fix(security): embed the Raycast signature secret via resolvePublicCred (HR#11)

The secret-scan ratchet only ran once the earlier Fast Quality Gates steps
stopped failing, and it lands at 1 finding vs baseline 0.

The finding is open-sse/services/raycast.ts:19 —
RAYCAST_DEFAULT_SIG_SECRET, a 64-hex request-signature secret that #8895
committed as a bare string literal. It is genuinely public (community-extracted
from the Raycast macOS client; the SAME value ships to every install, it is not
a per-user credential), which is exactly the category Hard Rule #11 governs:
public upstream credentials MUST go through resolvePublicCred()
(open-sse/utils/publicCreds.ts), never a literal — see
docs/security/PUBLIC_CREDS.md.

So the fix is the mandated pattern, not a .gitleaks.toml allowlist entry: added
`raycast_sig_secret` to EMBEDDED_DEFAULTS as the XOR-masked byte sequence and
resolved it with the existing RAYCAST_SIG_SECRET env override. The
providerSpecificData.sigSecret override is untouched. Verified the decoded value
is byte-identical to the literal it replaces.

check:secrets secretFindings 1 -> 0. check:public-creds exit 0.
publicCreds 12/12, raycast-auth 6/6, raycast-local-extract 1/1,
trae-publiccred 3/3. typecheck:core exit 0.

Refs #9298

AGENTS.md

@@ -58,7 +58,7 @@ Repository map and Reference Documentation sections below.
 | Services      | `open-sse/services/`    | Combo routing, rate limits, caching, etc                                                                                                                |
 | Database      | `src/lib/db/`           | SQLite domain modules (130 migrations)                                                                                                                  |
 | Domain/Policy | `src/domain/`           | Policy engine, cost rules, fallback logic                                                                                                               |
-| MCP Server    | `open-sse/mcp-server/`  | 104 tools (42 base + memory/skill/agentSkill/pool/notion/obsidian/gamification/plugin modules), 3 transports (stdio / SSE / Streamable HTTP), 31 scopes |
+| MCP Server    | `open-sse/mcp-server/`  | 105 tools (42 base + memory/skill/agentSkill/pool/notion/obsidian/gamification/plugin modules), 3 transports (stdio / SSE / Streamable HTTP), 31 scopes |
 | A2A Server    | `src/lib/a2a/`          | JSON-RPC 2.0 agent protocol                                                                                                                             |
 | Skills        | `src/lib/skills/`       | Extensible skill framework                                                                                                                              |
 | Memory        | `src/lib/memory/`       | Persistent conversational memory                                                                                                                        |

README.md

@@ -188,7 +188,7 @@ curl http://localhost:20128/v1/chat/completions \
 
 </div>
 
-<img src="./docs/diagrams/promise-pillars.svg" width="100%" alt="The Promise — One endpoint. 291 providers. Never stop building — OmniRoute picks the cheapest one that works. Six pillars: Never hit limits (auto-fallback across 291 providers in milliseconds, zero downtime) · Save up to 95% tokens (RTK + Caveman stacked compression cuts 15–95%, ~89% avg on tool-heavy sessions) · $0 to start (90+ free tiers, 40+ free forever — no card needed) · Every tool works (33 coding agents through one config) · One endpoint (OpenAI ↔ Claude ↔ Gemini ↔ Responses API at /v1) · Production-grade (circuit breakers, TLS stealth, MCP 104 tools, A2A, memory, guardrails, evals — 25,000+ tests)."/>
+<img src="./docs/diagrams/promise-pillars.svg" width="100%" alt="The Promise — One endpoint. 291 providers. Never stop building — OmniRoute picks the cheapest one that works. Six pillars: Never hit limits (auto-fallback across 291 providers in milliseconds, zero downtime) · Save up to 95% tokens (RTK + Caveman stacked compression cuts 15–95%, ~89% avg on tool-heavy sessions) · $0 to start (90+ free tiers, 40+ free forever — no card needed) · Every tool works (33 coding agents through one config) · One endpoint (OpenAI ↔ Claude ↔ Gemini ↔ Responses API at /v1) · Production-grade (circuit breakers, TLS stealth, MCP 105 tools, A2A, memory, guardrails, evals — 25,000+ tests)."/>
 
 <br/>
 <br/>
@@ -439,7 +439,7 @@ All **19** strategies — mix & match per combo step:
 
 </div>
 
-<img src="./docs/diagrams/comparison-table.svg" width="100%" alt="What sets OmniRoute apart — comparison table vs 9router, OpenRouter, CLIProxyAPI and LiteLLM across 13 capabilities. OmniRoute: 291 providers, 90+ free providers built-in, 19 routing strategies, 12-engine token compression, built-in MCP server with 104 tools, A2A agent protocol, persistent memory, guardrails, cloud agents, TLS fingerprint stealth, Desktop/Termux/PWA, 43 i18n UI locales, 100% MIT self-hosted. OmniRoute is the only one with the full set; competitors show a mix of checks, partials and crosses. Verified from each project&apos;s docs."/>
+<img src="./docs/diagrams/comparison-table.svg" width="100%" alt="What sets OmniRoute apart — comparison table vs 9router, OpenRouter, CLIProxyAPI and LiteLLM across 13 capabilities. OmniRoute: 291 providers, 90+ free providers built-in, 19 routing strategies, 12-engine token compression, built-in MCP server with 105 tools, A2A agent protocol, persistent memory, guardrails, cloud agents, TLS fingerprint stealth, Desktop/Termux/PWA, 43 i18n UI locales, 100% MIT self-hosted. OmniRoute is the only one with the full set; competitors show a mix of checks, partials and crosses. Verified from each project&apos;s docs."/>
 
 <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>
 
@@ -723,7 +723,7 @@ Expose OmniRoute over **MCP**, **A2A**, a **REST API**, **webhooks** or a **remo
 <table>
   <tr><th align="left">Interface</th><th align="left">Endpoint / command</th><th align="left">Use it for</th></tr>
   <tr><td align="left" nowrap>🧰 <b>MCP (stdio)</b></td><td align="left" nowrap><code>omniroute --mcp</code></td><td align="left">Plug into Claude Desktop, Cursor, any MCP client</td></tr>
-  <tr><td align="left" nowrap>🌊 <b>MCP (HTTP)</b></td><td align="left" nowrap><code>/api/mcp/stream</code></td><td align="left">Remote MCP — <b>104 tools</b>, 31 scopes, full audit trail</td></tr>
+  <tr><td align="left" nowrap>🌊 <b>MCP (HTTP)</b></td><td align="left" nowrap><code>/api/mcp/stream</code></td><td align="left">Remote MCP — <b>105 tools</b>, 31 scopes, full audit trail</td></tr>
   <tr><td align="left" nowrap>📡 <b>MCP (SSE)</b></td><td align="left" nowrap><code>/api/mcp/sse</code></td><td align="left">Streaming MCP transport</td></tr>
   <tr><td align="left" nowrap>🤝 <b>A2A</b></td><td align="left" nowrap><code>/.well-known/agent.json</code></td><td align="left">Agent-to-agent, <b>JSON-RPC 2.0</b> + SSE, 6 skills</td></tr>
   <tr><td align="left" nowrap>🌐 <b>REST API</b></td><td align="left" nowrap><code>/v1/*</code></td><td align="left">OpenAI-compatible — chat, embeddings, images, audio, OCR</td></tr>

changelog.d/features/9415-newapi-sub2api-aggregator-balance.md

@@ -1,6 +1 @@
----
-kind: feature
-ref: "#9415"
----
-
-New-API / One-API / Sub2API aggregator balance detection for compatible nodes. When a compatible provider node has the "Aggregator Gateway" toggle enabled, OmniRoute will query the aggregator's `/api/user/self` endpoint to detect the account balance. The dashboard shows the balance badge and quota-preflight routing skips exhausted accounts. The feature is gated by the `NEWAPI_AGGREGATOR_BALANCE` feature flag (default: off). A custom `quotaPerUnit` override is supported for aggregators that use a different rate than the default 500000 units/$1.
+- **sse:** New-API / One-API / Sub2API aggregator balance detection for compatible nodes — with the "Aggregator Gateway" toggle on, OmniRoute queries the aggregator's `/api/user/self` to read the account balance, shows it as a dashboard badge and lets quota-preflight routing skip exhausted accounts. Gated by the `NEWAPI_AGGREGATOR_BALANCE` feature flag (default off), with a `quotaPerUnit` override for aggregators that do not use the default 500000 units/$1 rate ([#9415](https://github.com/diegosouzapw/OmniRoute/issues/9415))

config/quality/eslint-suppressions.json

@@ -1,5 +1,4 @@
 {
-  "_comment": "Congelamento da divida ESLint da migracao TypeScript 7 (release/v3.8.50, 2026-08-05; regenerado 2026-08-06 apos prune de entradas orfas). Gerado pelo modo nativo `eslint --suppress-all --suppressions-location config/quality/eslint-suppressions.json` (NODE_OPTIONS=--max-old-space-size=12288). Politica: violacao PRE-EXISTENTE fica suprimida aqui; violacao NOVA (fora deste arquivo) e vermelho imediato e deve ser corrigida, nunca adicionada. Entradas que deixarem de ocorrer sao podadas com `eslint --prune-suppressions` (o job 'No new ESLint warnings' falha com supressoes orfas). A baseline eslintWarnings em config/quality/quality-baseline.json e 0 — o valor real medido com estas supressoes aplicadas.",
   "open-sse/executors/blackbox-web.ts": {
     "@typescript-eslint/no-explicit-any": {
       "count": 1
@@ -2080,11 +2079,6 @@
       "count": 5
     }
   },
-  "tests/unit/combo-cache-invalidation.test.ts": {
-    "@typescript-eslint/no-explicit-any": {
-      "count": 5
-    }
-  },
   "tests/unit/combo-context-length.test.ts": {
     "@typescript-eslint/no-explicit-any": {
       "count": 2
@@ -3328,11 +3322,6 @@
       "count": 2
     }
   },
-  "tests/unit/vertex-functioncall-id-3440.test.ts": {
-    "@typescript-eslint/no-explicit-any": {
-      "count": 2
-    }
-  },
   "tests/unit/vertex-media.test.ts": {
     "@typescript-eslint/no-explicit-any": {
       "count": 3
@@ -3363,4 +3352,4 @@
       "count": 5
     }
   }
-}
+}

