model rows**, **34 recurring pools** and **53 recurring/keyless free-forever providers**. These are different denominators by design; definitions and pool-deduped calculations live in the [Provider Reference](docs/reference/PROVIDER_REFERENCE.md) and [Free Tiers](docs/reference/FREE_TIERS.md).

 <div align="center">

@@ -1268,7 +1268,7 @@ Métricas canônicas em 2026-08-24: **1.029 vídeos únicos** · **11.132.922 vi
  <tr><td nowrap><b>Runtime</b></td><td>Node.js 22.x / 24.x LTS — <code>&gt;=22.22.2 &lt;23 || &gt;=24.0.0 &lt;27</code></td></tr>
  <tr><td nowrap><b>Language</b></td><td>TypeScript 6.0 — <b>100% TypeScript</b> across <code>src/</code> and <code>open-sse/</code> (zero <code>any</code> in core since v2.0)</td></tr>
  <tr><td nowrap><b>Framework</b></td><td>Next.js 16 + React 19 + Tailwind CSS 4</td></tr>
-  <tr><td nowrap><b>Database</b></td><td>better-sqlite3 (SQLite, WAL journaling) + LowDB (JSON legacy) — 122 domain modules, 176 migrations</td></tr>
+  <tr><td nowrap><b>Database</b></td><td>better-sqlite3 (SQLite, WAL journaling) + LowDB (JSON legacy) — 122 domain modules, 177 migrations</td></tr>
  <tr><td nowrap><b>Memory</b></td><td>SQLite FTS5 full-text + int8-quantized vector embeddings, typed decay</td></tr>
  <tr><td nowrap><b>Schemas</b></td><td>Zod 4 — MCP tool I/O validation + API contracts</td></tr>
  <tr><td nowrap><b>Protocols</b></td><td>MCP (stdio / HTTP / SSE) + A2A v0.3 (JSON-RPC 2.0 + SSE)</td></tr>
@@ -1331,7 +1331,7 @@ Métricas canônicas em 2026-08-24: **1.029 vídeos únicos** · **11.132.922 vi
  <tr><td nowrap><b><a href="docs/architecture/RESILIENCE_GUIDE.md">Resilience Guide</a></b></td><td>Circuit breakers, cooldowns, queue, anti-thundering herd, TLS spoofing</td></tr>
  <tr><td nowrap><b><a href="docs/routing/AUTO-COMBO.md">Auto-Combo Engine</a></b></td><td>16-factor scoring, mode packs, self-healing</td></tr>
  <tr><td nowrap><b><a href="docs/ops/PROXY_GUIDE.md">Proxy Guide</a></b></td><td>3-level proxy system, 1proxy marketplace, registry CRUD</td></tr>
-  <tr><td nowrap><b><a href="docs/reference/FREE_TIERS.md">Free Tiers</a></b></td><td>Consolidated directory: 34 documented recurring pools / 443 cataloged free-tier entries</td></tr>
+  <tr><td nowrap><b><a href="docs/reference/FREE_TIERS.md">Free Tiers</a></b></td><td>Consolidated directory: 34 documented recurring pools / 446 cataloged free-tier entries</td></tr>
  <tr><td nowrap><b><a href="docs/guides/FEATURES.md">Features Gallery</a></b></td><td>Visual dashboard tour with screenshots</td></tr>
  <tr><td nowrap><b><a href="docs/architecture/CODEBASE_DOCUMENTATION.md">Codebase Documentation</a></b></td><td>Beginner-friendly codebase walkthrough</td></tr>
 </table>
--- a/bin/cli/commands/mcp.mjs
+++ b/bin/cli/commands/mcp.mjs
@@ -9,6 +9,8 @@ function truncate(v, len = 60) {
  return s.length > len ? s.slice(0, len - 1) + "…" : s;
 }

+const VALID_MCP_TRANSPORTS = ["stdio", "sse", "streamable-http"];
+
 const mcpToolSchema = [
  { key: "name", header: "Tool", width: 36 },
  {
@@ -43,6 +45,25 @@ export function registerMcp(program) {
      if (exitCode !== 0) process.exit(exitCode);
    });

+  mcp
+    .command("enable")
+    .description(t("mcp.enable.description"))
+    .option("--transport <transport>", t("mcp.enable.transport"))
+    .action(async (opts, cmd) => {
+      const globalOpts = cmd.parent.optsWithGlobals();
+      const exitCode = await runMcpEnableCommand({ ...opts, output: globalOpts.output });
+      if (exitCode !== 0) process.exit(exitCode);
+    });
+
+  mcp
+    .command("disable")
+    .description(t("mcp.disable.description"))
+    .action(async (opts, cmd) => {
+      const globalOpts = cmd.parent.optsWithGlobals();
+      const exitCode = await runMcpDisableCommand({ ...opts, output: globalOpts.output });
+      if (exitCode !== 0) process.exit(exitCode);
+    });
+
  // 5.1 — mcp call + mcp scopes
  mcp
    .command("call <tool> [argsJson]")
@@ -61,10 +82,15 @@ export function registerMcp(program) {
            ? JSON.parse(argsPositional)
            : {};

-      const exitCode = await runMcpCallCommand(tool, args, {
-        ...opts,
-        stream: opts.stream,
-      }, globalOpts);
+      const exitCode = await runMcpCallCommand(
+        tool,
+        args,
+        {
+          ...opts,
+          stream: opts.stream,
+        },
+        globalOpts
+      );

      if (exitCode !== 0) process.exit(exitCode);
    });
@@ -127,7 +153,9 @@ async function mcpJsonRpcCall(tool, args, { stream = false, globalOpts = {} } =

  if (!initRes.ok) {
    const text = await initRes.text().catch(() => "");
-    process.stderr.write(`MCP initialize failed: HTTP ${initRes.status}${text ? ` — ${text}` : ""}\n`);
+    process.stderr.write(
+      `MCP initialize failed: HTTP ${initRes.status}${text ? ` — ${text}` : ""}\n`
+    );
    return 1;
  }

@@ -227,6 +255,7 @@ export async function runMcpStatusCommand(opts = {}) {
    });
    if (!res.ok) {
      console.log(t("mcp.stopped"));
+      console.log(t("mcp.stoppedHint"));
      return 0;
    }

@@ -240,6 +269,9 @@ export async function runMcpStatusCommand(opts = {}) {
    const transport = status.transport || "stdio";
    const online = status.online ?? status.running;
    console.log(online ? t("mcp.running", { transport }) : t("mcp.stopped"));
+    if (!online && status.enabled === false) {
+      console.log(t("mcp.stoppedHint"));
+    }
    if (status.toolsCount !== undefined) console.log(`  Tools: ${status.toolsCount}`);
    if (status.scopes?.length) {
      console.log("  Scopes:");
@@ -270,10 +302,76 @@ export async function runMcpRestartCommand(opts = {}) {
      console.log(t("mcp.restarted"));
      return 0;
    }
-    console.error(t("common.error", { message: `HTTP ${res.status}` }));
+    const body = await res.json().catch(() => null);
+    const message = body?.error || `HTTP ${res.status}`;
+    console.error(t("common.error", { message }));
    return 1;
  } catch (err) {
    console.error(t("common.error", { message: err instanceof Error ? err.message : String(err) }));
    return 1;
  }
 }
+
+export async function runMcpEnableCommand(opts = {}) {
+  const serverUp = await isServerUp();
+  if (!serverUp) {
+    console.error(t("common.serverOffline"));
+    return 1;
+  }
+
+  if (opts.transport && !VALID_MCP_TRANSPORTS.includes(opts.transport)) {
+    console.error(
+      t("common.error", {
+        message: `Invalid transport '${opts.transport}'. Valid: ${VALID_MCP_TRANSPORTS.join(", ")}`,
+      })
+    );
+    return 1;
+  }
+
+  try {
+    const body = { mcpEnabled: true };
+    if (opts.transport) body.mcpTransport = opts.transport;
+
+    const res = await apiFetch("/api/settings", {
+      method: "PATCH",
+      body,
+      retry: false,
+      acceptNotOk: true,
+    });
+    if (!res.ok) {
+      console.error(t("common.error", { message: `HTTP ${res.status}` }));
+      return 1;
+    }
+    console.log(t("mcp.enabled"));
+    return 0;
+  } catch (err) {
+    console.error(t("common.error", { message: err instanceof Error ? err.message : String(err) }));
+    return 1;
+  }
+}
+
+export async function runMcpDisableCommand(opts = {}) {
+  const serverUp = await isServerUp();
+  if (!serverUp) {
+    console.error(t("common.serverOffline"));
+    return 1;
+  }
+
+  try {
+    const res = await apiFetch("/api/settings", {
+      method: "PATCH",
+      body: { mcpEnabled: false },
+      retry: false,
+      acceptNotOk: true,
+    });
+    if (!res.ok) {
+      console.error(t("common.error", { message: `HTTP ${res.status}` }));
+      return 1;
+    }
+    console.log(t("mcp.disabled"));
+    return 0;
+  } catch (err) {
+    console.error(t("common.error", { message: err instanceof Error ? err.message : String(err) }));
+    return 1;
+  }
+}
--- a/bin/cli/locales/en.json
+++ b/bin/cli/locales/en.json
@@ -256,6 +256,7 @@
    "max_restarts": "Max crash restarts within 30s before giving up (default: 2)",
    "tray": "Start in the system tray (desktop only, opt-in)",
    "no_tray": "Disable system tray icon",
+    "ready_timeout": "Readiness probe timeout in ms (also OMNIROUTE_READY_TIMEOUT_MS, default 60000)",
    "tls_cert": "Path to a TLS certificate (PEM) to serve HTTPS (also OMNIROUTE_TLS_CERT)",
    "tls_key": "Path to the TLS private key (PEM) to serve HTTPS (also OMNIROUTE_TLS_KEY)"
  },
@@ -347,6 +348,16 @@
    "running": "MCP server running ({transport})",
    "stopped": "MCP server stopped.",
    "restarted": "MCP server restarted.",
+    "stoppedHint": "Run `omniroute mcp enable` to turn it on.",
+    "enabled": "MCP server enabled.",
+    "disabled": "MCP server disabled.",
+    "enable": {
+      "description": "Enable the MCP server",
+      "transport": "Transport to use: stdio|sse|streamable-http"
+    },
+    "disable": {
+      "description": "Disable the MCP server"
+    },
    "call": {
      "description": "Invoke an MCP tool directly",
      "args": "JSON arguments object (inline)",
--- a/bin/cli/locales/zh-CN.json
+++ b/bin/cli/locales/zh-CN.json
@@ -254,6 +254,7 @@
    "max_restarts": "30 秒内的最大崩溃重启次数（默认：2）",
    "tray": "显示系统托盘图标（仅桌面，选择加入）",
    "no_tray": "禁用系统托盘图标",
+    "ready_timeout": "就绪探测超时（毫秒）（也可用 OMNIROUTE_READY_TIMEOUT_MS，默认 60000）",
    "tls_cert": "用于提供 HTTPS 服务的 TLS 证书（PEM）路径（也可用 OMNIROUTE_TLS_CERT）",
    "tls_key": "用于提供 HTTPS 服务的 TLS 私钥（PEM）路径（也可用 OMNIROUTE_TLS_KEY）"
  },
@@ -345,6 +346,16 @@
    "running": "MCP 服务器正在运行（{transport}）",
    "stopped": "MCP 服务器已停止。",
    "restarted": "MCP 服务器已重启。",
+    "stoppedHint": "运行 `omniroute mcp enable` 以启用它。",
+    "enabled": "MCP 服务器已启用。",
+    "disabled": "MCP 服务器已禁用。",
+    "enable": {
+      "description": "启用 MCP 服务器",
+      "transport": "要使用的传输方式：stdio|sse|streamable-http"
+    },
+    "disable": {
+      "description": "禁用 MCP 服务器"
+    },
    "call": {
      "description": "直接调用 MCP 工具",
      "args": "JSON 参数对象（内联）",
--- a/bin/cli/locales/zh-TW.json
+++ b/bin/cli/locales/zh-TW.json
@@ -254,6 +254,7 @@
    "max_restarts": "30 秒內的最大崩潰重啟次數（預設：2）",
    "tray": "顯示系統托盤圖示（僅桌面，選擇加入）",
    "no_tray": "停用系統托盤圖示",
+    "ready_timeout": "就緒探測逾時（毫秒）（也可用 OMNIROUTE_READY_TIMEOUT_MS，預設 60000）",
    "tls_cert": "用於提供 HTTPS 服務的 TLS 憑證（PEM）路徑（也可用 OMNIROUTE_TLS_CERT）",
    "tls_key": "用於提供 HTTPS 服務的 TLS 私鑰（PEM）路徑（也可用 OMNIROUTE_TLS_KEY）"
  },
@@ -345,6 +346,16 @@
    "running": "MCP 伺服器正在執行（{transport}）",
    "stopped": "MCP 伺服器已停止。",
    "restarted": "MCP 伺服器已重啟。",
+    "stoppedHint": "執行 `omniroute mcp enable` 以啟用它。",
+    "enabled": "MCP 伺服器已啟用。",
+    "disabled": "MCP 伺服器已停用。",
+    "enable": {
+      "description": "啟用 MCP 伺服器",
+      "transport": "要使用的傳輸方式：stdio|sse|streamable-http"
+    },
+    "disable": {
+      "description": "停用 MCP 伺服器"
+    },
    "call": {
      "description": "直接呼叫 MCP 工具",
      "args": "JSON 引數物件（內聯）",
--- a/bin/cli/runtime/processSupervisor.mjs
+++ b/bin/cli/runtime/processSupervisor.mjs
@@ -14,6 +14,7 @@ import { stopProcessGracefully } from "../../../src/shared/platform/windowsProce
 import {
  isFatalInstrumentationHookFailure,
  formatAndroidInstrumentationFailureHint,
+  isFatalStartupDiagnostic,
 } from "../utils/ensureAndroidCacheDir.mjs";