config/quality/file-size-baseline.json

@@ -371,7 +371,7 @@
     "tests/unit/model-sync-route.test.ts": 1016,
     "tests/unit/models-catalog-route.test.ts": 1636,
     "tests/unit/perplexity-web.test.ts": 1355,
-    "tests/unit/provider-models-route.test.ts": 1784,
+    "tests/unit/provider-models-route.test.ts": 1787,
     "tests/unit/provider-validation-specialty.test.ts": 2985,
     "tests/unit/providers-page-utils.test.ts": 1106,
     "tests/unit/response-sanitizer.test.ts": 1063,
@@ -381,7 +381,7 @@
     "tests/unit/stream-utils.test.ts": 2445,
     "tests/unit/token-refresh-service.test.ts": 1378,
     "tests/unit/translator-openai-responses-req.test.ts": 1194,
-    "tests/unit/translator-openai-to-gemini.test.ts": 1616,
+    "tests/unit/translator-openai-to-gemini.test.ts": 1619,
     "tests/unit/translator-openai-to-kiro.test.ts": 1275,
     "tests/unit/translator-resp-gemini-to-openai.test.ts": 1234,
     "tests/unit/usage-service-hardening.test.ts": 1483,
@@ -607,5 +607,7 @@
   "_rebaseline_2026_08_05_9323_agentrouter_waf_retry": "PR #9323 (fix(agentrouter): retry on 400 content-blocked + burst guard) own growth: open-sse/executors/base.ts 1578->1623 (check-file-size.mjs conta via split(\"\\n\").length; wc -l ve 1622). As +45 linhas sao o WAF_RETRY_CONFIG + o burst guard via gateOutboundRequest() para o WAF do agentrouter.org, com comentarios explicando o porque de cada mitigacao e cobertos por tests/unit/base-executor-waf-retry.test.ts e tests/unit/wafRateLimit.test.ts. Crescimento funcional legitimo, nao inchaco.",
   "_rebaseline_2026_08_05_9529_own_growth": "PR #9529 own growth (base release/v3.8.50 medida EXATAMENTE nos frozen antigos, entao o modo base-relative #8522 nao cobre): open-sse/services/rateLimitManager.ts 1060->1105 (+45: helper applyLimiterSettings() que re-arma o heartbeat do reservoir apos updateSettings — fix do bug Bottleneck 2.19.5 que congelava a fila weighted; TDD em tests/unit/ratelimit-reservoir-refresh.test.ts); tests/integration/chat-pipeline.test.ts 1592->1598 (+6: User-Agent do codex derivado de getCodexClientVersion() em vez de literal pinado — teste-irmao alinhado ao contrato); tests/unit/provider-validation-specialty.test.ts 2980->2985 (+5: cobertura NOVA claude-web 429 -> valid:false, alinhamento #9406); open-sse/translator/response/openai-responses.ts 1174->1204 (+30: buildResponsesReasoningSummaryDelta MOVIDA do leaf pureHelpers.ts para o host — a funcao do #9500 muta stream state e violava o contrato do leaf puro; o LOC total do par host+leaf nao cresceu, o pureHelpers encolheu o mesmo tanto). Crescimento por fix de producao + cobertura adicional + realocacao arquitetural, nao inchaco.",
   "_rebaseline_2026_08_06_v3850_inherited_drift_reconcile": "Reconciliacao 2026-08-06 do drift ACUMULADO da release/v3.8.50 apos o lote de merges de 08-05/06: 13 arquivos acima do frozen no tip puro 8180b49ce1 (medidos pelo proprio gate). O modo PR base-relative (#8522) deixa PRs inocentes passarem, e os rebaselines individuais dos PRs se perderam nas resolucoes sucessivas de conflito deste hot-file — o drift so aparece no modo absoluto (nightly/local). Crescimentos funcionais dos PRs mergeados: #9024 topology click-nav src/app/(dashboard)/dashboard/HomePageClient.tsx; #9324 OpenRouter enrich src/app/(dashboard)/dashboard/providers/page.tsx; #9329 quota card ordering src/app/(dashboard)/dashboard/usage/components/ProviderLimits/index.tsx; #9193 context-window suffixes src/sse/handlers/chat.ts; #9332 nested Claude server tool ids open-sse/executors/base.ts; #9228 strip orphaned tool outputs open-sse/executors/codex.ts; #9236 nvidia tool-name normalize open-sse/executors/default.ts; #9314 nested tool_call validation open-sse/executors/kiro.ts; #9260 caller identity REST hops open-sse/mcp-server/server.ts; #8934 cache breakpoints tests tests/unit/chatcore-translation-paths.test.ts; #9193 suffix tests tests/unit/combo-routing-engine.test.ts; #9196 reasoning-on-tool-finish tests tests/unit/sse-auth.test.ts; #9163 GPT-5.6 Max reasoning tests tests/unit/translator-openai-to-kiro.test.ts. default.ts e kiro.ts entram no frozen (estavam sem entrada, acima do cap 1000). Atualizacao pos-medicao (a base avancou durante o ciclo do PR): src/sse/handlers/chat.ts 1857->1877 (#9184 affinity EOF evict) e open-sse/executors/default.ts 1027->1042 (#9005 Kimi K3 tool-name backfill).",
-  "_rebaseline_2026_08_06b_v3850_sweepreds_drift": "Segunda reconciliacao de 2026-08-06 (/sweep-reds sobre o tip puro 2ddbbc61a6): 3 arquivos voltaram a passar do frozen apos os merges do mesmo dia, com atribuicao 1:1 por commit. (1) src/app/(dashboard)/dashboard/providers/page.tsx 1928->1944 e (2) open-sse/executors/base.ts 1635->1640, ambos do #9515 (feat(radar): flag-gated signed free-model catalog overlay, commit e7f6b1d130) — o overlay do Radar entra por wiring nos chokepoints ja existentes (a resolucao/verificacao do catalogo assinado mora fora destes dois arquivos); +16 e +5 linhas liquidas nao sao extraiveis sem inventar um leaf por callsite. (3) open-sse/services/accountFallback.ts 1966->1972 do #8704 (commit c4527f97bd), +6 linhas de dados em CREDITS_EXHAUSTED_SIGNALS ('has been exhausted', fixes #8631). src/sse/handlers/chat.ts 1880>1877 tambem estava violando e NAO entra aqui de proposito: e drenado por encolhimento na PR #9598, sem rebaseline. Crescimento proprio DESTA PR: src/lib/db/migrationRunner.ts 1077->1084 (+7) — o guard retroativo em isSchemaAlreadyApplied para os arquivos renumerados 137/138, exigido pela propria mensagem de erro de colisao do runner (ambas as migracoes sao ALTER TABLE ADD COLUMN puro, nao idempotente). Dois `case` + dois `return hasColumn(...)` + 3 linhas de comentario dentro do switch existente; nao extraivel."
+  "_rebaseline_2026_08_06b_v3850_sweepreds_drift": "Segunda reconciliacao de 2026-08-06 (/sweep-reds sobre o tip puro 2ddbbc61a6): 3 arquivos voltaram a passar do frozen apos os merges do mesmo dia, com atribuicao 1:1 por commit. (1) src/app/(dashboard)/dashboard/providers/page.tsx 1928->1944 e (2) open-sse/executors/base.ts 1635->1640, ambos do #9515 (feat(radar): flag-gated signed free-model catalog overlay, commit e7f6b1d130) — o overlay do Radar entra por wiring nos chokepoints ja existentes (a resolucao/verificacao do catalogo assinado mora fora destes dois arquivos); +16 e +5 linhas liquidas nao sao extraiveis sem inventar um leaf por callsite. (3) open-sse/services/accountFallback.ts 1966->1972 do #8704 (commit c4527f97bd), +6 linhas de dados em CREDITS_EXHAUSTED_SIGNALS ('has been exhausted', fixes #8631). src/sse/handlers/chat.ts 1880>1877 tambem estava violando e NAO entra aqui de proposito: e drenado por encolhimento na PR #9598, sem rebaseline. Crescimento proprio DESTA PR: src/lib/db/migrationRunner.ts 1077->1084 (+7) — o guard retroativo em isSchemaAlreadyApplied para os arquivos renumerados 137/138, exigido pela propria mensagem de erro de colisao do runner (ambas as migracoes sao ALTER TABLE ADD COLUMN puro, nao idempotente). Dois `case` + dois `return hasColumn(...)` + 3 linhas de comentario dentro do switch existente; nao extraivel.",
+  "_rebaseline_2026_08_06c_v3850_sweepreds_pr2": "Segunda PR do /sweep-reds (fix/release-v3.8.50-basereds-0806b): tests/unit/provider-models-route.test.ts 1784->1787 (medido pelo gate, que conta split(\"\\n\").length) (+2 apos compressao de comentarios) — alinhamento de contrato forcado por dois merges do dia: #9106 tornou gemini-3.1-pro-high user-callable (a entry do alias entra na lista esperada do teste de discovery-retry, +1 linha de dado + 1 de comentario) e ff012ff420 adicionou onboardUser como bootstrap hop (exclusao no mock, ja comprimida a 1 linha). Nao ha o que encolher sem apagar o comentario que explica o porque.",
+  "_rebaseline_2026_08_07_v3850_sweepreds_pr2_toolnamemap": "tests/unit/translator-openai-to-gemini.test.ts 1616->1619 (+3). O frozen estava EXATAMENTE no tamanho da base, entao qualquer linha nova viola. #9568 (c9a3361e5a) fez buildChangedToolNameMap emitir entradas IDENTIDADE (o Gemini minusculiza nomes de tool nas respostas, entao o tradutor de resposta precisa da chave para mapear de volta), o que passou a incluir `_toolNameMap` no envelope Antigravity de qualquer request com tools. As 3 linhas sao: a chave nova na lista esperada de Object.keys, 1 comentario explicando POR QUE ela aparece (sem ele o proximo leitor tenta remove-la de novo) e 1 assert do CONTEUDO do map — presenca de chave sozinha nao provaria a entrada identidade, que e justamente o comportamento novo. Nao ha o que extrair: e alinhamento de contrato dentro de um teste existente."
 }

docs/frameworks/MCP-SERVER.md

@@ -6,7 +6,7 @@ lastUpdated: 2026-06-28
 
 # OmniRoute MCP Server Documentation
 
-> Model Context Protocol server with 104 tools across routing, cache, compression, memory, skills, proxy, pool, and context source operations.
+> Model Context Protocol server with 105 tools across routing, cache, compression, memory, skills, proxy, pool, and context source operations.
 >
 > Source of truth: `open-sse/mcp-server/server.ts` computes **104 unique tools** with `countUniqueMcpTools()`: 42 canonical definitions (including the six CCR lifecycle tools and the agent-skills trio), plus memory (3), skills (4), GitHub skills (3), pool (6), gamification (8), plugins (8), Notion (6), Obsidian (22), and two RTK-only compression tools.
 
@@ -369,7 +369,7 @@ MCP tool, prompt, and resource registries can compress descriptions at registrat
 
 Description compression shrinks each tool's metadata; **tool-cardinality reduction** goes one step further by reducing _how many_ tools are announced at all. Advertising fewer tools in the `tools/list` manifest cuts the per-request token cost the client's model pays for the tool catalog ("layer 5" compression). The implementation is a pure, stateless filter in `open-sse/mcp-server/toolCardinality.ts` (`reduceToolManifest`), wired into the registration loop in `createMcpServer()` (`open-sse/mcp-server/server.ts`).
 
-**Opt-in, off by default.** The filter only runs when at least one of two environment variables is set; with neither set, all 104 tools are announced unchanged.
+**Opt-in, off by default.** The filter only runs when at least one of two environment variables is set; with neither set, all 105 tools are announced unchanged.
 
 | Variable         | Mode                                                                                    |
 | :--------------- | :-------------------------------------------------------------------------------------- |

docs/i18n/pl/docs/ops/PROXY_GUIDE.md

@@ -645,7 +645,7 @@ for (const s of statuses) {
 }
 
 // Force re-check a specific proxy
-invalidateProxyHealth("http://user:pass@1.2.3.4:8080");
+invalidateProxyHealth("http://user:pass@203.0.113.7:8080");
 ```
 
 Flaga `stale` jest `true`, gdy wpis cache przekroczył `HEALTH_CACHE_TTL_MS` i następne żądanie wywoła świeży check.
@@ -807,7 +807,7 @@ Gdy proxy systematycznie pada, oznacz je ręcznie, by rotator je pomijał:
 ```ts
 import { failOneproxyProxy } from "omniroute/oneproxyRotator";
 
-const removed = await failOneproxyProxy("1.2.3.4", 8080);
+const removed = await failOneproxyProxy("203.0.113.7", 8080);
 if (removed) {
   console.log("Proxy marked as failed; rotator will skip it");
 }

docs/ops/PROXY_GUIDE.md

@@ -645,7 +645,7 @@ for (const s of statuses) {
 }
 
 // Force re-check a specific proxy
-invalidateProxyHealth("http://user:pass@1.2.3.4:8080");
+invalidateProxyHealth("http://user:pass@203.0.113.7:8080");
 ```
 
 The `stale` flag is `true` when the cache entry has exceeded `HEALTH_CACHE_TTL_MS` and the next request will trigger a fresh check.
@@ -807,7 +807,7 @@ When a proxy consistently fails, mark it manually so the rotator will skip it:
 ```ts
 import { failOneproxyProxy } from "omniroute/oneproxyRotator";
 
-const removed = await failOneproxyProxy("1.2.3.4", 8080);
+const removed = await failOneproxyProxy("203.0.113.7", 8080);
 if (removed) {
   console.log("Proxy marked as failed; rotator will skip it");
 }

docs/ops/VM_DEPLOYMENT_GUIDE.md

@@ -429,6 +429,7 @@ For deployments on small VPS instances (1 GB RAM or less):
 
 - **Disable background services** — set `OMNIROUTE_DISABLE_BACKGROUND_SERVICES=1` to skip scheduler, MCP server, and periodic maintenance tasks. See `docs/reference/ENVIRONMENT.md`.
 - **Use SQLite WAL mode** — enabled by default, reduces peak memory during concurrent reads.
-- **Limit connection concurrency** — reduce `OMNIROUTE_MAX_POOL_SIZE` and `OMNIROUTE_DB_POOL_SIZE` in your environment.
+- **Cap the V8 heap** — set `OMNIROUTE_MEMORY_MB` (e.g. `512`) so the runtime does not calibrate a ceiling larger than the VM. See `docs/reference/ENVIRONMENT.md`.
+- **Limit concurrent heavy requests** — lower `OMNIROUTE_CHAT_MAX_HEAVY_IN_FLIGHT` (default `1`); excess requests get a retryable `503` with `Retry-After` instead of competing for memory.
 - **Avoid `next build` on the VPS** — build locally and deploy the standalone output (`.next/standalone/`).
 - **Monitor with `top` / `free -m`** — OmniRoute typically uses 200-400 MB RSS at idle on a 1 GB VM.

open-sse/mcp-server/__tests__/essentialTools.test.ts

@@ -22,9 +22,10 @@ describe("MCP Essential Tools", () => {
   });
 
   describe("Tool schema validation", () => {
-    it("should have exactly 11 essential tools (includes web_search + web_fetch + tool_search)", () => {
+    it("should have exactly 12 essential tools (includes web_search + web_fetch + tool_search)", () => {
+      // 11 -> 12: #8925 shipped omniroute_create_combo as a phase-1 tool.
       const schemas = MCP_ESSENTIAL_TOOLS;
-      expect(schemas).toHaveLength(11);
+      expect(schemas).toHaveLength(12);
     });
 
     it("all tools should have omniroute_ prefix", () => {

open-sse/services/accountSemaphore.ts

@@ -54,8 +54,21 @@ export function buildAccountSemaphoreKey({
   return `${String(provider)}:${String(accountKey)}`;
 }
 
+/**
+ * Effective positive cap, or null when the semaphore is bypassed (unset/<=0).
+ *
+ * Narrowing companion of {@link isBypassed}: that one returns a plain boolean, so
+ * TypeScript cannot narrow `number | null` to `number` in its else-branch (a
+ * `x is null | undefined` predicate would be unsound — 0 bypasses too). Callers
+ * that need the VALUE after the guard go through here instead of casting.
+ */
+function resolveActiveCap(maxConcurrency?: number | null): number | null {
+  if (maxConcurrency == null || maxConcurrency <= 0) return null;
+  return maxConcurrency;
+}
+
 function isBypassed(maxConcurrency?: number | null): boolean {
-  return maxConcurrency == null || maxConcurrency <= 0;
+  return resolveActiveCap(maxConcurrency) === null;
 }
 
 function createNoopReleaseFn(): () => void {
@@ -192,7 +205,8 @@ export function acquire(
     maxQueueSize = DEFAULT_MAX_QUEUE_SIZE,
   }: AcquireAccountSemaphoreOptions = {}
 ): Promise<() => void> {
-  if (isBypassed(maxConcurrency)) {
+  const activeCap = resolveActiveCap(maxConcurrency);
+  if (activeCap === null) {
     return Promise.resolve(createNoopReleaseFn());
   }
 
@@ -200,7 +214,7 @@ export function acquire(
     return Promise.reject(makeAbortError(signal));
   }
 
-  const gate = ensureGate(semaphoreKey, maxConcurrency);
+  const gate = ensureGate(semaphoreKey, activeCap);
   clearCleanupTimer(gate);
 
   if (gate.running < gate.maxConcurrency && !isBlocked(gate)) {

open-sse/services/antigravityProjectPersist.ts

@@ -22,6 +22,43 @@ import { updateProviderConnection } from "@/lib/db/providers";
  * Best-effort / non-fatal by design: a persistence failure must never block
  * the in-flight request, which already has the discovered id in hand.
  */
+/**
+ * Selection-side companion of the persistence write path (#8894): given a pool
+ * of Antigravity/AGY connections, prefer the ones that already carry a stored
+ * projectId — they can serve a request without the `loadCodeAssist` discovery
+ * round-trip. "Prefer", not "require": when NO connection has a stored project
+ * the pool is returned unchanged, so a fresh install never empties its
+ * candidate list.
+ *
+ * Sync on purpose (called inside the quota-strategy connection expansion, which
+ * builds candidate lists without awaiting per-connection work). Tolerates
+ * `providerSpecificData` arriving either parsed or as the raw DB JSON string.
+ */
+export function preferAntigravityConnectionsWithStoredProject<T extends Record<string, unknown>>(
+  connections: T[]
+): T[] {
+  if (!Array.isArray(connections) || connections.length === 0) return connections;
+  const hasStoredProject = (connection: T): boolean => {
+    if (typeof connection.projectId === "string" && connection.projectId) return true;
+    let psd = connection.providerSpecificData;
+    if (typeof psd === "string") {
+      try {
+        psd = JSON.parse(psd);
+      } catch {
+        return false;
+      }
+    }
+    return Boolean(
+      psd &&
+      typeof psd === "object" &&
+      typeof (psd as Record<string, unknown>).projectId === "string" &&
+      (psd as Record<string, unknown>).projectId
+    );
+  };
+  const withStoredProject = connections.filter(hasStoredProject);
+  return withStoredProject.length > 0 ? withStoredProject : connections;
+}
+
 export async function persistDiscoveredAntigravityProjectId(
   connectionId: string | undefined | null,
   discoveredProjectId: string | undefined | null,

open-sse/services/combo/comboStructure.ts

@@ -137,7 +137,9 @@ function normalizeRuntimeStep(
       : {}),
     weight,
     label,
-    prompt: step.prompt || null,
+    // `prompt` is a per-step pipeline input and only exists on a model step —
+    // #8894 widened the union with ComboProviderWildcardStep, which has no prompt.
+    prompt: (step.kind === "model" ? step.prompt : null) || null,
   } satisfies ResolvedComboTarget;
 }
 

open-sse/services/combo/dispatchPrelude.ts

@@ -619,6 +619,7 @@ export async function tryRuntimeUnitDispatch(args: {
     nesting: nestingContext,
     baseOptions: buildBaseOptions(args),
     runCombo: args.runCombo,
+    hiddenModelsByProvider: args.hiddenModelsByProvider,
   });
   recordRuntimeUnitStickySuccess({
     strategy,

open-sse/services/combo/fusionPanel.ts

@@ -10,7 +10,7 @@
  * literal `auto/*` string panel member already behaves via the single-
  * dispatch safety net in src/sse/handlers/chat.ts.
  */
-import { normalizeComboStep } from "../../../src/lib/combos/steps.ts";
+import { getComboModelString, normalizeComboStep } from "../../../src/lib/combos/steps.ts";
 import { executeComboRefUnit } from "./runtimeUnits.ts";
 import type {
   ComboCollectionLike,
@@ -51,7 +51,11 @@ export function extractFusionPanelSpec(
       panel.push(step.comboName);
       return;
     }
-    panel.push(step.model);
+    // #8894 widened ComboStep with ComboProviderWildcardStep, which carries a
+    // modelPattern instead of a model. getComboModelString() already resolves any
+    // step shape (and returns null for the ones with no concrete model id).
+    const modelStr = getComboModelString(step);
+    if (modelStr) panel.push(modelStr);
   });
   return { panel, comboRefUnits };
 }

open-sse/services/combo/quotaStrategies.ts

@@ -45,7 +45,7 @@ import {
   type QuotaFetchCacheConfig,
 } from "./quotaScoring.ts";
 import { rankByHeadroom, type HeadroomSaturation } from "./headroomRanking.ts";
-import { preferAntigravityConnectionsWithStoredProject } from "../antigravityProjectPersistence.ts";
+import { preferAntigravityConnectionsWithStoredProject } from "../antigravityProjectPersist.ts";
 import { isQuotaExhaustedForRequest } from "../../../src/domain/quotaCache.ts";
 
 const RESET_AWARE_CONNECTION_CACHE_TTL_MS = 30_000;

open-sse/services/combo/runtimeUnitCapacity.ts

@@ -9,7 +9,12 @@
 import { isAccountSemaphoreFull } from "../accountSemaphore.ts";
 import { resolveComboTargets } from "./comboStructure.ts";
 import { lookupPositiveCap } from "./concurrencyCaps.ts";
-import type { ComboCollectionLike, ComboLike, ResolvedComboUnit } from "./types.ts";
+import type {
+  ComboCollectionLike,
+  ComboLike,
+  HiddenModelsByProvider,
+  ResolvedComboUnit,
+} from "./types.ts";
 
 type CapLookup = (connectionId: string) => Promise<number | null>;
 
@@ -45,7 +50,12 @@ async function isConnectionAtConcurrencyCap(
 export async function isRuntimeUnitAtConcurrencyCap(
   unit: ResolvedComboUnit,
   allCombos: ComboCollectionLike,
-  lookupCap: CapLookup = lookupPositiveCap
+  lookupCap: CapLookup = lookupPositiveCap,
+  // Threaded from the caller so the hidden-model snapshot resolved once per
+  // request is reused. Without it resolveComboTargets falls back to its default
+  // getHiddenModelsByProvider(), i.e. a fresh full key_value read per nested
+  // combo-ref unit on EVERY request (#8878 threaded the other call sites).
+  hiddenModelsByProvider?: HiddenModelsByProvider
 ): Promise<boolean> {
   if (unit.kind === "model") {
     if (!unit.connectionId || !unit.provider) return false;
@@ -55,7 +65,7 @@ export async function isRuntimeUnitAtConcurrencyCap(
   const childCombo = findComboByName(allCombos, unit.comboName);
   if (!childCombo) return false;
 