 const CRASH_LOG_LINES = 50;
@@ -55,12 +56,14 @@ export class ServerSupervisor {
    this.child = null;
    this.isShuttingDown = false;
    this.instrumentationFailureHintPrinted = false;
+    this.fatalStartupDiagnosticPrinted = false;
  }

  start() {
    this.startedAt = Date.now();
    this.crashLog = [];
    this.instrumentationFailureHintPrinted = false;
+    this.fatalStartupDiagnosticPrinted = false;

    const showLog = process.env.OMNIROUTE_SHOW_LOG === "1";
    // #6321: stdout used to be discarded (`"ignore"`) whenever `--log`/OMNIROUTE_SHOW_LOG
@@ -99,6 +102,15 @@ export class ServerSupervisor {
          )
        );
      }
+      // #13314: surface any `[STARTUP] Fatal:`-guarded boot diagnostic
+      // immediately, even without --log — otherwise it is only buffered and
+      // reaches the operator on exit/crash, which never happens when the
+      // HTTP listener still comes up after the fatal failure (every route
+      // then 500s with zero visible diagnostic anywhere).
+      if (!this.fatalStartupDiagnosticPrinted && isFatalStartupDiagnostic(text)) {
+        this.fatalStartupDiagnosticPrinted = true;
+        process.stderr.write(text.endsWith("\n") ? text : `${text}\n`);
+      }
    };

    if (this.child.stdout) {
--- a/bin/cli/tray/icon.ico
+++ b/bin/cli/tray/icon.ico
--- a/bin/cli/tray/icon.png
+++ b/bin/cli/tray/icon.png
--- a/bin/cli/utils/ensureAndroidCacheDir.mjs
+++ b/bin/cli/utils/ensureAndroidCacheDir.mjs
@@ -105,6 +105,27 @@ export function isFatalInstrumentationHookFailure(text) {
  return /Unsupported platform:\s*android/i.test(text);
 }

+/**
+ * Detect any fatal boot-time diagnostic guarded by the `[STARTUP] Fatal:`
+ * prefix (`src/instrumentation-node.ts::ensureDbReadyForBoot()`,
+ * `src/instrumentation.ts::register()`, and any future guard using the same
+ * marker). #13314: in the default `omniroute serve` mode (no `--log`),
+ * `ServerSupervisor` only buffers stdout/stderr and flushes it to the real
+ * console on exit/crash/readiness-timeout — so if the HTTP listener still
+ * comes up after a fatal boot diagnostic was already printed (e.g. the
+ * better-sqlite3 / node:sqlite driver cascade failing hard), the operator
+ * sees "OmniRoute is running!" with zero visible diagnostic anywhere, and
+ * every route 500s. This generalizes the #10028 Android/Termux carve-out to
+ * every `[STARTUP] Fatal:` guard, not just that one platform-specific string.
+ *
+ * @param {string} text
+ * @returns {boolean}
+ */
+export function isFatalStartupDiagnostic(text) {
+  if (!text) return false;
+  return /^\[STARTUP\] Fatal:/m.test(text);
+}
+
 /**
 * Operator-facing hint when that instrumentation failure shows up in child
 * output — defense in depth if prep was skipped or a future Next.js probe
--- a/changelog.d/features/13827-i18n-key-completeness-gate.md
+++ b/changelog.d/features/13827-i18n-key-completeness-gate.md
@@ -0,0 +1 @@
+- **feat(i18n):** new blocking gate `i18n:check-keys` (`scripts/i18n/check-key-completeness.mjs`) — every locale catalog must carry exactly the key set of `en.json`, whatever the age of the key; the percentage and new-key gates let batch 1 (#13044) ship 43 keys short and batch 2 (#13660) 10 keys short. The i18n guide now documents the post-merge re-sync and the retranslation flow. (#13827)
--- a/changelog.d/features/agnes-cn-provider.md
+++ b/changelog.d/features/agnes-cn-provider.md
@@ -0,0 +1 @@
+- feat(providers): **Added Agnes AI (China) as `agnes-cn` pointed at `https://api.agnes-ai.cn/v1`. Keys issued for `apihub.agnes-ai.com` stay on the existing `agnes` card. Live `/v1/models` on that host lists `agnes-3.0-flash` (same id as intl); the CN seed matches 2.0/2.5/3.0 and not retired 1.5.**
--- a/changelog.d/features/xai-oauth-live-models.md
+++ b/changelog.d/features/xai-oauth-live-models.md
@@ -0,0 +1 @@
+- **feat(providers):** share the existing `api.x.ai/v1/models` discovery config with `xai-oauth` so SuperGrok OAuth connections pick up new Grok ids without a registry seed edit.
--- a/changelog.d/fixes/12370-responses-function-call-name.md
+++ b/changelog.d/fixes/12370-responses-function-call-name.md
@@ -0,0 +1 @@
+- fix(api): restore the `name` field on non-streaming `/v1/responses` `function_call` output items — a plain (non-namespace) tool call's identity restore was blindly applying the `_toolNameMap` alias-table fallback as a `{namespace, name}` object, silently blanking `name` to `undefined` (dropped entirely by JSON.stringify) and leaving Codex unable to dispatch the call, so it re-narrated its intent in a loop instead (#12370)
--- a/changelog.d/fixes/12453-conversation-turn-nodes-retention.md
+++ b/changelog.d/fixes/12453-conversation-turn-nodes-retention.md
@@ -0,0 +1 @@
+- **fix(db):** give `conversation_turn_nodes` its own independent retention knob (`retention.conversationTurnNodes`, default 30 days — matching `callLogs` so upgrading changes nothing until an operator overrides it) instead of sharing `callLogs`, and sweep orphaned `agentic_conversations` after the nodes expire (#12453).
--- a/changelog.d/fixes/12491-codex-wreq-standalone-runtime.md
+++ b/changelog.d/fixes/12491-codex-wreq-standalone-runtime.md
@@ -0,0 +1 @@
+- **fix(sse):** Codex WebSocket transport (including the app-server) no longer fails to load in the Next.js standalone Docker runtime — the wreq-js loader now resolves its module name dynamically instead of a literal Turbopack could rewrite to an unreachable build-time symlink (#12491) — thanks @marshalfevzi
--- a/changelog.d/fixes/12692-xai-legacy-function-call.md
+++ b/changelog.d/fixes/12692-xai-legacy-function-call.md
@@ -0,0 +1 @@
+- **fix(providers):** xAI requests no longer silently drop an assistant tool call sent in the legacy OpenAI `function_call` shape (instead of `tool_calls[]`) — the call is now translated into the xAI request the same way modern tool calls are (#12692) — thanks @soroush5
--- a/changelog.d/fixes/12700-xai-usage-total.md
+++ b/changelog.d/fixes/12700-xai-usage-total.md
@@ -0,0 +1 @@
+- **fix(providers):** xAI responses no longer report `total_tokens`/`totalTokenCount` as `0` when upstream usage uses the legacy `prompt_tokens`/`completion_tokens` names instead of `input_tokens`/`output_tokens` (#12700) — thanks @soroush5
--- a/changelog.d/fixes/12861-direct-fetch-timeout-unhandled-rejection.md
+++ b/changelog.d/fixes/12861-direct-fetch-timeout-unhandled-rejection.md
@@ -0,0 +1 @@
+- **fix(resilience):** a recoverable direct-fetch response-start timeout (`DIRECT_RESPONSE_START_TIMEOUT`) could, in a narrow timer/promise-settlement race, escape as an `unhandledRejection` → `uncaughtException` and kill the server process — even though `proxyFetch` already retries this exact condition on a fresh socket. Guarded the timer callback so it can no longer fire against an already-settled attempt, and extended the process-level crash guard (already used by the WS/API-bridge servers) to recognize and swallow this code if it ever escapes anyway. Also installs that same guard in the production server entrypoint (`dist/server-ws.mjs`), which never had it even though the dev server already did ([#12861](https://github.com/diegosouzapw/OmniRoute/issues/12861)) — thanks @insoln
--- a/changelog.d/fixes/12927-magnific-key-validation-false-negative.md
+++ b/changelog.d/fixes/12927-magnific-key-validation-false-negative.md
@@ -0,0 +1 @@
+- **fix(providers):** correct Magnific API key validation, which reported every valid key as invalid due to a GET probe against a POST-only endpoint (#12927) — thanks @hubo1989
--- a/changelog.d/fixes/12958-gitlab-duo-403-entitlement-fallback.md
+++ b/changelog.d/fixes/12958-gitlab-duo-403-entitlement-fallback.md
@@ -0,0 +1 @@
+- **fix(providers):** GitLab Duo Retest and chat requests now fall back to the public Code Suggestions endpoint for ANY `direct_access` 403 (not only the "direct connections are disabled" tenant-config message), and surface the real upstream error body instead of a generic "Access denied" when both endpoints reject the token (#12958) — thanks @Rahulsharma0810
--- a/changelog.d/fixes/12968-anthropic-shim-tiny-probe-empty.md
+++ b/changelog.d/fixes/12968-anthropic-shim-tiny-probe-empty.md
@@ -0,0 +1 @@
+- **fix(sse):** stop misclassifying a truncated Anthropic-compatible `max_tokens` probe response (`content:[{type:"text",text:""}]`) as an empty upstream response (#12968) — thanks @pranay-gpt
--- a/changelog.d/fixes/13000-combo-context-limit-cold-start.md
+++ b/changelog.d/fixes/13000-combo-context-limit-cold-start.md
@@ -0,0 +1 @@
+- fix(providers): stop `@omniroute/opencode-plugin` combo context limits from downgrading to the raw `Math.min(member)` lower bound after a restart — the static catalog now honors the server-computed `computed_context_length` (mirroring the dynamic hook), and a background refresh with a degraded `/api/combos` response backfills the field from the last-known-good disk snapshot instead of overwriting it (#13000) — thanks @morpheus9393
--- a/changelog.d/fixes/13012-cli-mcp-restart-and-enable.md
+++ b/changelog.d/fixes/13012-cli-mcp-restart-and-enable.md
@@ -0,0 +1 @@
+- **fix(cli):** `omniroute mcp restart` no longer 404s — the missing `POST /api/mcp/restart` route now exists — and new `omniroute mcp enable`/`mcp disable [--transport]` subcommands give the CLI a way to turn the MCP server on without the dashboard ([#13012](https://github.com/diegosouzapw/OmniRoute/issues/13012)) — thanks @ricardusx
--- a/changelog.d/fixes/13022-desktop-exe-malformed-skill-tool-schemas.md
+++ b/changelog.d/fixes/13022-desktop-exe-malformed-skill-tool-schemas.md
@@ -0,0 +1 @@
+- **fix(skills):** repair nested malformed skill-tool schemas (bare property maps, boolean `required: true`) for OpenAI-compatible providers, not just the schema root (#13022) — thanks @ftevxk
--- a/changelog.d/fixes/13089-combo-live-roundrobin-missing-events.md
+++ b/changelog.d/fixes/13089-combo-live-roundrobin-missing-events.md
@@ -0,0 +1 @@
+- **fix(routing):** round-robin combos now show up in Combo Studio's Live dashboard — they were completing successfully but never publishing the attempt/success/failure events the dashboard listens for (#13089) — thanks @adityadwi21
--- a/changelog.d/fixes/13122-responses-custom-tool-choice.md
+++ b/changelog.d/fixes/13122-responses-custom-tool-choice.md
@@ -0,0 +1 @@
+- **fix(sse):** stop rejecting a Responses API `tool_choice.type: "custom"` (e.g. Codex CLI forcing `functions__exec`) with a 400 `unsupported_feature` error (#13122) — thanks @phamtienduceng-eng
--- a/changelog.d/fixes/13232-zai-web-missing-browser-executable.md
+++ b/changelog.d/fixes/13232-zai-web-missing-browser-executable.md
@@ -0,0 +1 @@
+- **fix(sse):** classify a missing Playwright Chromium install on the Z.ai web transport as an actionable 503 host/config cooldown instead of a generic 502 that trips the provider circuit breaker (#13232) — thanks @oleksandr1811
--- a/changelog.d/fixes/13234-embed-lan-keyed-auth.md
+++ b/changelog.d/fixes/13234-embed-lan-keyed-auth.md
@@ -0,0 +1 @@
+- **fix(embeddings):** LAN/CGNAT OpenAI-compatible embeddings nodes with a stored API key now send `Authorization: Bearer` on the outbound request, matching dashboard Check. Keyless LAN nodes stay no-auth ([#6925](https://github.com/diegosouzapw/OmniRoute/issues/6925)) ([#13234](https://github.com/diegosouzapw/OmniRoute/issues/13234))
--- a/changelog.d/fixes/13306-windows-libuv-abort-sqljs-exit.md
+++ b/changelog.d/fixes/13306-windows-libuv-abort-sqljs-exit.md
@@ -0,0 +1 @@
+- **fix(db):** defer `process.exit(0)` on graceful shutdown by one macrotask, avoiding a Windows-only libuv abort when the sql.js fallback driver has a statement in flight (#13306) — thanks @anhtahaylove
--- a/changelog.d/fixes/13314-supervisor-surfaces-fatal-startup-diagnostic.md
+++ b/changelog.d/fixes/13314-supervisor-surfaces-fatal-startup-diagnostic.md
@@ -0,0 +1 @@
+- **fix(cli):** `omniroute serve` now surfaces a fatal `[STARTUP] Fatal: ...` boot diagnostic (e.g. a DB driver init failure) to the console immediately, even without `--log`, instead of only when the process later crashes or restarts (#13314) — thanks @Orion1943
--- a/changelog.d/fixes/13326-memory-fts-skip-access-updates.md
+++ b/changelog.d/fixes/13326-memory-fts-skip-access-updates.md
@@ -0,0 +1 @@
+- **fix(memory):** stop FTS5 rewrite on access-count updates; rebuild the index on cleanup so leftover tombstones shrink (#13326).
--- a/changelog.d/fixes/13364-zed-hosted-haiku-thinking-inflation.md
+++ b/changelog.d/fixes/13364-zed-hosted-haiku-thinking-inflation.md
@@ -0,0 +1 @@
+- **fix(providers):** stop zed-hosted `claude-haiku-4-5` extended-thinking requests from inflating `max_tokens` past the model's real 64000 output cap (#13364) — thanks @ThiagoMafra-Integrare
--- a/changelog.d/fixes/13380-gemini-web-system-and-tool-prompt.md
+++ b/changelog.d/fixes/13380-gemini-web-system-and-tool-prompt.md
@@ -0,0 +1 @@
+- **fix(providers):** gemini-web no longer drops the system instruction on single-turn requests or the tool contract when a client system message is present, and switches to an atomic composer insert so embedded newlines can't submit the message early (#13380) — thanks @formilw
--- a/changelog.d/fixes/13389-catalog-cache-backoff-reset.md
+++ b/changelog.d/fixes/13389-catalog-cache-backoff-reset.md
@@ -0,0 +1 @@
+- **fix(db):** stop routine connection-backoff auto-recovery from busting the entire `/v1/models` response cache, which was causing intermittent 75-120s/502 responses on deployments routing many providers (#13389) — thanks @RaviTharuma
--- a/changelog.d/fixes/13429-lite-redundant-remove-tool-call-id.md
+++ b/changelog.d/fixes/13429-lite-redundant-remove-tool-call-id.md
@@ -0,0 +1 @@
+- **fix(compression):** stop lite compression from dropping a `role:"tool"` message when it is byte-identical to the previous message, which orphaned a `tool_call_id` and triggered upstream 400 errors on parallel tool calls (#13429) — thanks @tolgaaksoy
--- a/changelog.d/fixes/13431-responses-post-keepalive-error-type.md
+++ b/changelog.d/fixes/13431-responses-post-keepalive-error-type.md
@@ -0,0 +1 @@
+- **fix(sse):** frame post-keepalive `/v1/responses` stream errors with a top-level `type` field so Responses clients (Codex) surface the real upstream error instead of reporting "stream disconnected before completion" (#13431) — thanks @andrea-kingautomation
--- a/changelog.d/fixes/13432-incremental-auto-vacuum-drift.md
+++ b/changelog.d/fixes/13432-incremental-auto-vacuum-drift.md
@@ -0,0 +1 @@
+- **fix(db):** reconcile `auto_vacuum` drift between the configured INCREMENTAL mode and the live SQLite pragma — detected at startup and reconciled out-of-request by the vacuum scheduler, which now also runs a bounded `PRAGMA incremental_vacuum` reclaim instead of an unconditional full `VACUUM` once INCREMENTAL is actually in effect (#13432) — thanks @tolgaaksoy
--- a/changelog.d/fixes/13445-groq-tls-fingerprint.md
+++ b/changelog.d/fixes/13445-groq-tls-fingerprint.md
@@ -0,0 +1 @@
+- **fix(network):** skip Chrome TLS impersonation for Groq (`api.groq.com`); Cloudflare 1010s that JA3 while native undici reaches the API ([#13445](https://github.com/diegosouzapw/OmniRoute/pull/13445)) (#13225)
--- a/changelog.d/fixes/13452-provider-node-baseurl-ignored.md
+++ b/changelog.d/fixes/13452-provider-node-baseurl-ignored.md
@@ -0,0 +1 @@
+- **fix(sse):** stop an unhydrated `openai-compatible-*`/`anthropic-compatible-*` connection from silently routing chat requests (and its stored credential) to the real OpenAI/Anthropic API instead of the operator's configured provider-node endpoint (#13452) — thanks @DenXio101
--- a/changelog.d/fixes/13470-background-oauth-refresh-proxy-guard.md
+++ b/changelog.d/fixes/13470-background-oauth-refresh-proxy-guard.md
@@ -0,0 +1 @@
+- **fix(resilience):** background OAuth token refresh (proactive health-check sweep and the shared refresh helper behind `refreshAccessToken`/`refreshClaudeOAuthToken`/etc.) now fails closed like the interactive chat path when a connection's assigned proxy pool is entirely dead, instead of silently sending the refresh-token exchange out direct or via a stray `HTTPS_PROXY` (#13470) — thanks @elielsousa-pathbit
--- a/changelog.d/fixes/13472-responses-cache-creation-tokens.md
+++ b/changelog.d/fixes/13472-responses-cache-creation-tokens.md
@@ -0,0 +1 @@
+- **fix(sse):** forward Anthropic prompt-cache-creation tokens through the `/v1/responses` usage hop so cache-write counts stop logging as zero (#13472) — thanks @fidelix
--- a/changelog.d/fixes/13488-pii-sanitizer-splices-openrouter-metadata.md
+++ b/changelog.d/fixes/13488-pii-sanitizer-splices-openrouter-metadata.md
@@ -0,0 +1 @@
+- **fix(sse):** stop the streaming PII sanitizer from splicing OpenRouter metadata (`provider`, `native_finish_reason`, `reasoning_details[].format`) into the answer text buffer (#13488) — thanks @Xore
--- a/changelog.d/fixes/13535-windows-tray-icon-contrast.md
+++ b/changelog.d/fixes/13535-windows-tray-icon-contrast.md
@@ -0,0 +1 @@