-  const targets = resolveComboTargets(childCombo, allCombos, 1);
+  const targets = resolveComboTargets(childCombo, allCombos, 1, hiddenModelsByProvider);
   const byConnection = new Map<string, { provider: string; connectionId: string }>();
   for (const target of targets) {
     if (!target.connectionId || !target.provider) continue;

open-sse/services/combo/runtimeUnits.ts

@@ -18,6 +18,7 @@ import type {
   ComboNestingContext,
   HandleComboChatOptions,
   HandleSingleModel,
+  HiddenModelsByProvider,
   IsModelAvailable,
   ResolvedComboRefTarget,
   ResolvedComboUnit,
@@ -186,6 +187,7 @@ export async function executeRuntimeUnitCombo(args: {
   nesting: ComboNestingContext;
   baseOptions: HandleComboChatOptions;
   runCombo: RuntimeUnitRunner;
+  hiddenModelsByProvider?: HiddenModelsByProvider;
 }): Promise<RuntimeUnitExecutionResult> {
   const maxRetries = Number(args.config.maxRetries ?? 1);
   const retryDelayMs = resolveDelayMs(args.config.retryDelayMs, 2000);
@@ -197,7 +199,14 @@ export async function executeRuntimeUnitCombo(args: {
   let fallbackCount = 0;
 
   for (const unit of orderedUnits) {
-    if (await isRuntimeUnitAtConcurrencyCap(unit, args.allCombos)) {
+    if (
+      await isRuntimeUnitAtConcurrencyCap(
+        unit,
+        args.allCombos,
+        undefined,
+        args.hiddenModelsByProvider
+      )
+    ) {
       args.log.info(
         "COMBO",
         `Skipping ${unit.kind} ${unitDisplayName(unit)} — concurrency cap reached`

open-sse/services/compression/engines/ccr/index.ts

@@ -292,7 +292,8 @@ function rehydrateEntry(hash: string, principalId: string, now: number): CcrEntr
 
   // Re-admit through the same budgets a fresh store would face. If the block no longer
   // fits, it stays on disk and is served straight from the row instead of being cached.
-  if (enforcePrincipalBudget(entry.principalId, entry.bytes) && enforceGlobalBudget(entry.bytes)) {
+  const { principalId: owner, bytes } = entry;
+  if (enforcePrincipalBudget(owner, bytes) && enforceGlobalBudget(owner, bytes)) {
     const key = buildStoreKey(hash, principalId === ANON ? undefined : principalId);
     ccrStore.set(key, entry);
     ccrTotalBytes += entry.bytes;

open-sse/services/firecrawlQuotaFetcher.ts

@@ -110,7 +110,8 @@ export function getFirecrawlBaseUrl(connection?: Record<string, unknown>): strin
     return envBase.replace(/\/+$/, "");
   }
   const providerData = toRecord(connection?.providerSpecificData);
-  const connBase = typeof connection?.baseUrl === "string" ? connection.baseUrl : providerData?.baseUrl;
+  const connBase =
+    typeof connection?.baseUrl === "string" ? connection.baseUrl : providerData?.baseUrl;
   if (typeof connBase === "string" && connBase.trim() && !connBase.includes("api.firecrawl.dev")) {
     return connBase.trim().replace(/\/+$/, "");
   }
@@ -120,7 +121,11 @@ export function getFirecrawlBaseUrl(connection?: Record<string, unknown>): strin
 export async function fetchFirecrawlQuota(
   connectionId: string,
   connection?: Record<string, unknown>
-): Promise<QuotaInfo | null> {
+  // FirecrawlQuota, not the base QuotaInfo: every return here is a full credit
+  // breakdown (remainingCredits / planCredits / extraCreditsInferred / overPlan),
+  // and the narrower annotation made the custom-base literal below an excess-
+  // property error. FirecrawlQuota extends QuotaInfo, so callers are unaffected.
+): Promise<FirecrawlQuota | null> {
   const cached = quotaCache.get(connectionId);
   if (cached && Date.now() - cached.fetchedAt < CACHE_TTL_MS) {
     return cached.quota;

open-sse/services/raycast.ts

@@ -9,15 +9,23 @@
 
 import { createHmac, createHash, randomUUID } from "node:crypto";
 
+import { resolvePublicCred } from "../utils/publicCreds.ts";
+
 export const RAYCAST_CHAT_URL = "https://backend.raycast.com/api/v1/ai/chat_completions";
 export const RAYCAST_MODELS_URL = "https://backend.raycast.com/api/v1/ai/models";
-export const RAYCAST_DEFAULT_USER_AGENT =
-  "Raycast/1.104.20 (macOS Version 26.5.1 (Build 25F80))";
+export const RAYCAST_DEFAULT_USER_AGENT = "Raycast/1.104.20 (macOS Version 26.5.1 (Build 25F80))";
 export const RAYCAST_DEFAULT_EXPERIMENTAL = "chatBranching, mcpHTTPServer";
 
-/** Community-extracted default; override via providerSpecificData.sigSecret or RAYCAST_SIG_SECRET. */
-export const RAYCAST_DEFAULT_SIG_SECRET =
-  "6bc455473576ce2cd6f70426caff867aabbe3f7291c1a79681af5e8ce0ca1408";
+/**
+ * Community-extracted default; override via providerSpecificData.sigSecret or
+ * RAYCAST_SIG_SECRET. Embedded through resolvePublicCred() per Hard Rule #11 —
+ * a public upstream credential must never be a string literal in the source
+ * (see docs/security/PUBLIC_CREDS.md).
+ */
+export const RAYCAST_DEFAULT_SIG_SECRET = resolvePublicCred(
+  "raycast_sig_secret",
+  "RAYCAST_SIG_SECRET"
+);
 
 export type RaycastCredentials = {
   accessToken?: string;
@@ -69,9 +77,7 @@ export function raycastJwt(aid: string, secret: string): string {
   const iat = Date.now() / 1000;
   const header = base64UrlJson({ typ: "JWT", alg: "HS256" });
   const payload = base64UrlJson({ aid, exp: iat + 60, iat });
-  const signature = createHmac("sha256", secret)
-    .update(`${header}.${payload}`)
-    .digest("base64url");
+  const signature = createHmac("sha256", secret).update(`${header}.${payload}`).digest("base64url");
   return `${header}.${payload}.${signature}`;
 }
 
@@ -111,7 +117,10 @@ export function resolveRaycastSecrets(credentials: RaycastCredentials): {
   return { bearerToken, deviceId, aid, sigSecret };
 }
 
-export function buildRaycastHeaders(payload: string, credentials: RaycastCredentials): Record<string, string> {
+export function buildRaycastHeaders(
+  payload: string,
+  credentials: RaycastCredentials
+): Record<string, string> {
   const { bearerToken, deviceId, aid, sigSecret } = resolveRaycastSecrets(credentials);
   const psd = credentials.providerSpecificData || {};
   const timestamp = Math.floor(Date.now() / 1000).toString();
@@ -137,7 +146,12 @@ export function contentToText(content: unknown): string {
   return content
     .map((part) => {
       if (typeof part === "string") return part;
-      if (part && typeof part === "object" && "type" in part && (part as { type?: string }).type === "text") {
+      if (
+        part &&
+        typeof part === "object" &&
+        "type" in part &&
+        (part as { type?: string }).type === "text"
+      ) {
         return String((part as { text?: string }).text || "");
       }
       return "";

open-sse/services/usage/firecrawl.ts

@@ -5,7 +5,11 @@
  * credits into the standard `{ plan, quotas }` response.
  */
 
-import { fetchFirecrawlQuota, getFirecrawlBaseUrl, type FirecrawlQuota } from "../firecrawlQuotaFetcher.ts";
+import {
+  fetchFirecrawlQuota,
+  getFirecrawlBaseUrl,
+  type FirecrawlQuota,
+} from "../firecrawlQuotaFetcher.ts";
 import { createQuotaFromUsage, parseResetTime } from "./quota.ts";
 
 function createFirecrawlPlanQuota(q: FirecrawlQuota) {
@@ -29,7 +33,11 @@ function createFirecrawlPlanQuota(q: FirecrawlQuota) {
   };
 }
 
-export async function getFirecrawlUsage(connectionId: string, apiKey?: string, connection?: Record<string, unknown>) {
+export async function getFirecrawlUsage(
+  connectionId: string,
+  apiKey?: string,
+  connection?: Record<string, unknown>
+) {
   if (!connectionId) {
     return { message: "Firecrawl: connection id unavailable." };
   }
@@ -44,7 +52,12 @@ export async function getFirecrawlUsage(connectionId: string, apiKey?: string, c
   }
 
   try {
-    const live = await fetchFirecrawlQuota(connectionId, connection);
+    // The explicit `apiKey` argument was silently dropped when #91bb6aa619 moved
+    // this to fetchFirecrawlQuota(connectionId, connection): the fetcher reads the
+    // key off the connection record, so a caller that passes the key directly —
+    // without a connection carrying it — always got "API key not available".
+    const resolvedConnection = apiKey ? { ...(connection || {}), apiKey } : connection;
+    const live = await fetchFirecrawlQuota(connectionId, resolvedConnection);
     if (!live) {
       return { message: "Firecrawl API key not available or credit usage unavailable." };
     }

open-sse/utils/publicCreds.ts

@@ -207,6 +207,15 @@ const EMBEDDED_DEFAULTS = {
   // Firefly credits balance endpoint public x-api-key (`SunbreakWebUI1`) from
   // GET firefly.adobe.io/v1/credits/balance browser traffic.
   adobe_firefly_balance_api_key: [60, 24, 0, 11, 0, 10, 20, 31, 50, 72, 18, 32, 43, 93],
+  // Raycast Pro V2 request-signature secret (#8895). Community-extracted from the
+  // public Raycast macOS client — the SAME value ships to every install, so it is
+  // public by design, not a per-user credential. Overridable via RAYCAST_SIG_SECRET
+  // or providerSpecificData.sigSecret.
+  raycast_sig_secret: [
+    89, 15, 13, 93, 71, 90, 65, 67, 86, 24, 71, 67, 1, 9, 91, 0, 73, 64, 87, 88, 93, 90, 91, 68, 12,
+    20, 18, 3, 21, 70, 66, 3, 13, 11, 1, 72, 69, 87, 88, 95, 87, 88, 17, 94, 20, 67, 92, 27, 72, 68,
+    3, 10, 92, 6, 21, 21, 84, 95, 14, 15, 88, 70, 95, 77,
+  ],
 } as const;
 
 export type EmbeddedDefaultKey = keyof typeof EMBEDDED_DEFAULTS;

scripts/check/check-db-rules.mjs

@@ -63,6 +63,7 @@ export const INTENTIONALLY_INTERNAL = new Set([
   "optimizationSettings", // db-internal: imported by db/core.ts for SQLite PRAGMA application helpers that require the live adapter
   "pluginMetrics", // DEAD? (production): write path não foi conectado ainda (documentado no cabeçalho do módulo); testado por tests/unit/plugins-metrics.test.ts
   "prompts", // DEAD? (production): zero callers de produção encontrados; domínio domain/prompts.ts é independente; testado por tests/integration/proxy-pipeline.test.ts
+  "probeUtils", // db-internal: importado so por db/core.ts (retryProbeIfTransient no caminho da corruption-probe, #9541); testado por tests/unit/probe-9541-repro.test.ts
   "providerNodeSelect", // db-internal: importado só por db/providers.ts (selectProviderNodeForConnection — lógica pura de seleção de provider node split do providers.ts, #4421)
   "providerStats", // intentionally-internal: src/app/api/provider-stats/route.ts
   "proxyLatency", // intentionally-internal: imported directly by src/lib/db/proxies.ts (anti-barrel, #6798)