+- **fix(cli):** redraw the CLI/Electron system tray icon with a dark outline and ship a native multi-res `icon.ico` so it is no longer a pure-white, nearly invisible glyph on the Windows light-theme taskbar and hidden-icons flyout (#13535) — thanks @ProphetOfDoom-PoD
--- a/changelog.d/fixes/13544-audio-transcription-call-log.md
+++ b/changelog.d/fixes/13544-audio-transcription-call-log.md
@@ -0,0 +1 @@
+- **fix(api):** `/v1/audio/transcriptions`, `/v1/audio/translations` and `/v1/audio/speech` requests now show up in Dashboard → Request Logs — the three routes never called the shared call-log pipeline, so every successful (and failed) transcription/translation/speech request was silently dropped from `call_logs` ([#13544](https://github.com/diegosouzapw/OmniRoute/issues/13544)) — thanks @delafu
--- a/changelog.d/fixes/13558-minimax-m3-reasoning-leak.md
+++ b/changelog.d/fixes/13558-minimax-m3-reasoning-leak.md
@@ -0,0 +1 @@
+- **fix(providers):** MiniMax-M3's inline `<think>...</think>` reasoning no longer leaks into `message.content`/`delta.content` on the `minimax`/`minimax-cn` routes — it is now stripped and surfaced as `reasoning_content`, in both streaming and non-streaming responses (#13558) — thanks @pan17
--- a/changelog.d/fixes/13562-hide-auto-nothink-settings-schema.md
+++ b/changelog.d/fixes/13562-hide-auto-nothink-settings-schema.md
@@ -0,0 +1 @@
+- **fix(api):** `PATCH /api/settings` now persists `hideAutoCombos` and `hideNoThinkVariants` instead of silently dropping them (#13562) — thanks @texastoland
--- a/changelog.d/fixes/13591-antigravity-opaque-400-error-surfaced.md
+++ b/changelog.d/fixes/13591-antigravity-opaque-400-error-surfaced.md
@@ -0,0 +1,3 @@
+- **fix(providers):** Antigravity error responses and logs now surface the real upstream
+  message (e.g. Gemini field-path rejections) instead of the generic "Antigravity upstream
+  error (400)" placeholder (#13591) — thanks @afonsoft
--- a/changelog.d/fixes/13597-calllogs-worker-error-detail.md
+++ b/changelog.d/fixes/13597-calllogs-worker-error-detail.md
@@ -0,0 +1 @@
+- **fix(usage):** the call-logs artifact worker's failure warning now includes the underlying error's message/code instead of the generic "detail omitted" — a crashed or non-zero-exit worker was previously undiagnosable in the logs (#13597) — thanks @afonsoft
--- a/changelog.d/fixes/13599-deepseek-bai-reasoning-content-echo.md
+++ b/changelog.d/fixes/13599-deepseek-bai-reasoning-content-echo.md
@@ -0,0 +1 @@
+- **fix(providers):** echo back `reasoning_content` on `bai` DeepSeek thinking-mode follow-up turns, fixing the upstream 400 "reasoning_content must be passed back" (#13599) — thanks @afonsoft
--- a/changelog.d/fixes/13602-i18n-duplicated-flag-description-key.md
+++ b/changelog.d/fixes/13602-i18n-duplicated-flag-description-key.md
@@ -0,0 +1 @@
+- **fix(i18n):** drop the second copy of `featureFlagProxySkipRecentlyFailedDescription` that the 2026-09-15 batch merges left in 59 dashboard catalogs (a scripted keep-both conflict resolution concatenated the key both PRs carried; `JSON.parse` silently kept the last copy) and the duplicated `ERROR_TYPE_CONTRACT` import in `src/lib/db/callLogStats.ts` (TS2300); adds `tests/unit/i18n-catalogs-no-duplicate-keys.test.ts`, a raw-text guard that fails on any key declared twice in one object of `src/i18n/messages/*.json` or `bin/cli/locales/*.json` ([#13602](https://github.com/diegosouzapw/OmniRoute/pull/13602), [#13641](https://github.com/diegosouzapw/OmniRoute/pull/13641))
--- a/changelog.d/fixes/13609-mistral-401-ambiguous-auth.md
+++ b/changelog.d/fixes/13609-mistral-401-ambiguous-auth.md
@@ -0,0 +1 @@
+- **fix(providers):** opt-in `MISTRAL_AMBIGUOUS_401_SOFT_LOCKOUT` flag (default off): a bare Mistral 401 with no explicit auth signal (identical for a revoked key and an exhausted quota) cools the connection down instead of parking it as expired, at most 3 times per hour per connection before it parks, so a revoked key still converges; the ambiguity check is now one implementation shared by the connection test and the runtime ([#13609](https://github.com/diegosouzapw/OmniRoute/pull/13609)) — thanks @maxmad64bis
--- a/changelog.d/fixes/13620-combo-reasoning-only-sse-burst.md
+++ b/changelog.d/fixes/13620-combo-reasoning-only-sse-burst.md
@@ -0,0 +1 @@
+- **fix(sse):** stream reasoning deltas from combo targets incrementally instead of buffering them into a single burst, and stop rejecting reasoning-only streams as an empty completion (#13620) — thanks @NaNomicon
--- a/changelog.d/fixes/13628-grok46-default-effort.md
+++ b/changelog.d/fixes/13628-grok46-default-effort.md
@@ -0,0 +1,2 @@
+- fix(providers): restore grok-4.6/4.5 default reasoning effort so requests without an explicit effort keep reasoning enabled (#13628)
+- fix(registry): declare supportedThinkingEfforts on claude-opus-5 and claude-fable-5 across the anthropic/claude/claude-web/ghe-copilot/github registries (#13628)
--- a/changelog.d/fixes/13636-pack-artifact-policy-abort-guard-test.md
+++ b/changelog.d/fixes/13636-pack-artifact-policy-abort-guard-test.md
@@ -0,0 +1 @@
+- **fix(tests):** add `dist/httpClientAbortGuard.mjs` to the expected missing-paths list in `tests/unit/pack-artifact-policy.test.ts` — [#13636](https://github.com/diegosouzapw/OmniRoute/pull/13636) registered the file in `PACK_ARTIFACT_REQUIRED_PATHS` without updating the assertion, leaving the test red on the release tip for every PR that runs it ([#12732](https://github.com/diegosouzapw/OmniRoute/issues/12732))
--- a/changelog.d/fixes/13652-kiro-tooldocs-repeat-every-turn.md
+++ b/changelog.d/fixes/13652-kiro-tooldocs-repeat-every-turn.md
@@ -0,0 +1 @@
+- **fix(sse):** Kiro translator no longer re-prepends the full relocated tool-documentation block onto every subsequent turn of a multi-turn conversation; it now stays anchored to the turn that originally carried it. (#13652) — thanks @KelvinKSPS
--- a/changelog.d/fixes/13657-rate-limited-429-single-retry-drain.md
+++ b/changelog.d/fixes/13657-rate-limited-429-single-retry-drain.md
@@ -0,0 +1 @@
+- **fix(sse):** opt-in `OPENCODE_RATE_LIMITED_429_EARLY_STOP` flag (default off): an opencode 429 classified as a real rate limit (parseable `Retry-After`, or a body naming a rate/usage limit) stops the cross-account wave and returns that upstream 429 unchanged — body, `Retry-After` and quota headers intact, so the opencode quota error rules still apply; unclassified 429s keep rotating, and with the flag off every 429 rotates as before (#9611) ([#13657](https://github.com/diegosouzapw/OmniRoute/pull/13657)) — thanks @maxmad64bis
--- a/changelog.d/fixes/13679-cdp-proxy-auth-network-isolation.md
+++ b/changelog.d/fixes/13679-cdp-proxy-auth-network-isolation.md
@@ -0,0 +1 @@
+- **fix(docker):** isolate the ChatGPT Web (Codex) CDP proxy sidecar onto its own Compose network, add an opt-in `CDP_PROXY_TOKEN` auth gate to `cdp-proxy.mjs`, and stop the VNC browser-login CDP bridge from starting when no token is configured (#13679)
--- a/changelog.d/fixes/13679-cloudsync-hmac-fail-open.md
+++ b/changelog.d/fixes/13679-cloudsync-hmac-fail-open.md
@@ -0,0 +1 @@
+- **fix(auth):** `verifyCloudSignature()` no longer accepts an unverifiable `X-Cloud-Sig` when `OMNIROUTE_CLOUD_SYNC_SECRET` is unset — a forged/garbage signature is rejected outright, and the new opt-in `OMNIROUTE_CLOUD_SYNC_ENFORCE_SIGNATURE=true` flag rejects unsigned Cloud-sync payloads too (default stays legacy pass-through for v3.8.x; the default flips in v3.9) ([#13679](https://github.com/diegosouzapw/OmniRoute/issues/13679))
--- a/changelog.d/fixes/13679-deploy-manifest-literal-secrets.md
+++ b/changelog.d/fixes/13679-deploy-manifest-literal-secrets.md
@@ -0,0 +1 @@
+- **fix(security):** removed the copy-pasteable placeholder `JWT_SECRET`/`API_KEY_SECRET`/`INITIAL_PASSWORD` values from the Podman Quadlet deploy manifest, and blocked remote dashboard logins with the well-known default `INITIAL_PASSWORD=CHANGEME` (#13679)
--- a/changelog.d/fixes/13679-selfloop-bearer-random.md
+++ b/changelog.d/fixes/13679-selfloop-bearer-random.md
@@ -0,0 +1 @@
+- **fix(security):** the internal self-loop admission-bypass bearer is now a random per-process secret instead of the checked-in literal `"sk_omniroute"` when no `OMNIROUTE_API_KEY`/`ROUTER_API_KEY` is configured (#13679)
--- a/changelog.d/fixes/13680-13681-batches-sweep-cap-shared-file.md
+++ b/changelog.d/fixes/13680-13681-batches-sweep-cap-shared-file.md
@@ -0,0 +1 @@
+- **fix(db):** `DELETE /v1/batches/delete-completed` now caps the work it does per request and reports `hasMore` so a caller can resume, and the sweep no longer deletes a file that another batch still references (#13680, #13681)
--- a/changelog.d/fixes/13782-i18n-retranslate-english-leaves.md
+++ b/changelog.d/fixes/13782-i18n-retranslate-english-leaves.md
@@ -0,0 +1 @@
+- **fix(i18n):** retranslate the English strings that had been copied verbatim into the locale catalogs (Spanish alone carried 7,142) and turn the real-translation ratio gate into a blocking ratchet. (#13782)
--- a/changelog.d/fixes/13795-opencode-429-proxy-dedup.md
+++ b/changelog.d/fixes/13795-opencode-429-proxy-dedup.md
@@ -0,0 +1 @@
+- **fix(sse):** stop re-sending a request to an already-refused route after a 429 — each refused route is tried once per request ([#13795](https://github.com/diegosouzapw/OmniRoute/pull/13795)) — thanks @maxmad64bis
--- a/changelog.d/fixes/agnes-cn-thinking-effort-tiers.md
+++ b/changelog.d/fixes/agnes-cn-thinking-effort-tiers.md
@@ -0,0 +1 @@
+- fix(providers): **declare Agnes CN chat models' live `reasoning_effort` vocabulary** so the catalog and sanitizer stop inventing tiers the CN API rejects. Probes on api.agnes-ai.cn (2026-09-14) match the international endpoint: 2.0/2.5 accept `none/low/medium/high/max`, 3.0 also accepts `minimal` and `xhigh`; `off`/`ultra` clamp off the wire and Hermes' default `xhigh` clamps to `max` on 2.x.
--- a/changelog.d/fixes/agnes-thinking-effort-tiers.md
+++ b/changelog.d/fixes/agnes-thinking-effort-tiers.md
@@ -0,0 +1 @@
+- fix(providers): **declare Agnes chat models' live `reasoning_effort` vocabulary so catalog/builder/sanitizer stop inventing aliases the API 400s.** 2.0/2.5 accept `none/low/medium/high/max`; 3.0 also accepts `minimal` and `xhigh`. `off`/`ultra` still clamp off the wire.
--- a/changelog.d/fixes/basereds-forgotten-sibling-report-bounds.md
+++ b/changelog.d/fixes/basereds-forgotten-sibling-report-bounds.md
@@ -0,0 +1 @@
+- **fix(ci):** the advisory `forgotten-sibling-tests` step no longer fails "Fast Quality Gates" when a PR touches a hub module — the cross-product of consumers × candidate tests reached millions of rows and rendering them exceeded V8's maximum string length, so the throw hit `main()`'s catch and exited 1. The report now lists at most 200 rows per section (and 5 000 per array in the JSON artifact) while the header keeps the exact totals
--- a/changelog.d/fixes/claude-assistant-prefill.md
+++ b/changelog.d/fixes/claude-assistant-prefill.md
@@ -0,0 +1 @@
+- Strip a trailing text-only assistant turn before official Claude OAuth dispatch. Claude returns 400 `This model does not support assistant message prefill` for that shape; the shared strip only covered Mistral.
--- a/changelog.d/fixes/codex-reasoning-object-whitelist.md
+++ b/changelog.d/fixes/codex-reasoning-object-whitelist.md
@@ -0,0 +1 @@
+- Fixed Codex executor forwarding client `reasoning` sub-fields (`enabled`, `max_tokens`, `exclude`) that the Codex Responses API rejects with HTTP 400, taking down every combo target with a deterministic client error. The reasoning object is now whitelisted to `effort`/`summary`, and `enabled: false` maps to effort `none` when no more specific effort was requested.
--- a/changelog.d/fixes/combo-probe-no-thinking.md
+++ b/changelog.d/fixes/combo-probe-no-thinking.md
@@ -0,0 +1 @@
+- **fix(combos):** Gemini combo probes send `reasoning_effort: none` so thinking does not eat the health-check budget; truncated `finish_reason: length` responses are no longer rewritten as empty-content 502s
--- a/changelog.d/fixes/gemini-38-think-level.md
+++ b/changelog.d/fixes/gemini-38-think-level.md
@@ -0,0 +1 @@
+- **fix(gemini):** send Gemini 3.8 `thinkingLevel` instead of a numeric `thinkingBudget`, and omit `includeThoughts` unless the client asked, so hidden thoughts stop eating `maxOutputTokens`
--- a/changelog.d/fixes/release-acceptance-shadow.md
+++ b/changelog.d/fixes/release-acceptance-shadow.md
@@ -0,0 +1 @@
+- Add a shadow release-acceptance report next to release-green.json. It does not close #12732 and is not a Mergify required check.
--- a/changelog.d/fixes/sensenova-deepseek-v4-effort-clamp.md
+++ b/changelog.d/fixes/sensenova-deepseek-v4-effort-clamp.md
@@ -0,0 +1 @@
+- **fix(providers):** clamp SenseNova DeepSeek V4 Flash `reasoning_effort` `xhigh`/`max` to `high` (upstream lists `xhigh` then 400s it)
--- a/changelog.d/maintenance/12732-agent-skills-sync-omni-settings.md
+++ b/changelog.d/maintenance/12732-agent-skills-sync-omni-settings.md
@@ -0,0 +1 @@
+- **chore(skills):** regenerate the `omni-settings` agent skill after the pool egress-observation route landed (#13581), clearing the `check:agent-skills-sync` base-red (#12732)
--- a/changelog.d/maintenance/12732-attempt-log-tests-trace-id.md
+++ b/changelog.d/maintenance/12732-attempt-log-tests-trace-id.md
@@ -0,0 +1 @@
+- **test(call-logs):** the early-keepalive merge and video-bridge redaction tests pass a `traceId` (defaulting to `pendingRequestId`) now that #13546 keys each attempt's call-log row on it, the dashboard `request.failed` redaction probe reads the persisted row by `traceId`, and the keepalive test polls against a 30s wall-clock deadline like the video-bridge test instead of a 2.4s try count ([#12732](https://github.com/diegosouzapw/OmniRoute/issues/12732))
--- a/changelog.d/maintenance/12732-call-log-stats-duplicate-import.md
+++ b/changelog.d/maintenance/12732-call-log-stats-duplicate-import.md
@@ -0,0 +1 @@
+- **fix(db):** drop the duplicated `ERROR_TYPE_CONTRACT` import in `src/lib/db/callLogStats.ts` left by the #13641 merge; the TS2300 duplicate-identifier error failed the API-route and dashboard typecheck gates on every PR ([#12732](https://github.com/diegosouzapw/OmniRoute/issues/12732))
--- a/changelog.d/maintenance/12732-cli-serve-ready-timeout-i18n.md
+++ b/changelog.d/maintenance/12732-cli-serve-ready-timeout-i18n.md
@@ -0,0 +1 @@
+- **fix(cli):** add the `serve.ready_timeout` string to the `en`, `zh-CN` and `zh-TW` CLI catalogs; `--ready-timeout` shipped calling `t("serve.ready_timeout")` without a catalog entry, which the CLI i18n key-coverage and parity tests report ([#12732](https://github.com/diegosouzapw/OmniRoute/issues/12732))
--- a/changelog.d/maintenance/12732-file-size-test-growth-13748-13749.md
+++ b/changelog.d/maintenance/12732-file-size-test-growth-13748-13749.md
@@ -0,0 +1 @@
+- **fix(ci):** re-freeze `tests/unit/image-generation-handler.test.ts` (2133→2235, #13748) and `tests/unit/batch_api.test.ts` (1345→1348, #13749) at their merged size; PR-mode `check:file-size` does not relax `testFrozen` against the base, so that regression coverage turned the gate red on every PR into the release line ([#12732](https://github.com/diegosouzapw/OmniRoute/issues/12732))
--- a/changelog.d/maintenance/12732-gitleaks-uzbek-token-desc.md
+++ b/changelog.d/maintenance/12732-gitleaks-uzbek-token-desc.md
@@ -0,0 +1 @@
+- **fix(ci):** allowlist the Uzbek `outputTokenDesc` translation ("Yakunlash/javob tokenlari") and the `PROTECTED_PRIORITY_INFRA_502_ENABLED` feature-flag id (#13439) in `.gitleaks.toml`; the `generic-api-key` rule reads the `...TokenDesc` key and the flag `key:` as token assignments, which the secrets ratchet reported as new findings ([#12732](https://github.com/diegosouzapw/OmniRoute/issues/12732))
--- a/changelog.d/maintenance/12732-jina-custom-rows-provider-prefix.md
+++ b/changelog.d/maintenance/12732-jina-custom-rows-provider-prefix.md
@@ -0,0 +1 @@
+- **test(models):** the custom Jina specialty-model catalog test expects the `jina-ai/` prefix again: custom rows keep the connection provider id, only synced rows resolve through the `jina` alias, and #13403 had switched the custom assertion to `jina/` ([#12732](https://github.com/diegosouzapw/OmniRoute/issues/12732))
--- a/changelog.d/maintenance/12732-pack-artifact-abort-guard.md
+++ b/changelog.d/maintenance/12732-pack-artifact-abort-guard.md
@@ -0,0 +1 @@
+- **chore(build):** ship `httpClientAbortGuard.mjs` in the published tarball — the #13636 crash guard was a new `server-ws.mjs` import missing from both pack-artifact allowlists (#12732)
--- a/changelog.d/maintenance/12732-paid-target-fixture-gemini-pro.md
+++ b/changelog.d/maintenance/12732-paid-target-fixture-gemini-pro.md
@@ -0,0 +1 @@
+- **test(settings):** the #6540 paid-target tests now use `gemini/gemini-3.1-pro-preview` as the paid fixture and assert the fixtures still classify as paid/free/unknown; the old Together target became "unknown" once #13407 removed Together's one-time signup credit from the free catalog, so the three save-time blocking tests read a correct 200 as a missing guard ([#12732](https://github.com/diegosouzapw/OmniRoute/issues/12732))
--- a/changelog.d/maintenance/12732-stryker-noauth-token-budget-coverage.md
+++ b/changelog.d/maintenance/12732-stryker-noauth-token-budget-coverage.md
@@ -0,0 +1 @@
+- **fix(ci):** register `noauth-model-lockout`, `local-token-budget-429-skips-cooldown`, `free-badge-provider-gate` (#13645) and `daily-reset-tz-threading` (#13440) in `stryker.conf.json` `tap.testFiles`; they cover `accountFallback.ts`/`auth.ts`/`comboPredicates.ts`/`rrState.ts`, so the strict `mutation-test-coverage` gate failed Fast Quality Gates on every PR into the release line ([#12732](https://github.com/diegosouzapw/OmniRoute/issues/12732))
--- a/changelog.d/maintenance/12732-stryker-provider-401-ambiguous.md
+++ b/changelog.d/maintenance/12732-stryker-provider-401-ambiguous.md
@@ -0,0 +1 @@
+- **chore(ci):** register the #13609 ambiguous-401 regression test in the mutation-coverage config, clearing the second `check:agent-skills-sync`/`mutation-test-coverage` base-red (#12732)
--- a/config/quality/eslint-suppressions.json
+++ b/config/quality/eslint-suppressions.json
@@ -21,7 +21,7 @@
  },
  "open-sse/config/providers/registry/claude/index.ts": {
    "@typescript-eslint/no-unused-vars": {
-      "count": 7
+      "count": 6
    }
  },
  "open-sse/config/providers/registry/vertex/index.ts": {
--- a/config/quality/file-size-baseline.json
+++ b/config/quality/file-size-baseline.json
@@ -1,4 +1,9 @@
 {
+  "_rebaseline_2026_09_15_13572_combined_growth": "Combined growth of the 2026-09-15 maxmad64bis uplift batch (each PR rebaselined its own growth; the merged sum is larger): open-sse/executors/base.ts->1754. Every hunk is flag-gated or a verified fix covered by that PR's tests; see the batch report.",
+  "_rebaseline_2026_09_15_13445_combined_growth": "Combined growth of the 2026-09-15 maxmad64bis uplift batch (each PR rebaselined its own growth; the merged sum is larger): open-sse/utils/proxyFetch.ts->1296. Every hunk is flag-gated or a verified fix covered by that PR's tests; see the batch report.",
+  "_rebaseline_2026_09_15_13643_combined_growth": "Combined growth of the 2026-09-15 maxmad64bis uplift batch (each PR rebaselined its own growth; the merged sum is larger): open-sse/executors/codex.ts->1528. Every hunk is flag-gated or a verified fix covered by that PR's tests; see the batch report.",
+  "_rebaseline_2026_09_15_13344_conversation_turn_nodes_retention_field": "PR #13344 rework: src/app/(dashboard)/dashboard/settings/components/SystemStorageTab.tsx 1596->1597 (+1 checker count on the pure tip, matching the existing frozen 1597; this rework's own edit only changes the literal default shown in the retentionFields row from 1 to 30, no line added/removed). The one added line is the PR's own retentionFields row (conversationTurnNodes) exposing the new independent retention.conversationTurnNodes knob (src/types/databaseSettings.ts) added by the same PR. Irreducible: one row per existing retention setting in this table. Covered by tests/unit/db-cleanup-conversation-nodes-12453.test.ts.",
+  "_rebaseline_2026_09_15_13609_combined_growth": "Combined growth of the 2026-09-15 maxmad64bis uplift batch (each PR rebaselined its own growth; the merged sum is larger): open-sse/services/accountFallback.ts->2507. Every hunk is flag-gated or a verified fix covered by that PR's tests; see the batch report.",
  "_rebaseline_2026_09_15_13602_combined_growth": "Combined growth of the 2026-09-15 maxmad64bis uplift batch (each PR rebaselined its own growth; the merged sum is larger): src/sse/handlers/chatHelpers.ts->1214. Every hunk is flag-gated or a verified fix covered by that PR's tests; see the batch report.",
  "_rebaseline_2026_09_15_13580_combined_growth": "Combined growth of the 2026-09-15 maxmad64bis uplift batch (each PR rebaselined its own growth; the merged sum is larger): src/sse/handlers/chatHelpers.ts->1202. Every hunk is flag-gated or a verified fix covered by that PR's tests; see the batch report.",
  "_rebaseline_2026_09_13_13581_pool_egress_observation": "PR #13581 own growth: src/app/(dashboard)/dashboard/settings/components/ProxyRegistryManager.tsx 1475->1477 (+2 = the PoolEgressObservation import and its one-line mount under the pool members label). The observation itself lives outside the frozen file, all under cap: PoolEgressObservation.tsx, the dedicated GET /api/settings/proxies/pool/egress-observation route, src/lib/proxyPoolEgressObservation.ts and getPoolEgressObservation in src/lib/db/proxyLogs.ts. Only the mount point is irreducible. Covered by tests/unit/proxy-pool-egress-observation.test.ts, tests/unit/proxy-pool-egress-observation-route.test.ts and tests/unit/ui/PoolEgressObservation.test.tsx.",
@@ -234,12 +239,13 @@
    "_rebaseline_2026_07_22_8213_combo_config_cooldown_wait_tests": "PR #8213 (hartmark, fix/gemini-tpm-quota-cooldown-wait) own growth: tests/unit/combo-config.test.ts 880->940 (+60, entirely this PR's diff — testFrozen add covering isComboCooldownWaitEligible (gating cooldown-wait to auto/quota-share strategies with the feature enabled) and resolveComboTargetTimeoutMsForCombo (raising the per-target timeout floor to cover the cooldown-wait budget + buffer for eligible strategies, fixing the 120s default cutting off a 130s wait early and returning a synthetic 524)). Covered by the new assertions themselves.",
    "_rebaseline_2026_07_23_8122_codex_image_edits": "#8122 (@xiaoyaner0201) own growth: tests/unit/image-generation-handler.test.ts 2019->2029 (+10) — new coverage for Codex reference image edits (POST /v1/images/edits) plus the sanitizeImageProviderError/redactSensitiveErrorText hardening it introduces. Test-only growth at the existing handler test file.",
    "_rebaseline_2026_07_25_8510_adobe_firefly_reference_images_tests": "#8510 (artickc, feat/adobe-firefly-reference-images) own test growth: tests/unit/adobe-firefly.test.ts 711->871 (+159, entirely this PR's diff — new referenceBlobs upload/dispatch coverage for handleAdobeFireflyImageGeneration, resolveAdobeSourceImageIds, and the storage-upload wire contract). Route-level /v1/images/edits coverage (credentials/rate-limit/4-ref-cap branches added to route.ts) lives in the new tests/unit/8510-adobe-firefly-edits-route.test.ts instead of growing this file further.",
+    "_rebaseline_2026_09_15_13748_13749_merged_test_growth_basereds": "Base-red drain (#12732): two security fixes merged on 2026-09-15 each grew a frozen test file with their own regression coverage, and PR-mode check:file-size does not relax testFrozen against the base, so every PR into release/v3.8.51 went red on file-size. Recorded against the merged state: tests/unit/image-generation-handler.test.ts 2133->2235 (#13748 public-only guard on client-supplied image URLs); tests/unit/batch_api.test.ts 1345->1348 (#13749 API-key ownership on files and batches). No cap is raised beyond the merged LOC.",
    "_rebaseline_2026_08_24_video_bridge_fu01_fu03_fu04_result_cache_tests": "PRs #11362 (FU-01 cache hardening) + #11382 (FU-03 visual dedup policy identity) + #11383 (FU-04 focused analysis mode) own test growth: videoBridgeResultCache.test.ts <1000->1040, +40 (sum of three stacked PRs boarded together in the same merge-batch, each adding its own cache-identity assertions on the shared result-cache seam). Owner pre-authorized rebaseline for legitimate PR growth (2026-08-19 directive).",
    "_rebaseline_basered_codebuddy_cn": "Base-red fix (#4664 CodeBuddy CN): oauth-providers-config.test.ts 867->870 (+3) to align the EXPECTED provider list/config with the codebuddy-cn provider that #4664 added to the registry without updating this test (it asserts 'exactly once').",
    "_rebaseline_pr4613_compatible_provider_groups": "Reconcile #4613 already-merged growth: providers-page-utils.test.ts 1004->1052 (+48, buildCompatibleProviderGroups partition unit test). Fast-gate PR->release does not run check:file-size, so this surfaced post-merge.",
    "tests/integration/chat-pipeline.test.ts": 1736,
    "tests/unit/account-fallback-service.test.ts": 2056,
-    "tests/unit/batch_api.test.ts": 1345,
+    "tests/unit/batch_api.test.ts": 1348,
    "tests/unit/cc-compatible-provider.test.ts": 1225,
    "tests/unit/chatcore-translation-paths.test.ts": 3447,
    "tests/unit/chatgpt-web.test.ts": 4911,
@@ -248,7 +254,7 @@
    "tests/unit/executor-codex.test.ts": 1465,
    "tests/unit/executor-default-base.test.ts": 1632,
    "tests/unit/grok-web.test.ts": 2985,
-    "tests/unit/image-generation-handler.test.ts": 2133,
+    "tests/unit/image-generation-handler.test.ts": 2235,
    "tests/unit/models-catalog-route.test.ts": 1653,
    "tests/unit/perplexity-web.test.ts": 1384,
    "tests/unit/provider-models-route.test.ts": 1783,
@@ -345,6 +351,7 @@
  "_rebaseline_2026_07_27_v3849_train1h": "Merge-train 1H (31 PRs) — owner-approved 2026-07-27. Two distinct causes, kept separate on purpose: (1) GENUINE irreducible growth at existing chokepoints — providerLimits/auth (#8632 Kimi quota-reset recovery), rateLimitManager (#8616 idle wedged limiters), models-catalog-route.test (#8610 OpenCode Go effort aliases); (2) COLLISION with #8585, which banked shrinks measured on the pre-train release tip while 30 sibling PRs in the SAME train grew those files again — chat/accountFallback (#8628), chatCore (#8613), videoGeneration (#8581), imageGeneration. The zero-headroom frozen entries cannot absorb either. Ceilings re-pinned to the post-merge tip; #8612 (also in this train) automates shrink-banking so this self-inflicted drift stops recurring. Detail: src/lib/usage/providerLimits.ts 1006->1013 (#8632); src/sse/services/auth.ts 2492->2508 (#8632); open-sse/services/rateLimitManager.ts 1014->1060 (#8616); src/sse/handlers/chat.ts 1842->1845 (#8628); open-sse/handlers/chatCore.ts 4939->4955 (#8613); open-sse/handlers/imageGeneration.ts 3100->3101 ((sem PR — teto do #8585)); open-sse/handlers/videoGeneration.ts 1038->1063 (#8581); open-sse/services/accountFallback.ts 1965->1966 (#8628); tests/unit/models-catalog-route.test.ts 1608->1636 (#8610)",
  "frozen": {
    "src/sse/handlers/chatHelpers.ts": 1214,
+    "_rebaseline_2026_09_15_13609_mistral_ambiguous_401": "PR #13609 rework (maxmad64bis, bare Mistral 401 soft lockout behind MISTRAL_AMBIGUOUS_401_SOFT_LOCKOUT, default off). open-sse/services/accountFallback.ts 2469->2501 (+32): +14 are the change itself (shared-predicate + flag imports, the documented ambiguousAuth field on the checkFallbackError return type, and the flag-gated 401 branch formatted normally instead of the PR's 139-char squeezed configuredRule line); +18 are the lint-staged prettier pass normalizing lines that were already unformatted on the release tip (multi-import, ISO_RETRY_RE, two regex arrays, persistAntigravityFamilyCooldownIfQuota call, applyErrorState guard, trailing commas) — pure formatting, no logic. src/sse/services/auth.ts 3556->3557 (+1): markAccountUnavailable passes connectionId to resolveTerminalConnectionStatus so the soft-strike bound is per connection. The predicate and strike tracker live in the leaf open-sse/services/accountFallback/mistralAmbiguousAuth.ts (under cap). Covered by tests/unit/provider-401-ambiguous-runtime.test.ts (flag off/on, end-to-end through markAccountUnavailable).",
    "_rebaseline_2026_06_22_4644_deepseek_web_tools": "PR #4644 (BugsBag/robust deepseek-web tool-call parsing): open-sse/executors/deepseek-web.ts 1117->1125 (+8). The new agentic tool-call path emits surrounding text + reasoning before tool_calls and swaps to the dedicated deepseekWebTools.ts parser; the +8 lines are cohesive wiring at the existing transformSSE chokepoint (the parser itself lives in the new deepseekWebTools.ts file, already under cap). The PR's own fast-gate (PR->release) does not run check:file-size, so this surfaced only at release reconcile. Covered by tests/unit/deepseek-web-tools-variants.test.ts + deepseek-web-tools-execute.test.ts.",
    "_rebaseline_2026_06_23_4712_deepseek_web_tool_results": "PR for #4712 (deepseek-web drops role:tool): open-sse/executors/deepseek-web.ts 1125->1148 (+23). messagesToPrompt() now folds role:\"tool\" results into the single-prompt transcript (recovering the tool name from the preceding assistant tool_calls by tool_call_id) instead of silently dropping them; the lines are cohesive wiring inside the existing function. Covered by tests/unit/deepseek-web-tool-result-prompt-4712.test.ts.",
    "_rebaseline_2026_06_24_headroom_strategy": "Headroom-aware connection selection (dario technique): combo.ts 3168->3180 (+12 = a new `else if (strategy === \"headroom\")` dispatch branch in handleComboChat that delegates to orderTargetsByHeadroom + its log line, plus the import). The actual logic lives OUT of the god-file: the pure ranker rankByHeadroom/computeHeadroom is the new leaf open-sse/services/combo/headroomRanking.ts (91 LOC, <cap) and the async orderer orderTargetsByHeadroom is appended to the existing open-sse/services/combo/quotaStrategies.ts (<cap) next to its sibling reset-aware/reset-window orderers (reuses their connection-expansion machinery). headroom = 1 - max(util_5h, util_7d) from getSaturation (src/lib/quota/saturationSignals.ts), prefers the connection with the most free capacity. Only the dispatch wiring is irreducible at the existing combo strategy chokepoint (mirrors the reset-aware/reset-window/context-optimized branches); not extractable without hiding the call site. fill-first stays default; all existing strategies untouched. Covered by tests/unit/combo-headroom-ranking.test.ts (pure helper) + tests/unit/combo-headroom-strategy.test.ts (orderer, saturation injected). Structural shrink of combo.ts tracked in #3501.",
@@ -445,9 +452,9 @@