scripts/check/check-test-discovery.mjs

@@ -107,6 +107,11 @@ export const COLLECTORS = [
     glob: "open-sse/services/__tests__/antigravity-quota-family.test.ts",
     sources: ["vitest.mcp.config.ts"],
   },
+  // #8890 landed this suite here without wiring a runner, so it had never run once.
+  {
+    glob: "open-sse/services/__tests__/fail-fast-concurrency-gate.test.ts",
+    sources: ["vitest.mcp.config.ts"],
+  },
   { glob: "tests/unit/autoCombo/**/*.test.ts", sources: ["vitest.mcp.config.ts"] },
   { glob: "src/lib/memory/__tests__/generic-backend.test.ts", sources: ["vitest.mcp.config.ts"] },
   { glob: "tests/unit/encryption.spec.ts", sources: ["vitest.mcp.config.ts"] },

src/app/api/providers/[id]/test/oauthTestConfig.ts

@@ -111,6 +111,14 @@ export const OAUTH_TEST_CONFIG = {
     // Validate using token presence/expiry as a lightweight auth check.
     checkExpiry: true,
   },
+  raycast: {
+    // #8895 — Raycast Pro is an `import_token` provider: the token is imported
+    // from the local Raycast install, `refreshToken` is always null and the
+    // stored `expiresIn` defaults to 30 days. There is nothing to refresh, so
+    // the test is the expiry check on the imported token; without an entry here
+    // Test Connection persists testStatus="error" on a healthy account (#8408).
+    checkExpiry: true,
+  },
   cline: CLINE_OAUTH_TEST_CONFIG,
   // ClinePass reuses the same WorkOS OAuth flow and token lifecycle as Cline.
   clinepass: CLINE_OAUTH_TEST_CONFIG,

src/i18n/messages/pt-BR.json

@@ -5475,13 +5475,13 @@
     "newApiUserIdLabel": "ID de Usuário New-API",
     "newApiUserIdPlaceholder": "ex.: 12345",
     "newApiUserIdHint": "Valor do cabeçalho New-Api-User do AgentRouter, usado junto com a chave de API do console para consultar o saldo de cota.",
-    "newApiAggregatorToggleLabel": "Gateway Agregador",
-    "newApiAggregatorToggleHint": "Ativar detecção de saldo para nós agregadores New-API / One-API / Sub2API. O painel mostrará o badge de saldo e o roteamento de pré-voo de cota ignorará contas esgotadas.",
-    "newApiAggregatorConsoleApiKeyHint": "Token de Acesso do Sistema para o endpoint /api/user/self do agregador. Não é a chave de API de roteamento.",
-    "newApiAggregatorUserIdHint": "Valor do cabeçalho New-Api-User usado para consultar o saldo de cota do usuário do agregador.",
-    "newApiAggregatorQuotaPerUnitLabel": "Cota por Unidade",
-    "newApiAggregatorQuotaPerUnitHint": "Unidades de crédito New-API por $1 (padrão: 500000). Substitua se seu agregador usar uma taxa diferente.",
-    "featureFlagNewApiAggregatorBalanceDescription": "Ativar detecção de saldo para nós compatíveis de agregadores New-API / One-API / Sub2API",
+    "newApiAggregatorToggleLabel": "Gateway agregador",
+    "newApiAggregatorToggleHint": "Ativa a detecção de saldo para nós agregadores New-API / One-API / Sub2API. O painel passa a mostrar o selo de saldo e o roteamento com quota-preflight pula contas esgotadas.",
+    "newApiAggregatorConsoleApiKeyHint": "System Access Token para o endpoint /api/user/self do agregador. Não é a chave de API de roteamento.",
+    "newApiAggregatorUserIdHint": "Valor do cabeçalho New-Api-User usado para buscar o saldo de quota do usuário do agregador.",
+    "newApiAggregatorQuotaPerUnitLabel": "Quota por unidade",
+    "newApiAggregatorQuotaPerUnitHint": "Unidades de crédito New-API por US$ 1 (padrão: 500000). Substitua se o seu agregador usar outra taxa.",
+    "featureFlagNewApiAggregatorBalanceDescription": "Ativa a detecção de saldo para nós compatíveis New-API / One-API / Sub2API",
     "cpaModeDisabledTitle": "Habilitar backend CLIProxyAPI para emulação OAuth mais profunda do Claude Code",
     "cpaModeEnabledTitle": "Usando CLIProxyAPI para uma emulação mais profunda do Claude Code (uTLS, multi-conta, perfis de dispositivo)",
     "customUserAgentHint": "Override opcional enviado upstream como cabeçalho User-Agent desta conexão.",
@@ -5597,7 +5597,7 @@
     "tagGroupPlaceholder": "ex.: personal, work, team-a",
     "testModel": "Test Model",
     "testingModel": "Testing Model",
-    "modelTestQuotaTooltip": "Cota esgotada — reinicia amanhã ou precisa de recarga",
+    "modelTestQuotaTooltip": "Quota esgotada — reseta amanhã ou precisa de recarga",
     "toggleOffShort": "OFF",
     "toggleOnShort": "ON",
     "tokenExpiredBadge": "Expirado",

src/i18n/messages/vi.json

@@ -6020,7 +6020,15 @@
     "ccAliasAddModelPlaceholder": "Id mô hình (ví dụ: gpt-4o)",
     "ccAliasAddModelButton": "Thêm ghi đè",
     "ccAliasLoadError": "Không tải được cài đặt bí danh khám phá: {error}",
-    "ccAliasSaveError": "Không lưu được cài đặt bí danh khám phá: {error}"
+    "ccAliasSaveError": "Không lưu được cài đặt bí danh khám phá: {error}",
+    "newApiAggregatorToggleLabel": "Cổng tổng hợp",
+    "newApiAggregatorToggleHint": "Bật phát hiện số dư cho các node tổng hợp New-API / One-API / Sub2API. Bảng điều khiển sẽ hiển thị huy hiệu số dư và định tuyến quota-preflight sẽ bỏ qua các tài khoản đã cạn.",
+    "newApiAggregatorConsoleApiKeyHint": "System Access Token cho endpoint /api/user/self của bộ tổng hợp. Không phải khóa API định tuyến.",
+    "newApiAggregatorUserIdHint": "Giá trị header New-Api-User dùng để lấy số dư quota của người dùng bộ tổng hợp.",
+    "newApiAggregatorQuotaPerUnitLabel": "Quota mỗi đơn vị",
+    "newApiAggregatorQuotaPerUnitHint": "Số đơn vị tín dụng New-API cho mỗi 1 USD (mặc định: 500000). Ghi đè nếu bộ tổng hợp của bạn dùng tỷ lệ khác.",
+    "featureFlagNewApiAggregatorBalanceDescription": "Bật phát hiện số dư cho các node tương thích New-API / One-API / Sub2API",
+    "modelTestQuotaTooltip": "Đã hết quota — sẽ đặt lại vào ngày mai hoặc cần nạp thêm"
   },
   "settings": {
     "title": "Cài đặt",

src/lib/db/contextHandoffs.ts

@@ -254,40 +254,3 @@ export function deleteSessionModelHistory(sessionId: string, comboName: string):
     .run(sessionId, comboName);
   return result.changes ?? 0;
 }
-
-/**
- * Get usage counts for ALL models in a session's history.
- * Returns a Map<{model}> -> {count} for least-used strategy.
- *
- * Queries the session_model_history table and aggregates by model_str.
- * Can optionally filter by connectionId if provided.
- *
- * @param connectionId - Optional connection ID to filter by. If not provided, returns all connections.
- * @returns Promise<Map<string, number>> of model strings to their usage count.
- */
-export async function getSessionModelUsageCounts(
-  connectionId?: string
-): Promise<Map<string, number>> {
-  const db = getDbInstance() as any;
-
-  let sql = `SELECT model_str, COUNT(*) as count 
-             FROM session_model_history 
-             WHERE 1=1`;
-  const params: unknown[] = [];
-
-  if (connectionId) {
-    sql += ` AND connection_id = ?`;
-    params.push(connectionId);
-  }
-
-  sql += ` GROUP BY model_str ORDER BY count ASC`;
-
-  const rows = db.prepare(sql).all(...params) as Array<{ model_str: string; count: number }>;
-
-  const usageMap = new Map<string, number>();
-  rows.forEach((row: { model_str: string; count: number }) => {
-    usageMap.set(row.model_str, row.count);
-  });
-
-  return usageMap;
-}

src/lib/db/models/activeSyncedCatalog.ts

@@ -1,6 +1,10 @@
 import { providerUsesAuthoritativeLiveCatalog } from "@omniroute/open-sse/config/providerRegistry";
 import { PROVIDER_ID_TO_ALIAS } from "@omniroute/open-sse/config/providerModels.ts";
-import { getSyncedAvailableModelsByConnection, type SyncedAvailableModel } from "../models";
+import {
+  getSyncedAvailableModels,
+  getSyncedAvailableModelsByConnection,
+  type SyncedAvailableModel,
+} from "../models";
 import { getRawProviderConnections } from "../providers";
 
 export type ActiveSyncedCatalog = {
@@ -102,11 +106,26 @@ export async function getActiveSyncedCatalog(providerId: string): Promise<Active
       .map((connection) => connection.id);
 
     const models = collectModelsForConnections(modelsByConnection, activeConnectionIds);
+    if (models.length > 0) {
+      return {
+        authoritative: providerUsesAuthoritativeLiveCatalog(providerId),
+        models,
+      };
+    }
 
-    return {
-      authoritative: models.length > 0 && providerUsesAuthoritativeLiveCatalog(providerId),
-      models,
-    };
+    // No ACTIVE CONNECTION carries a catalog for this provider — but a provider
+    // NODE can: nodes live in `provider_nodes`, never in `provider_connections`,
+    // so filtering by active connection ids drops their synced catalog entirely.
+    // Before #9294 this path read the provider-wide key_value set, and losing it
+    // took every node's runtime metadata with it (supportedThinkingEfforts, so
+    // `-high`/`-low` effort suffixes stopped resolving, plus contextWindow /
+    // maxInputTokens used by the combo context-window filter).
+    //
+    // Fall back to that provider-wide set, and deliberately keep it
+    // NON-authoritative: #9294's live-catalog gating is about what an active
+    // connection actually serves, so a node-backed catalog must inform metadata
+    // without ever being used to reject a model as unavailable.
+    return { authoritative: false, models: await getSyncedAvailableModels(storedProviderId) };
   } catch {
     return { authoritative: false, models: [] };
   }

src/lib/guardrails/visionBridgeHelpers.ts

@@ -232,13 +232,19 @@ export async function callVisionModel(
   imageDataUri: string,
   config: VisionModelConfig,
   apiKey?: string,
-  routerConfig?: Partial<import("./visionBridgeRouter").VisionBridgeRouterConfig>
+  routerConfig?: Partial<import("./visionBridgeRouter").VisionBridgeRouterConfig>,
+  deps?: import("./visionBridgeRouter").VisionBridgeRouterDeps
 ): Promise<string> {
-  // Auto-select the best vision model
-  const modelToUse = await getBestVisionModel({
-    fixedModel: config.model,
-    ...routerConfig,
-  });
+  // Auto-select the best vision model. `deps` is the router's existing
+  // injectable credential-check seam — without forwarding it, tests (and any
+  // embedder) cannot keep model selection away from the live connections DB.
+  const modelToUse = await getBestVisionModel(
+    {
+      fixedModel: config.model,
+      ...routerConfig,
+    },
+    deps
+  );
   // (#8430) When no vision-capable provider has usable credentials on this
   // instance, surface a clear error instead of attempting a describe call that
   // would fail with an opaque auth/serde error upstream.
@@ -248,7 +254,7 @@ export async function callVisionModel(
   let lastError: Error | null = null;
 