    "_rebaseline_pr1043_minimax_tts": "Upstream port decolua/9router#1043 (toanalien) own growth: audioSpeech.ts 965->1061 (+96). Adds MiniMax T2A v2 TTS dispatch (handleMinimaxSpeech + hexToBytes helper) — provider entry was already in audioRegistry (format: minimax-tts) but no handler existed, falling through to the OpenAI-compatible default that fails (T2A has custom shape + hex-encoded audio + base_resp envelope). New branch sits next to the other inline provider branches (xiaomi-mimo, coqui, tortoise, aws-polly) — extracting would just create indirection. Covered by tests/unit/minimax-tts-1043.test.ts (3 tests, GREEN: success, base_resp error, invalid-hex).",
    "_rebaseline_pr4592_exclude_exhausted_auto": "Reconcile #4592 already-merged growth: combo.ts 2991->3036 (+45, terminal-status quota-cutoff exclusion in buildAutoCandidates + opt-in gate). Fast-gate PR->release does not run check:file-size.",
    "open-sse/executors/antigravity.ts": 1665,
-    "open-sse/executors/base.ts": 1753,
+    "open-sse/executors/base.ts": 1754,
    "open-sse/executors/chatgpt-web.ts": 5056,
-    "open-sse/executors/codex.ts": 1505,
+    "open-sse/executors/codex.ts": 1528,
    "open-sse/executors/cursor.ts": 1759,
    "open-sse/executors/muse-spark-web.ts": 1405,
    "open-sse/handlers/chatCore.ts": 6146,
@@ -455,13 +462,13 @@
    "open-sse/handlers/search.ts": 1789,
    "open-sse/mcp-server/schemas/tools.ts": 1621,
    "open-sse/mcp-server/server.ts": 1572,
-    "open-sse/services/accountFallback.ts": 2493,
+    "open-sse/services/accountFallback.ts": 2507,
    "open-sse/services/adobeFireflyBrowserLogin.ts": 1401,
    "open-sse/services/combo.ts": 4080,
    "open-sse/services/combo/executeTargetAttempt.ts": 1228,
    "open-sse/translator/response/openai-responses.ts": 1466,
    "open-sse/utils/cursorAgentProtobuf.ts": 1547,
-    "open-sse/utils/proxyFetch.ts": 1275,
+    "open-sse/utils/proxyFetch.ts": 1296,
    "open-sse/utils/stream.ts": 3098,
    "open-sse/vendor/codex-chatgpt-web/adapters/chatgpt-web/browser-worker.ts": 4398,
    "open-sse/vendor/codex-chatgpt-web/bridge.ts": 1335,
@@ -476,7 +483,7 @@
    "src/app/(dashboard)/dashboard/settings/components/ProxyRegistryManager.tsx": 1477,
    "src/app/(dashboard)/dashboard/settings/components/ResilienceTab.tsx": 1271,
    "src/app/(dashboard)/dashboard/settings/components/RoutingTab.tsx": 1607,
-    "src/app/(dashboard)/dashboard/settings/components/SystemStorageTab.tsx": 1597,
+    "src/app/(dashboard)/dashboard/settings/components/SystemStorageTab.tsx": 1598,
    "src/app/(dashboard)/dashboard/usage/components/EvalsTab.tsx": 2152,
    "src/app/api/providers/[id]/models/route.ts": 2432,
    "src/app/api/providers/[id]/test/route.ts": 1252,
@@ -491,12 +498,13 @@
    "src/shared/constants/providers/apikey/gateways.ts": 1502,
    "src/shared/services/cliRuntime.ts": 1296,
    "src/sse/handlers/chat.ts": 2490,
-    "src/sse/services/auth.ts": 3556,
+    "src/sse/services/auth.ts": 3557,
    "tests/unit/account-fallback-service.test.ts": 2453,
    "tests/unit/provider-validation-specialty.test.ts": 4656,
    "open-sse/services/autoCombo/virtualFactory.ts": 1230,
    "open-sse/services/combo/roundRobinCombo.ts": 1213
  },
+  "_rebaseline_2026_09_15_roundrobin_dashboard_events": "Fix #13089 (Combo Studio Live dashboard shows an empty backlog for round-robin combos): open-sse/services/combo/roundRobinCombo.ts 1205->1213. Round-robin is the only combo strategy that bypasses handleComboChat/executeTargetAttempt.ts, the path that publishes the combo.target.attempt/succeeded/failed EventBus events the Live dashboard listens for — so round-robin completions never showed up. The new call-site wiring (createRRDashboardEvents(...) instantiated once per target, one-line .attempt()/.succeeded()/.failed() calls at the 6 existing dispatch/outcome points) is the emitter logic actually extracted into a new module, open-sse/services/combo/rrDashboardEvents.ts — this is the minimum irreducible footprint for wiring 6 required call sites into 6 fixed control-flow points of the frozen file. Covered by tests/unit/issue-13089-roundrobin-live-ws-events.test.ts (2 tests: success + failure paths).",
  "_rebaseline_base_2026_08_10_proxyfetch": "Base-red fix (green-prs sweep, issue #9985): open-sse/utils/proxyFetch.ts 1207 > cap 1000 — new proxied-TLS fetch helper introduced by the Fal reference-image work. Owner-authorized quick rebaseline to green; structural slim tracked for v3.9.0.",
  "_rebaseline_2026_07_27_v3849_train2": "Merge-train 2 (7 PRs) — owner-approved 2026-07-27. Single entry: chatCore.ts 4955->5006 (#8595, Responses multi-turn image compaction before the context hard-reject). Genuine irreducible growth at the existing compaction chokepoint in handleChatCore — the PR adds a last-resort retry against the concrete budget plus the estimateFinalInputTokens helper, both wired at the pre-existing call site rather than a new branch. Covered by tests/unit/8560-responses-image-compaction.test.ts (4 tests).",
  "_rebaseline_2026_07_27_v3849_train3": "Merge-train 3 (13 PRs) — owner-approved 2026-07-27. Both entries are genuine irreducible growth at existing chokepoints, not new branches: src/lib/db/apiKeys.ts 1518->1529 (#8805 cx/* ≡ codex/* API-key model permissions); open-sse/handlers/chatCore.ts 5006->5020 (#8806 real response payload into plugin onResponse hooks). Covered by tests/unit/db-apiKeys-crud.test.ts (4 new cases) and the two plugin-hook test files updated in #8806 respectively.",
--- a/config/quality/i18n-translation-baseline.json
+++ b/config/quality/i18n-translation-baseline.json
@@ -2,70 +2,70 @@
  "_comment": "Catraca de tradução real (valor idêntico ao en.json, placeholder ou ausente, fora do allowlist untranslatable-keys.json) em % por locale. Só pode cair. Atualize via `npm run i18n:check-ratio:update` quando um locale melhora. Valores medidos, nunca chutados.",
  "slack": 0.5,
  "locales": {
-    "am": 1.8,
-    "ar": 4,
-    "az": 48.8,
-    "bg": 24.7,
-    "bn": 44.1,
-    "cs": 22.7,
-    "da": 27.9,
-    "de": 26,
-    "el": 3.2,
-    "es": 56.2,
-    "et": 2.4,
-    "fa": 43.8,
-    "fi": 25.1,
-    "fr": 25.9,
-    "ga": 2.7,
-    "gu": 43.5,
-    "ha": 2.8,
-    "he": 25.4,
-    "hi": 25.8,
-    "hr": 3.5,
-    "hu": 27,
-    "hy": 1.6,
-    "id": 27.2,
-    "ig": 2.8,
-    "it": 26.9,
-    "ja": 25.6,
-    "ka": 1.7,
-    "km": 2.1,
-    "kn": 2,
-    "ko": 26.8,
-    "lt": 2.1,
-    "lv": 2.6,
-    "ml": 1.9,
-    "mr": 44,
-    "ms": 26.8,
-    "mt": 3.5,
-    "my": 3.5,
-    "ne": 1.8,
-    "nl": 28.6,
-    "no": 28.4,
-    "or": 1.8,
-    "pa": 1.8,
-    "phi": 33.4,
-    "pl": 10.5,
-    "pt": 5.5,
-    "pt-BR": 16.9,
-    "ro": 28.4,
-    "ru": 20.3,
-    "si": 1.9,
-    "sk": 27.3,
-    "sl": 2.4,
-    "sr": 2.8,
-    "sv": 27.6,
-    "sw": 44,
-    "ta": 43.7,
-    "te": 43.2,
-    "th": 25.4,
-    "tr": 22.4,
-    "uk-UA": 23.4,
-    "ur": 43.6,
-    "uz": 2.7,
-    "vi": 4.9,
-    "yo": 2.1,
-    "zh-CN": 3.7,
-    "zh-TW": 4.3
+    "am": 1.4,
+    "ar": 0.9,
+    "az": 2.2,
+    "bg": 1.2,
+    "bn": 1.2,
+    "cs": 2.1,
+    "da": 3.6,
+    "de": 3,
+    "el": 1.5,
+    "es": 2.1,
+    "et": 1.8,
+    "fa": 1.1,
+    "fi": 1.5,
+    "fr": 3.4,
+    "ga": 1.5,
+    "gu": 1.2,
+    "ha": 1.5,
+    "he": 1.1,
+    "hi": 1,
+    "hr": 2.2,
+    "hu": 1.6,
+    "hy": 1.3,
+    "id": 2.5,
+    "ig": 1.6,
+    "it": 2.6,
+    "ja": 1.1,
+    "ka": 1.4,
+    "km": 1.4,
+    "kn": 1.3,
+    "ko": 1.3,
+    "lt": 1.5,
+    "lv": 1.6,
+    "ml": 1.4,
+    "mr": 1.3,
+    "ms": 2.4,
+    "mt": 2.3,
+    "my": 1.5,
+    "ne": 1.4,
+    "nl": 3.8,
+    "no": 2.5,
+    "or": 1.4,
+    "pa": 1.4,
+    "phi": 3.2,
+    "pl": 2.4,
+    "pt": 2,
+    "pt-BR": 2.5,
+    "ro": 2.7,
+    "ru": 1,
+    "si": 1.4,
+    "sk": 2,
+    "sl": 1.8,
+    "sr": 1.4,
+    "sv": 2.7,
+    "sw": 1.5,
+    "ta": 1.3,
+    "te": 1.2,
+    "th": 1.1,
+    "tr": 1.6,
+    "uk-UA": 1.2,
+    "ur": 1.2,
+    "uz": 2.2,
+    "vi": 1.5,
+    "yo": 1.4,
+    "zh-CN": 1,
+    "zh-TW": 1.1
  }
 }
--- a/config/quality/release-acceptance.schema.json
+++ b/config/quality/release-acceptance.schema.json
@@ -0,0 +1,183 @@
+{
+  "$id": "https://omniroute.local/quality/release-acceptance.schema.json",
+  "title": "Release acceptance report",
+  "type": "object",
+  "additionalProperties": false,
+  "required": [
+    "schema_version",
+    "identity",
+    "required_gates",
+    "gates",
+    "evidence_errors",
+    "verdict",
+    "artifact"
+  ],
+  "properties": {
+    "schema_version": { "type": "integer", "const": 1 },
+    "identity": { "$ref": "#/$defs/identity" },
+    "required_gates": {
+      "type": "array",
+      "uniqueItems": true,
+      "items": { "$ref": "#/$defs/gateInstanceKey" }
+    },
+    "gates": {
+      "type": "array",
+      "items": { "$ref": "#/$defs/gateResult" }
+    },
+    "evidence_errors": {
+      "type": "array",
+      "items": { "$ref": "#/$defs/evidenceError" }
+    },
+    "verdict": { "enum": ["VERIFIED", "FAILED", "UNVERIFIED"] },
+    "artifact": {
+      "anyOf": [
+        { "type": "null" },
+        { "$ref": "#/$defs/artifact" }
+      ]
+    }
+  },
+  "allOf": [
+    {
+      "if": { "properties": { "verdict": { "const": "VERIFIED" } }, "required": ["verdict"] },
+      "then": { "properties": { "required_gates": { "minItems": 1 } } }
+    }
+  ],
+  "$defs": {
+    "sha": {
+      "type": "string",
+      "pattern": "^[0-9a-f]{40}$"
+    },
+    "digest": {
+      "type": "string",
+      "pattern": "^[0-9a-f]{64}$"
+    },
+    "gateInstanceKey": {
+      "type": "object",
+      "additionalProperties": false,
+      "required": ["gate_id", "suite_id", "shard_index", "shard_total"],
+      "properties": {
+        "gate_id": { "type": "string", "minLength": 1 },
+        "suite_id": { "type": ["string", "null"] },
+        "shard_index": { "type": ["integer", "null"], "minimum": 0 },
+        "shard_total": { "type": ["integer", "null"], "minimum": 1 }
+      }
+    },
+    "identity": {
+      "type": "object",
+      "additionalProperties": false,
+      "required": [
+        "repository",
+        "run_id",
+        "run_attempt",
+        "workflow",
+        "trigger",
+        "scope",
+        "requested_ref",
+        "base_sha",
+        "candidate_sha",
+        "tested_sha"
+      ],
+      "properties": {
+        "repository": { "type": "string", "minLength": 1 },
+        "run_id": { "type": "string", "minLength": 1 },
+        "run_attempt": { "type": "integer", "minimum": 1 },
+        "workflow": { "type": "string", "minLength": 1 },
+        "trigger": { "type": "string", "minLength": 1 },
+        "scope": { "enum": ["pr", "release", "scheduled"] },
+        "requested_ref": { "type": "string", "minLength": 1 },
+        "base_sha": { "$ref": "#/$defs/sha" },
+        "candidate_sha": { "$ref": "#/$defs/sha" },
+        "tested_sha": { "$ref": "#/$defs/sha" }
+      }
+    },
+    "evidenceRef": {
+      "type": "object",
+      "additionalProperties": false,
+      "required": ["artifact_id", "member", "algorithm", "digest"],
+      "properties": {
+        "artifact_id": { "type": "string", "minLength": 1 },
+        "member": {
+          "type": "string",
+          "minLength": 1,
+          "pattern": "^(?!/)(?!.*(?:^|[/\\\\])\\.\\.(?:[/\\\\]|$))[^\\s]+$"
+        },
+        "algorithm": { "const": "sha256" },
+        "digest": { "$ref": "#/$defs/digest" }
+      }
+    },
+    "artifact": {
+      "type": "object",
+      "additionalProperties": false,
+      "required": ["algorithm", "digest", "identity"],
+      "properties": {
+        "algorithm": { "const": "sha256" },
+        "digest": { "$ref": "#/$defs/digest" },
+        "identity": { "type": "string", "minLength": 1 }
+      }
+    },
+    "evidenceError": {
+      "type": "object",
+      "additionalProperties": false,
+      "required": ["code", "gate", "detail"],
+      "properties": {
+        "code": { "type": "string", "minLength": 1 },
+        "gate": { "$ref": "#/$defs/gateInstanceKey" },
+        "detail": { "type": "string" }
+      }
+    },
+    "gateResult": {
+      "type": "object",
+      "additionalProperties": false,
+      "required": [
+        "gate_id",
+        "suite_id",
+        "shard_index",
+        "shard_total",
+        "tested_sha",
+        "run_id",
+        "run_attempt",
+        "command_id",
+        "gate_type",
+        "status",
+        "cause",
+        "exit_code",
+        "duration_ms",
+        "evidence"
+      ],
+      "properties": {
+        "gate_id": { "type": "string", "minLength": 1 },
+        "suite_id": { "type": ["string", "null"] },
+        "shard_index": { "type": ["integer", "null"], "minimum": 0 },
+        "shard_total": { "type": ["integer", "null"], "minimum": 1 },
+        "tested_sha": { "$ref": "#/$defs/sha" },
+        "run_id": { "type": "string", "minLength": 1 },
+        "run_attempt": { "type": "integer", "minimum": 1 },
+        "command_id": { "type": "string", "minLength": 1 },
+        "gate_type": { "enum": ["static", "test", "artifact"] },
+        "status": { "enum": ["PASS", "FAIL", "INFRA_ERROR", "SKIPPED"] },
+        "reason": { "type": "string", "minLength": 1 },
+        "cause": {
+          "anyOf": [
+            { "type": "null" },
+            { "$ref": "#/$defs/gateInstanceKey" }
+          ]
+        },
+        "exit_code": { "type": ["integer", "null"] },
+        "duration_ms": { "type": "integer", "minimum": 0 },
+        "evidence": {
+          "type": "array",
+          "items": { "$ref": "#/$defs/evidenceRef" }
+        }
+      },
+      "allOf": [
+        {
+          "if": {
+            "properties": { "status": { "const": "SKIPPED" } },
+            "required": ["status"]
+          },
+          "then": { "required": ["reason"] }
+        }
+      ]
+    }
+  }
+}
--- a/contrib/podman/README.md
+++ b/contrib/podman/README.md
@@ -40,7 +40,28 @@ cp contrib/podman/*.network ~/.config/containers/systemd/omniroute/
 cp contrib/podman/*.volume ~/.config/containers/systemd/omniroute/
 ```

-### 3. Mount the project .env for secrets
+### 3. Generate secrets before first start
+
+`omniroute.container` no longer ships `JWT_SECRET` / `API_KEY_SECRET` /
+`INITIAL_PASSWORD` values — earlier versions shipped copy-pasteable
+placeholders (`change-me-to-a-random-base64-string`,
+`change-me-to-a-random-hex-string`) that an operator could forget to
+rotate, leaving the deployment with a public, guessable secret/password
+(#13679). Generate real ones and put them in your project `.env`:
+
+```bash
+echo "JWT_SECRET=$(openssl rand -base64 48)" >> .env
+echo "API_KEY_SECRET=$(openssl rand -hex 32)" >> .env
+echo "INITIAL_PASSWORD=$(openssl rand -hex 24)" >> .env
+```
+
+If you skip this: `JWT_SECRET`/`API_KEY_SECRET` are auto-generated and
+persisted on first boot, and the dashboard requires setup from `localhost`
+before it accepts any password — safer than a literal default, but a real
+`INITIAL_PASSWORD` is still recommended so a non-interactive first boot has
+a known credential to log in with.
+
+### 4. Mount the project .env for secrets

 Edit `~/.config/containers/systemd/omniroute/omniroute.container` and
 uncomment/replace the `EnvironmentFile` line with the absolute path to
@@ -54,7 +75,7 @@ Make sure `CONTAINER_HOST=podman` is set in that `.env`.

 Alternatively, edit the env vars directly in the `.container` file.

-### 4. Reload systemd and start
+### 5. Reload systemd and start

 ```bash
 systemctl --user daemon-reload
@@ -62,7 +83,7 @@ systemctl --user start omniroute-redis
 systemctl --user start omniroute
 ```

-### 5. Verify
+### 6. Verify

 ```bash
 systemctl --user status omniroute
--- a/contrib/podman/omniroute.container
+++ b/contrib/podman/omniroute.container
@@ -28,14 +28,23 @@ Environment=DASHBOARD_PORT=20128
 Environment=API_PORT=20129
 Environment=API_HOST=0.0.0.0
 Environment=REDIS_URL=redis://redis:6379
-Environment=JWT_SECRET=change-me-to-a-random-base64-string
-Environment=API_KEY_SECRET=change-me-to-a-random-base64-string
-Environment=INITIAL_PASSWORD=change-me-to-a-random-hex-string
 Environment=NODE_ENV=production
 Environment=REQUIRE_API_KEY=true

-# Load additional secrets (API keys, OAuth creds) from the project .env:
+# JWT_SECRET, API_KEY_SECRET and INITIAL_PASSWORD are deliberately NOT set here.
+# This unit used to ship copy-pasteable "replace-me" placeholder literals — an
+# operator who forgot to replace them ran production with a public, guessable
+# secret and dashboard password (#13679). Generate real values and load them from
+# your project .env before the FIRST start — see "Generate secrets before first
+# start" in contrib/podman/README.md — by uncommenting and pointing this at your
+# project .env:
 # EnvironmentFile=%h/code/docker/OmniRoute/.env
+#
+# If left unset: JWT_SECRET and API_KEY_SECRET are auto-generated and persisted
+# on first boot, and the dashboard requires setup from localhost before it
+# accepts any password (see managementPassword.ts / apiAuth.ts) — safer than a
+# known-literal default either way, but a real INITIAL_PASSWORD is still
+# recommended for non-interactive first boots.