   // Try primary model + fallbacks
-  const modelsToTry = [modelToUse, ...(await getFallbackModels(modelToUse, routerConfig))];
+  const modelsToTry = [modelToUse, ...(await getFallbackModels(modelToUse, routerConfig, deps))];
   const maxAttempts = Math.min(modelsToTry.length, routerConfig?.maxFallbackAttempts ?? 3);
 
   for (let attempt = 0; attempt < maxAttempts; attempt++) {

src/lib/oauth/constants/oauth.ts

@@ -531,6 +531,10 @@ export const PROVIDERS = {
   KIRO: "kiro",
   AMAZON_Q: "amazon-q",
   CURSOR: "cursor",
+  // #8895 — registered in src/lib/oauth/providers/index.ts but missing here, so
+  // every consumer reading PROVIDERS (onboarding wizard, test-connection routing)
+  // did not know Raycast Pro exists as an OAuth provider.
+  RAYCAST: "raycast",
   KILOCODE: "kilocode",
   CLINE: "cline",
   CLINEPASS: "clinepass",

stryker.conf.json

@@ -64,6 +64,7 @@
       "tests/unit/adobe-firefly.test.ts",
       "tests/unit/anthropic-thinking-signature-recovery.test.ts",
       "tests/unit/antigravity-429-quota-tdd.test.ts",
+      "tests/unit/antigravity-prefer-stored-project.test.ts",
       "tests/unit/api-key-rotator-health.test.ts",
       "tests/unit/appearance-widget-settings-schema.test.ts",
       "tests/unit/auth-antigravity-account-retry-v2.test.ts",
@@ -206,6 +207,7 @@
       "tests/unit/error-sensitive-redaction.test.ts",
       "tests/unit/execute-chat-resource-pressure-breaker.test.ts",
       "tests/unit/executor-antigravity.test.ts",
+      "tests/unit/executor-devin-cli-agentic-acp.test.ts",
       "tests/unit/executor-web-cookie-sweep.test.ts",
       "tests/unit/format-provider-error-cause.test.ts",
       "tests/unit/forwarded-header-budget.test.ts",

tests/snapshots/provider/translate-path.json

@@ -1488,6 +1488,29 @@
       "stream": "devin://acp/stdio"
     }
   },
+  "devin-cli-agentic": {
+    "format": "claude",
+    "headers": {
+      "apiKey": {
+        "Accept": "text/event-stream",
+        "Authorization": "Bearer <TOK>",
+        "Content-Type": "application/json"
+      },
+      "nonStream": {
+        "Authorization": "Bearer <TOK>",
+        "Content-Type": "application/json"
+      },
+      "oauth": {
+        "Accept": "text/event-stream",
+        "Authorization": "Bearer <TOK>",
+        "Content-Type": "application/json"
+      }
+    },
+    "url": {
+      "nonStream": "devin://acp/stdio",
+      "stream": "devin://acp/stdio"
+    }
+  },
   "dgrid": {
     "format": "openai",
     "headers": {
@@ -4274,6 +4297,52 @@
       "stream": "https://chat.qwen.ai/api/v2/chat/completions"
     }
   },
+  "raycast": {
+    "format": "openai",
+    "headers": {
+      "apiKey": {
+        "Accept": "text/event-stream",
+        "Authorization": "Bearer <TOK>",
+        "Content-Type": "application/json"
+      },
+      "nonStream": {
+        "Authorization": "Bearer <TOK>",
+        "Content-Type": "application/json"
+      },
+      "oauth": {
+        "Accept": "text/event-stream",
+        "Authorization": "Bearer <TOK>",
+        "Content-Type": "application/json"
+      }
+    },
+    "url": {
+      "nonStream": "https://backend.raycast.com/api/v1/ai",
+      "stream": "https://backend.raycast.com/api/v1/ai"
+    }
+  },
+  "regolo": {
+    "format": "openai",
+    "headers": {
+      "apiKey": {
+        "Accept": "text/event-stream",
+        "Authorization": "Bearer <TOK>",
+        "Content-Type": "application/json"
+      },
+      "nonStream": {
+        "Authorization": "Bearer <TOK>",
+        "Content-Type": "application/json"
+      },
+      "oauth": {
+        "Accept": "text/event-stream",
+        "Authorization": "Bearer <TOK>",
+        "Content-Type": "application/json"
+      }
+    },
+    "url": {
+      "nonStream": "https://api.regolo.ai",
+      "stream": "https://api.regolo.ai"
+    }
+  },
   "reka": {
     "format": "openai",
     "headers": {
@@ -4829,6 +4898,29 @@
       "stream": "https://hermes.ai.unturf.com/v1/chat/completions"
     }
   },
+  "unorouter": {
+    "format": "openai",
+    "headers": {
+      "apiKey": {
+        "Accept": "text/event-stream",
+        "Authorization": "Bearer <TOK>",
+        "Content-Type": "application/json"
+      },
+      "nonStream": {
+        "Authorization": "Bearer <TOK>",
+        "Content-Type": "application/json"
+      },
+      "oauth": {
+        "Accept": "text/event-stream",
+        "Authorization": "Bearer <TOK>",
+        "Content-Type": "application/json"
+      }
+    },
+    "url": {
+      "nonStream": "https://api.unorouter.ai/v1/chat/completions",
+      "stream": "https://api.unorouter.ai/v1/chat/completions"
+    }
+  },
   "upstage": {
     "format": "openai",
     "headers": {

tests/unit/antigravity-prefer-stored-project.test.ts

@@ -0,0 +1,52 @@
+/**
+ * Regression guard for the #8894 import break: quotaStrategies.ts imported
+ * `preferAntigravityConnectionsWithStoredProject` from a module that never
+ * landed (`antigravityProjectPersistence.ts`), killing the whole combo module
+ * graph with ERR_MODULE_NOT_FOUND on the release tip. The helper now lives in
+ * the real persistence module (`antigravityProjectPersist.ts`); these tests pin
+ * its selection semantics.
+ *
+ * Run: node --import tsx/esm --test tests/unit/antigravity-prefer-stored-project.test.ts
+ */
+import { test } from "node:test";
+import assert from "node:assert/strict";
+
+import { preferAntigravityConnectionsWithStoredProject } from "../../open-sse/services/antigravityProjectPersist.ts";
+
+const withProject = { id: "a", projectId: "proj-1" };
+const withNestedProject = { id: "b", providerSpecificData: { projectId: "proj-2" } };
+const withStringPsd = { id: "c", providerSpecificData: '{"projectId":"proj-3"}' };
+const withoutProject = { id: "d", projectId: null, providerSpecificData: {} };
+const withBrokenPsd = { id: "e", providerSpecificData: "{not json" };
+
+test("prefers connections that already carry a stored projectId", () => {
+  const pool = [withoutProject, withProject, withNestedProject];
+  assert.deepEqual(
+    preferAntigravityConnectionsWithStoredProject(pool).map((c) => c.id),
+    ["a", "b"]
+  );
+});
+
+test("reads projectId from a raw JSON-string providerSpecificData", () => {
+  const pool = [withoutProject, withStringPsd];
+  assert.deepEqual(
+    preferAntigravityConnectionsWithStoredProject(pool).map((c) => c.id),
+    ["c"]
+  );
+});
+
+test("never empties the pool: no stored project anywhere → unchanged", () => {
+  const pool = [withoutProject, withBrokenPsd];
+  assert.deepEqual(preferAntigravityConnectionsWithStoredProject(pool), pool);
+});
+
+test("empty input passes through", () => {
+  assert.deepEqual(preferAntigravityConnectionsWithStoredProject([]), []);
+});
+
+test("the quota-strategy module graph resolves (the #8894 break shape)", async () => {
+  // Importing quotaStrategies transitively exercises the fixed import path; the
+  // pre-fix tip died here with ERR_MODULE_NOT_FOUND before any test could run.
+  const mod = await import("../../open-sse/services/combo/quotaStrategies.ts");
+  assert.ok(mod, "quotaStrategies must be importable");
+});

tests/unit/autoCombo/provider-family-combos.test.ts

@@ -87,10 +87,15 @@ describe("detectModelFamily (pure)", () => {
   });
 
   it("advertises exactly one auto/<family> catalog id per family", () => {
-    assert.deepEqual(
-      [...AUTO_FAMILY_IDS].sort(),
-      ["auto/gemini", "auto/gemma", "auto/glm", "auto/llama", "auto/mimo", "auto/minimax", "auto/zai"]
-    );
+    assert.deepEqual([...AUTO_FAMILY_IDS].sort(), [
+      "auto/gemini",
+      "auto/gemma",
+      "auto/glm",
+      "auto/llama",
+      "auto/mimo",
+      "auto/minimax",
+      "auto/zai",
+    ]);
   });
 });
 
@@ -133,7 +138,10 @@ describe("auto/<family> materialization (#6453)", () => {
     // "degrades gracefully" test below documents for opencode/minimax — a
     // no-auth backend that genuinely serves a family model IS a legitimate
     // member of the family pool, not just credentialed provider_connections rows.
-    assert.deepEqual(providerIds, ["auggie", "glm", "zai"]);
+    // `devin-cli-agentic` joined for the same documented reason as `auggie`:
+    // #8914 added the Devin ACP bridge whose catalog (registry/devin/catalog.ts)
+    // advertises the glm-5-2* line, so it genuinely serves the family.
+    assert.deepEqual(providerIds, ["auggie", "devin-cli-agentic", "glm", "zai"]);
     // Every candidate must be a glm-family model (the Cartesian pool now surfaces
     // each backend's full glm line-up, not only the glm-5.2 default), and the
     // connected openai/gpt-4o-mini backend must be excluded — same family
@@ -201,7 +209,6 @@ describe("auto/<family> materialization (#6453)", () => {
     );
   });
 
-
   it("rejects auto/<unknownfamily> with the same clean error as any unknown combo", async () => {
     await assert.rejects(
       () => builtinCatalog.createBuiltinAutoCombo("auto/unknownfam", "unknownfam"),

tests/unit/ccr-durable-store-9061.test.ts

@@ -119,6 +119,32 @@ describe("CCR engine survives losing its in-memory map (#9061)", () => {
     );
   });
 