 HealthCmd=node /app/healthcheck.mjs
 HealthInterval=30s
--- a/docker-compose.yml
+++ b/docker-compose.yml
@@ -142,11 +142,24 @@ services:
      - "${APP_BIND_HOST:-127.0.0.1}:${DASHBOARD_PORT:-20128}:${DASHBOARD_PORT:-20128}"
      - "${APP_BIND_HOST:-127.0.0.1}:${API_PORT:-20129}:${API_PORT:-20129}"
      - "${APP_BIND_HOST:-127.0.0.1}:${LIVE_WS_PORT:-20132}:${LIVE_WS_PORT:-20132}"
+    # SECURITY (#13679): joins BOTH `default` (to keep reaching redis and the
+    # other sidecars) AND the dedicated `chatgpt-web-codex-net` (the one
+    # legitimate consumer of the CDP proxy below).
+    networks:
+      - default
+      - chatgpt-web-codex-net
    profiles:
      - web

  # Internal-only Chromium runtime for ChatGPT Web (Codex). No CDP or browser
  # UI port is published to the host.
+  #
+  # SECURITY (#13679): isolated onto its own `chatgpt-web-codex-net` network
+  # instead of the shared implicit default bridge — its cdp-proxy.mjs
+  # sidecar republishes Chromium's CDP on 0.0.0.0:9223, and CDP grants full
+  # control over a live browser session. Without this isolation, any
+  # compromised sibling container (redis, qdrant, bifrost, cliproxyapi,
+  # codex-app-server, ...) on the default network could reach it.
  chatgpt-web-codex-browser:
    build:
      context: .
@@ -154,8 +167,12 @@ services:
    image: omniroute:chatgpt-web-codex-browser
    restart: unless-stopped
    shm_size: "2gb"
+    environment:
+      - CDP_PROXY_TOKEN=${CDP_PROXY_TOKEN:-}
    volumes:
      - chatgpt-web-codex-browser-data:/browser-profile
+    networks:
+      - chatgpt-web-codex-net
    profiles:
      - web

@@ -370,7 +387,12 @@ services:
    # compose network by the omniroute app.
    healthcheck:
      test:
-        ["CMD", "node", "-e", "require('http').get('http://127.0.0.1:1456/readyz',r=>process.exit(r.statusCode===200?0:1)).on('error',()=>process.exit(1))"]
+        [
+          "CMD",
+          "node",
+          "-e",
+          "require('http').get('http://127.0.0.1:1456/readyz',r=>process.exit(r.statusCode===200?0:1)).on('error',()=>process.exit(1))",
+        ]
      interval: 30s
      timeout: 5s
      retries: 3
@@ -378,6 +400,13 @@ services:
    profiles:
      - codex-app-server

+networks:
+  # SECURITY (#13679): dedicated network for the unauthenticated-by-default
+  # CDP proxy sidecar (docker/chatgpt-web-codex-browser/cdp-proxy.mjs) —
+  # shared only with omniroute-web, not with redis/qdrant/bifrost/cliproxyapi/
+  # codex-app-server or any other sibling on the implicit default network.
+  chatgpt-web-codex-net: {}
+
 volumes:
  chatgpt-web-codex-browser-data:
    name: omniroute-chatgpt-web-codex-browser-data
--- a/docker/chatgpt-web-codex-browser/cdp-proxy.mjs
+++ b/docker/chatgpt-web-codex-browser/cdp-proxy.mjs
@@ -5,6 +5,31 @@ const listenPort = 9223;
 const upstreamHost = "127.0.0.1";
 const upstreamPort = 9222;

+// SECURITY (#13679): this proxy republishes Chromium's loopback CDP onto
+// 0.0.0.0:9223 with no auth of its own — CDP grants full control over a
+// live browser session (Runtime.evaluate, cookie theft, etc). When the
+// operator sets CDP_PROXY_TOKEN, every request/WS-upgrade MUST present it as
+// an `X-Omni-Cdp-Token: <token>` header before a single byte is forwarded
+// upstream, mirroring the gate docker/vnc-browser/chromium/cdp-bridge.py
+// already has (#12571). Left unset, the proxy keeps its historical
+// zero-config behavior — the primary mitigation for the shared-bridge risk
+// is docker-compose.yml isolating this service onto its own network so no
+// unrelated sibling container can reach it at all.
+const TOKEN = process.env.CDP_PROXY_TOKEN || "";
+const TOKEN_HEADER = "x-omni-cdp-token";
+
+if (!TOKEN) {
+  console.error(
+    "[cdp-proxy] WARNING: running without CDP_PROXY_TOKEN — every request is forwarded " +
+      "unauthenticated. Set CDP_PROXY_TOKEN to require an X-Omni-Cdp-Token header (#13679)."
+  );
+}
+
+function hasValidToken(headers) {
+  if (!TOKEN) return true;
+  return headers[TOKEN_HEADER] === TOKEN;
+}
+
 function proxyHeaders(headers) {
  const next = { ...headers, host: `${upstreamHost}:${upstreamPort}` };
  delete next.connection;
@@ -13,6 +38,11 @@ function proxyHeaders(headers) {
 }

 const server = http.createServer((request, response) => {
+  if (!hasValidToken(request.headers)) {
+    response.writeHead(403, { "content-type": "application/json" });
+    response.end(JSON.stringify({ error: "missing or invalid X-Omni-Cdp-Token" }));
+    return;
+  }
  const upstream = http.request(
    {
      host: upstreamHost,
@@ -48,6 +78,10 @@ const server = http.createServer((request, response) => {
 });

 server.on("upgrade", (request, socket, head) => {
+  if (!hasValidToken(request.headers)) {
+    socket.destroy();
+    return;
+  }
  const upstream = net.connect(upstreamPort, upstreamHost, () => {
    const upgradeHeaders = {
      ...request.headers,
@@ -69,4 +103,6 @@ server.on("upgrade", (request, socket, head) => {
  upstream.on("error", () => socket.destroy());
 });

-server.listen(listenPort, "0.0.0.0");
+server.listen(listenPort, "0.0.0.0", () => {
+  console.error(`[cdp-proxy] listening on 0.0.0.0:${listenPort}`);
+});
--- a/docker/vnc-browser/chromium/svc-de-run
+++ b/docker/vnc-browser/chromium/svc-de-run
@@ -10,7 +10,13 @@ if [[ "${PIXELFLUX_WAYLAND,,}" == "true" ]]; then
  echo "[svc-de] ${SOCKET_PATH} found launching de"
  cd $HOME
  # OmniRoute: bridge Chromium DevTools (127.0.0.1:9222) to 0.0.0.0:9223.
-  ( sleep 8; python3 /usr/local/bin/cdp-bridge.py >/proc/1/fd/2 2>&1 ) &
+  # SECURITY (#13679): only start the bridge when CDP_BRIDGE_TOKEN is
+  # configured — cdp-bridge.py already fails closed for every caller when
+  # it is unset (#12571), so an unconfigured container gains nothing by
+  # running an always-listening 0.0.0.0:9223 process anyway.
+  if [ -n "${CDP_BRIDGE_TOKEN:-}" ]; then
+    ( sleep 8; python3 /usr/local/bin/cdp-bridge.py >/proc/1/fd/2 2>&1 ) &
+  fi
  exec s6-setuidgid abc \
    /bin/bash /defaults/startwm_wayland.sh &
  PID=$!
@@ -57,7 +63,13 @@ chmod 777 /tmp/selkies*
 # run
 cd $HOME
 # OmniRoute: bridge Chromium DevTools (127.0.0.1:9222) to 0.0.0.0:9223.
-( sleep 8; python3 /usr/local/bin/cdp-bridge.py >/proc/1/fd/2 2>&1 ) &
+# SECURITY (#13679): only start the bridge when CDP_BRIDGE_TOKEN is
+# configured — cdp-bridge.py already fails closed for every caller when it
+# is unset (#12571), so an unconfigured container gains nothing by running
+# an always-listening 0.0.0.0:9223 process anyway.
+if [ -n "${CDP_BRIDGE_TOKEN:-}" ]; then
+  ( sleep 8; python3 /usr/local/bin/cdp-bridge.py >/proc/1/fd/2 2>&1 ) &
+fi
 exec s6-setuidgid abc \
  /bin/bash /defaults/startwm.sh &
 PID=$!
--- a/Show More
+++ b/Show More
				`@@ -0,0 +1 @@`
				- feat(i18n): new blocking gate `i18n:check-keys` (`scripts/i18n/check-key-completeness.mjs`) — every locale catalog must carry exactly the key set of `en.json`, whatever the age of the key; the percentage and new-key gates let batch 1 (#13044) ship 43 keys short and batch 2 (#13660) 10 keys short. The i18n guide now documents the post-merge re-sync and the retranslation flow. (#13827)
				`@@ -0,0 +1 @@`
				- feat(providers): Added Agnes AI (China) as `agnes-cn` pointed at `https://api.agnes-ai.cn/v1`. Keys issued for `apihub.agnes-ai.com` stay on the existing `agnes` card. Live `/v1/models` on that host lists `agnes-3.0-flash` (same id as intl); the CN seed matches 2.0/2.5/3.0 and not retired 1.5.
				`@@ -0,0 +1 @@`
				- feat(providers): share the existing `api.x.ai/v1/models` discovery config with `xai-oauth` so SuperGrok OAuth connections pick up new Grok ids without a registry seed edit.
				`@@ -0,0 +1 @@`
				- fix(api): restore the `name` field on non-streaming `/v1/responses` `function_call` output items — a plain (non-namespace) tool call's identity restore was blindly applying the `_toolNameMap` alias-table fallback as a `{namespace, name}` object, silently blanking `name` to `undefined` (dropped entirely by JSON.stringify) and leaving Codex unable to dispatch the call, so it re-narrated its intent in a loop instead (#12370)
				`@@ -0,0 +1 @@`
				- fix(db): give `conversation_turn_nodes` its own independent retention knob (`retention.conversationTurnNodes`, default 30 days — matching `callLogs` so upgrading changes nothing until an operator overrides it) instead of sharing `callLogs`, and sweep orphaned `agentic_conversations` after the nodes expire (#12453).
				`@@ -0,0 +1 @@`
				`- fix(sse): Codex WebSocket transport (including the app-server) no longer fails to load in the Next.js standalone Docker runtime — the wreq-js loader now resolves its module name dynamically instead of a literal Turbopack could rewrite to an unreachable build-time symlink (#12491) — thanks @marshalfevzi`
				`@@ -0,0 +1 @@`
				- fix(providers): xAI requests no longer silently drop an assistant tool call sent in the legacy OpenAI `function_call` shape (instead of `tool_calls[]`) — the call is now translated into the xAI request the same way modern tool calls are (#12692) — thanks @soroush5
				`@@ -0,0 +1 @@`
				- fix(providers): xAI responses no longer report `total_tokens`/`totalTokenCount` as `0` when upstream usage uses the legacy `prompt_tokens`/`completion_tokens` names instead of `input_tokens`/`output_tokens` (#12700) — thanks @soroush5
				`@@ -0,0 +1 @@`
				- fix(resilience): a recoverable direct-fetch response-start timeout (`DIRECT_RESPONSE_START_TIMEOUT`) could, in a narrow timer/promise-settlement race, escape as an `unhandledRejection` → `uncaughtException` and kill the server process — even though `proxyFetch` already retries this exact condition on a fresh socket. Guarded the timer callback so it can no longer fire against an already-settled attempt, and extended the process-level crash guard (already used by the WS/API-bridge servers) to recognize and swallow this code if it ever escapes anyway. Also installs that same guard in the production server entrypoint (`dist/server-ws.mjs`), which never had it even though the dev server already did ([#12861](https://github.com/diegosouzapw/OmniRoute/issues/12861)) — thanks @insoln
				`@@ -0,0 +1 @@`
				`- fix(providers): correct Magnific API key validation, which reported every valid key as invalid due to a GET probe against a POST-only endpoint (#12927) — thanks @hubo1989`