+  it("re-admits the disk row into the fresh map (the enforceGlobalBudget arity bug)", async () => {
+    // The restart path re-admits a disk-served block through the same budgets a fresh
+    // store would face. #9061 called enforceGlobalBudget(entry.bytes) with ONE argument
+    // against a (owner, bytes) signature: `bytes` arrived undefined, `ccrTotalBytes +
+    // undefined` is NaN, and `NaN <= MAX` is false — so the re-admit never happened and
+    // the map stayed empty, re-reading from disk on every single retrieve. Typecheck
+    // caught the arity; this pins the observable behaviour.
+    const text = "z".repeat(2_000);
+    const stored = ccr.tryStoreBlock(text, "principal-readmit");
+    assert.equal(stored.stored, true);
+    await ccr.flushCcrDurableWrites();
+
+    const restarted = await import(`${ccrPath}?restart=9061-readmit`);
+    assert.equal(
+      restarted.getCcrStoreStats("principal-readmit").entries,
+      0,
+      "a fresh instance starts with an empty map"
+    );
+    assert.equal(restarted.retrieveBlock(stored.hash, "principal-readmit"), text);
+    assert.equal(
+      restarted.getCcrStoreStats("principal-readmit").entries,
+      1,
+      "the disk-served block must be re-admitted into the map, not re-read every time"
+    );
+  });
+
   it("keeps the principal boundary across the restart", async () => {
     const text = "y".repeat(2_000);
     const stored = ccr.tryStoreBlock(text, "principal-a");

tests/unit/chat-combo-live-test.test.ts

@@ -250,7 +250,13 @@ test("chat completions route emits early keepalive while waiting for stream read
   await seedHealthyConnection();
 
   globalThis.fetch = async () => {
-    await new Promise((resolve) => setTimeout(resolve, 100));
+    // Must exceed resolveKeepaliveThreshold()'s DEFAULT_THRESHOLD_MS (2000ms) for
+    // openai/* — otherwise withEarlyStreamKeepalive takes the FAST path, forwards
+    // the handler response as-is and no keepalive frame is ever emitted. The old
+    // 100ms only worked while unrelated handler latency happened to push the
+    // total past the threshold, which made this assertion incidental rather than
+    // deterministic; it stopped holding once the handler got faster.
+    await new Promise((resolve) => setTimeout(resolve, 2_400));
     return new Response(
       [
         `data: ${JSON.stringify({

tests/unit/check-db-rules-classification.test.ts

@@ -121,7 +121,7 @@ test("INTENTIONALLY_INTERNAL is exported from check-db-rules.mjs", () => {
   assert.ok(INTENTIONALLY_INTERNAL.size > 0, "INTENTIONALLY_INTERNAL must not be empty");
 });
 
-test("INTENTIONALLY_INTERNAL contains the expected 37 audited modules", () => {
+test("INTENTIONALLY_INTERNAL contains the expected 38 audited modules", () => {
   const expected = [
     "_rowTypes",
     "accessTokens",
@@ -147,6 +147,7 @@ test("INTENTIONALLY_INTERNAL contains the expected 37 audited modules", () => {
     "optimizationSettings",
     "pluginMetrics",
     "prompts",
+    "probeUtils",
     "providerNodeSelect",
     "providerStats",
     "proxyLatency",

tests/unit/combo-context-window-filter.test.ts

@@ -56,7 +56,11 @@ function capabilityEntry(limitContext: number | null) {
   };
 }
 
-function capabilityEntryWithLimits(limitInput: number | null, limitContext: number | null, limitOutput = 4096) {
+function capabilityEntryWithLimits(
+  limitInput: number | null,
+  limitContext: number | null,
+  limitOutput = 4096
+) {
   return {
     ...capabilityEntry(limitContext),
     limit_input: limitInput,
@@ -325,10 +329,12 @@ test("small input-only maxInputTokens keeps a target whose input fits even thoug
   );
 });
 
-test("input-only maxInputTokens still rejects when the input itself exceeds the cap", () => {
-  // The fix must not let a genuinely-too-small input cap pass. `too-small` has
-  // maxInputTokens = 1, which cannot even hold the ~11-token input, so it must
-  // still be dropped while the compatible target survives.
+test("input-only maxInputTokens is demoted when the input itself exceeds the cap", () => {
+  // #8944 made context metadata ADVISORY: a catalog-too-small target is no longer
+  // removed (a stale catalog entry must never delete the only target that could
+  // accept the request at runtime), it is ordered AFTER the known-fitting ones.
+  // `too-small` has maxInputTokens = 1, which cannot hold the ~11-token input, so
+  // it must lose the ordering to `huge` while remaining available as a fallback.
   saveModelsDevCapabilities({
     "unit-7039-too-small": {
       "too-small": capabilityEntryWithLimits(1, 1_000_000, 500),
@@ -344,14 +350,14 @@ test("input-only maxInputTokens still rejects when the input itself exceeds the
 
   assert.deepEqual(
     out.map((entry) => entry.modelStr),
-    ["unit-7039-too-small/huge"]
+    ["unit-7039-too-small/huge", "unit-7039-too-small/too-small"]
   );
 });
 
-test("maxInputTokens defaulting to contextWindow still rejects when input + output exceeds the total window (#7039 follow-up)", () => {
-  // Shared-window model where maxInputTokens equals the total window size.
-  // The input alone fits the input cap, but input + output overflows the
-  // window, so the target must be rejected instead of passing on the input cap.
+test("maxInputTokens defaulting to contextWindow is demoted when input + output exceeds the total window (#7039 follow-up)", () => {
+  // Shared-window model where maxInputTokens equals the total window size. The
+  // input alone fits the input cap but input + output overflows the window, so the
+  // target must not be PREFERRED — since #8944 it is demoted rather than dropped.
   saveModelsDevCapabilities({
     "unit-7039-window": {
       "shared-window": capabilityEntryWithLimits(400_000, 400_000, 200_000),
@@ -367,7 +373,7 @@ test("maxInputTokens defaulting to contextWindow still rejects when input + outp
 
   assert.deepEqual(
     out.map((entry) => entry.modelStr),
-    ["unit-7039-window/huge"]
+    ["unit-7039-window/huge", "unit-7039-window/shared-window"]
   );
 });
 
@@ -391,16 +397,18 @@ test("model_context_override lets a small-catalog target survive a large-context
       largeContextBody(),
       noopLog
     );
-    assert.deepEqual(
-      out.map((entry) => entry.modelStr).sort(),
-      ["unit-override/big", "unit-override/capped"]
-    );
+    assert.deepEqual(out.map((entry) => entry.modelStr).sort(), [
+      "unit-override/big",
+      "unit-override/capped",
+    ]);
   } finally {
     removeModelContextOverride("unit-override", "capped");
   }
 });
 
-test("without an override the small-catalog target is still dropped for the large request", () => {
+// #8944: "dropped" became "demoted" — the small-catalog target survives as a
+// runtime fallback but must never outrank the one whose known limit fits.
+test("without an override the small-catalog target is ordered last for the large request", () => {
   saveModelsDevCapabilities({
     "unit-override": {
       big: capabilityEntry(1_000_000),
@@ -417,6 +425,6 @@ test("without an override the small-catalog target is still dropped for the larg
 
   assert.deepEqual(
     out.map((entry) => entry.modelStr),
-    ["unit-override/big"]
+    ["unit-override/big", "unit-override/capped"]
   );
 });

tests/unit/compression/session-dedup-memory-7849.test.ts

@@ -8,8 +8,15 @@ import { sessionDedupEngine } from "../../../open-sse/services/compression/engin
 
 const REPO_ROOT = join(dirname(fileURLToPath(import.meta.url)), "../../..");
 const FIXTURE = join(REPO_ROOT, "tests/fixtures/compression/session-dedup-memory-7849.ts");
+// #7849 originally bounded session-dedup with a shared "suffix work budget" that
+// emitted SUFFIX_WORK_BUDGET_WARNING when exhausted. Commit 7f36b192f0 REPLACED
+// that mechanism with the MAX_SUFFIX_STARTS / MAX_TOTAL_BLOCK_BYTES guards and
+// removed both the budget and its warning. The invariant #7849 exists for — a
+// line-rich long context must not blow the heap, and the engine must fail open —
+// is unchanged and still guarded below; only the pins on the removed mechanism
+// were rewritten. The budget size is kept as the input-shaping constant that
+// produces the pathological pair.
 const SUFFIX_WORK_BUDGET = 32 * 1024 * 1024;
-const SUFFIX_WORK_BUDGET_WARNING = "session-dedup: skipped (suffix work budget exceeded)";
 
 function makeFixedWidthText(lineCount: number, lineChars: number, tag: string): string {
   return Array.from({ length: lineCount }, (_, index) => {
@@ -38,7 +45,7 @@ function makeSharedBudgetBody(): Record<string, unknown> {
   };
 }
 
-test("#7849: shares the two-pass suffix-work budget across all messages", () => {
+test("#7849: the two-message pathological pair stays bounded", () => {
   const body = makeSharedBudgetBody();
   const messages = body.messages as Array<{ content: string }>;
   const perMessageWork = messages.map(({ content }) => projectedSuffixWork(content, 2));
@@ -64,20 +71,29 @@ test("#7849: shares the two-pass suffix-work budget across all messages", () =>
     assert.equal(individualResult.stats, null, "each message must be accepted individually");
   }
 
+  // The pathological pair must be processed BOUNDED — quickly and without
+  // corrupting the body. Pre-#7849 this shape retained one full-length suffix
+  // string per line and OOM-killed the heap.
+  const started = Date.now();
   const result = sessionDedupEngine.apply(body);
-  assert.deepEqual(result.stats?.validationWarnings, [SUFFIX_WORK_BUDGET_WARNING]);
+  assert.ok(
+    Date.now() - started < 4000,
+    "the pathological pair must stay fast; quadratic work would take seconds"
+  );
+  assert.ok(Array.isArray((result.body as { messages?: unknown[] }).messages));
 });
 
-test("#7849: exhausted suffix-work budget fails open with exact zero-savings stats", () => {
+test("#7849: the pathological pair fails open, returning the input body untouched", () => {
   const body = makeSharedBudgetBody();
   const result = sessionDedupEngine.apply(body);
 
-  assert.strictEqual(result.body, body, "budget exhaustion must return the input body by identity");
+  // Fail-open is the surviving contract: nothing deduplicable in this shape, so
+  // the ORIGINAL body comes back by identity and no compression is claimed. The
+  // explanatory zero-savings stats belonged to the removed budget path — the
+  // current guards skip before producing any, so stats is null.
+  assert.strictEqual(result.body, body, "failing open must return the input body by identity");
   assert.equal(result.compressed, false);
-  assert.ok(result.stats, "budget exhaustion must return explanatory stats");
-  assert.equal(result.stats.originalTokens, result.stats.compressedTokens);
-  assert.equal(result.stats.savingsPercent, 0);
-  assert.deepEqual(result.stats.validationWarnings, [SUFFIX_WORK_BUDGET_WARNING]);
+  assert.equal(result.stats, null);
 });
 
 test("#7849: near-boundary under-budget request still deduplicates", () => {
@@ -130,9 +146,13 @@ test(
       warnings: string[];
     };
     assert.deepEqual(output.enginesRun, ["session-dedup", "lite", "rtk", "headroom", "caveman"]);
+    // The OOM guard is `child.status === 0` plus the full engine chain above:
+    // pre-fix this fixture killed the 512 MiB heap before the pipeline finished.
+    // session-dedup must still REPORT its skip; the exact reason string moved
+    // with the mechanism (7f36b192f0), so only the prefix is pinned.
     assert.ok(
-      output.warnings.includes("session-dedup: skipped (suffix work budget exceeded)"),
-      `expected an explicit session-dedup work-budget warning, got ${JSON.stringify(output.warnings)}`
+      output.warnings.some((warning) => warning.startsWith("session-dedup: skipped")),
+      `expected a session-dedup skip warning, got ${JSON.stringify(output.warnings)}`
     );
   }
 );

tests/unit/feature-flags-settings.test.ts

@@ -30,13 +30,13 @@ const {
   isControlPlaneProxyDirectFallbackEnabled,
 } = await import("../../src/shared/utils/featureFlags.ts");
 
-const EXPECTED_FEATURE_FLAG_COUNT = 45;
+const EXPECTED_FEATURE_FLAG_COUNT = 46;
 
 // ──────────────────────────────────────────────────────
 // Test group 1 — Flag definitions registry
 // ──────────────────────────────────────────────────────
 describe("featureFlagDefinitions", () => {
-  it("has exactly 45 flag definitions", () => {
+  it("has exactly 46 flag definitions", () => {
     assert.strictEqual(FEATURE_FLAG_DEFINITIONS.length, EXPECTED_FEATURE_FLAG_COUNT);
   });
 
@@ -332,7 +332,7 @@ describe("resolveFeatureFlag", () => {
   });
 
   describe("resolveAllFeatureFlags", () => {
-    it("returns all 45 flags", () => {
+    it("returns all 46 flags", () => {
       const all = resolveAllFeatureFlags();
       assert.strictEqual(all.length, EXPECTED_FEATURE_FLAG_COUNT);
     });

tests/unit/guardrails/vision-bridge-sse-and-reasoning.test.ts

@@ -21,7 +21,19 @@
  */
 import test from "node:test";
 import assert from "node:assert/strict";
-import { callVisionModel, type VisionModelConfig } from "@/lib/guardrails/visionBridgeHelpers";
+import {
+  callVisionModel as callVisionModelRaw,
+  type VisionModelConfig,
+} from "@/lib/guardrails/visionBridgeHelpers";
+
+// Inject the router's credential-check seam as INDETERMINATE (null): the fixed
+// model is used as-is and selection never touches the live connections DB — on a
+// clean box the suite otherwise dies with "No vision-capable provider connected",
+// and on a dev box auto-selection may swap the model under the assertions.
+const callVisionModel = (img: string, config: VisionModelConfig) =>
+  callVisionModelRaw(img, config, undefined, undefined, {
+    hasUsableCredentials: async () => null,
+  });
 
 const originalFetch = globalThis.fetch;
 

tests/unit/image-generation-route.test.ts

@@ -510,8 +510,14 @@ test("v1 image edit POST executes Codex through the configured connection proxy"
     host: "127.0.0.1",
     port: 1,
   });
+  // #9100: the reachability probe is NON-BLOCKING — dispatch is optimistic and the
+  // probe aborts the request only while it is still in flight (t14 pattern). The
+  // mock must stay pending: an instantly-throwing fetch would settle the race
+  // first and surface as a generic 502 upstream error instead of the proxy 503.
+  // Never resolved on purpose so the aborted continuation cannot proceed.
   globalThis.fetch = async () => {
-    throw new Error("Direct fetch must not run when the configured proxy is unreachable");
+    await new Promise(() => {});
+    throw new Error("unreachable");
   };
 
   const response = await imageEditRoute.POST(
@@ -537,8 +543,11 @@ test("v1 image generation POST resolves proxy and executes with proxy context wh
     port: 1, // intentionally unreachable — proves proxy path was taken
   });
 
+  // #9100 non-blocking probe: keep the request in flight so the fast-fail can
+  // abort it with the proxy-specific 503 (see the edit-route case above).
   globalThis.fetch = async () => {
-    throw new Error("fetch should not be called — proxy fast-fail should trigger first");
+    await new Promise(() => {});
+    throw new Error("unreachable");
   };
 
   const response = await imageRoute.POST(

tests/unit/login-bootstrap-route.test.ts

@@ -46,6 +46,8 @@ test("public login bootstrap route exposes the metadata the login page consumes"
 
   assert.equal(response.status, 200);
   assert.deepEqual(body, {
+    // #9491 added `authenticated` so /login can redirect an active session.
+    authenticated: false,
     requireLogin: true,
     hasPassword: false,
     setupComplete: true,
@@ -68,6 +70,8 @@ test("public login bootstrap route reports env-provided bootstrap password metad
 
   assert.equal(response.status, 200);
   assert.deepEqual(body, {
+    // #9491 added `authenticated` so /login can redirect an active session.
+    authenticated: false,
     requireLogin: true,
     hasPassword: true,
     setupComplete: true,
@@ -89,6 +93,8 @@ test("public login bootstrap route reports stored password metadata and disabled
 
   assert.equal(response.status, 200);
   assert.deepEqual(body, {
+    // #9491 added `authenticated` so /login can redirect an active session.
+    authenticated: false,
     requireLogin: false,
     hasPassword: true,
     setupComplete: true,

tests/unit/mcp-tool-count-dedup-6854.test.ts

@@ -6,7 +6,7 @@ import assert from "node:assert/strict";
 // (omniroute_agent_skills_list/get/coverage) are intentionally defined in BOTH
 // MCP_TOOLS (open-sse/mcp-server/schemas/tools.ts) and agentSkillTools
 // (open-sse/mcp-server/tools/agentSkillTools.ts), so the additive sum reported 121
-// while only 107 distinct tool names actually exist. countUniqueMcpTools
+// while only 108 distinct tool names actually exist. countUniqueMcpTools
 // (open-sse/mcp-server/toolCount.ts) fixes this by unioning tool names from every
 // registered collection into a Set, so each user-visible tool is counted once.
 
@@ -60,7 +60,7 @@ test("#6854: countUniqueMcpTools de-duplicates tools registered in multiple coll
   };
 
   const total = countUniqueMcpTools(collections);
-  assert.equal(total, 107, "the published MCP inventory must match the registered tool set");
+  assert.equal(total, 108, "the published MCP inventory must match the registered tool set");
 
   // Independently compute the "true" unique count by unioning every collection's
   // tool names into a Set — this must equal countUniqueMcpTools's own result AND

tests/unit/oauth-providers-config.test.ts

@@ -40,6 +40,7 @@ const {
   OAUTH_TIMEOUT,
   PROVIDERS: OAUTH_PROVIDER_IDS,
   QODER_CONFIG,
+  RAYCAST_CONFIG,
   TRAE_CONFIG,
   WINDSURF_CONFIG,
   XAI_OAUTH_CONFIG,
@@ -63,6 +64,7 @@ const EXPECTED_PROVIDER_KEYS = [
   "amazon-q",
   "cursor",
   "trae",
+  "raycast",
   "kilocode",
   "cline",
   "clinepass",
@@ -100,6 +102,7 @@ const EXPECTED_CONFIG_BY_PROVIDER = {
   clinepass: CLINE_CONFIG, // reuses the Cline WorkOS flow (clinepass: cline in providers/index.ts)
   windsurf: WINDSURF_CONFIG,
   "devin-cli": WINDSURF_CONFIG,
+  raycast: RAYCAST_CONFIG,
   trae: TRAE_CONFIG,
   "grok-cli": GROK_BUILD_OAUTH_CONFIG,
   "xai-oauth": XAI_OAUTH_CONFIG,

tests/unit/provider-limits-proxy-fail-closed.test.ts

@@ -101,6 +101,13 @@ test("Claude provider limits fail closed when an account proxy is unreachable",
   await withMockedFetch(
     (async (url) => {
       directFetchUrls.push(String(url));
+      // #9100: the reachability probe is NON-BLOCKING — dispatch is optimistic and
+      // the probe aborts the request only while it is still in flight (same shape
+      // as t14-proxy-fast-fail). The mock must therefore stay pending: an instant
+      // response would win the race and the fast-fail would never be observable.
+      // Never resolved on purpose — the aborted continuation must NOT proceed to a
+      // real (unmocked) fetch after this block restores globalThis.fetch.
+      await new Promise(() => {});
       return claudeUsageResponse();
     }) as typeof fetch,
     async () => {
@@ -108,10 +115,20 @@ test("Claude provider limits fail closed when an account proxy is unreachable",
         () => providerLimits.fetchAndPersistProviderLimits(connectionId, "manual"),
         /Proxy unreachable|fetch failed|ECONNREFUSED|UND_ERR_CONNECT_TIMEOUT/i
       );
+      // The fail-closed proof is twofold: (1) the rejection above settled at all —
+      // a direct retry would await the hung mock and never reject; (2) nothing
+      // egresses AFTER the fast-fail. The in-flight count itself may be >1: the
+      // Claude flow dispatches bootstrap + oauth/usage concurrently, and both are
+      // optimistic pre-abort attempts, not retries.
+      const urlsAtRejection = directFetchUrls.length;
+      await new Promise((resolve) => setTimeout(resolve, 50));
+      assert.equal(
+        directFetchUrls.length,
+        urlsAtRejection,
+        "account-proxied Claude usage must not egress anything after the fast-fail"
+      );
     }
   );
-
-  assert.deepEqual(directFetchUrls, [], "account-proxied Claude usage must not retry direct");
 });
 
 test("non-Claude OAuth provider limits fail closed when an account proxy is unreachable", async () => {
@@ -136,6 +153,9 @@ test("non-Claude OAuth provider limits fail closed when an account proxy is unre
   await withMockedFetch(
     (async (url) => {
       directFetchUrls.push(String(url));
+      // #9100: non-blocking probe — keep the request in flight so the fast-fail
+      // can abort it (see the Claude case above for the full rationale).
+      await new Promise(() => {});
       return new Response(
         JSON.stringify({
           copilot_plan: "free",
@@ -153,7 +173,10 @@ test("non-Claude OAuth provider limits fail closed when an account proxy is unre
     }
   );
 
-  assert.deepEqual(directFetchUrls, [], "account-proxied OAuth usage must not retry direct");
+  assert.ok(
+    directFetchUrls.length <= 1,
+    "at most the single optimistic in-flight attempt — account-proxied OAuth usage must never retry direct after the fast-fail"
+  );
 });
 
 test("Claude provider limits preserve direct retry for non-account proxy failures", async () => {

tests/unit/provider-models-route.test.ts

@@ -934,7 +934,8 @@ test("provider models route retries Antigravity discovery endpoints before retur
     // After PR #2219, the discovery flow calls loadCodeAssist first as a project
     // bootstrap; treat all bootstrap calls as non-fatal failures so the test
     // exercises the discovery retry path.
-    if (urlString.includes("/v1internal:loadCodeAssist")) {
+    // onboardUser is a bootstrap hop too (ff012ff420) — else it eats the single 503 below.
+    if (urlString.includes(":loadCodeAssist") || urlString.includes(":onboardUser")) {
       return new Response("nope", { status: 503 });
     }
     seenUrls.push(urlString);
@@ -983,6 +984,8 @@ test("provider models route retries Antigravity discovery endpoints before retur
     "https://cloudcode-pa.googleapis.com/v1internal:fetchAvailableModels",
   ]);
   assert.deepEqual(body.models, [
+    // #9106: both alias ids are user-callable now, so the upstream echo survives the filter.
+    { id: "gemini-3.1-pro-high", name: "Gemini 3.1 Pro (High)" },
     { id: "gemini-pro-agent", name: "Gemini 3.1 Pro (High)" },
     { id: "gemini-3.6-flash-high", name: "Gemini 3.6 Flash (High)" },
     { id: "gemini-3.6-flash-medium", name: "Gemini 3.6 Flash (Medium)" },

tests/unit/providers-constants-split.test.ts

@@ -1,7 +1,7 @@
 // Characterization of the providers.ts catalog split (god-file decomposition): the host became a
 // barrel that re-exports 10 data catalogs now living under constants/providers/*, and APIKEY is
 // merged from 6 semantic family files (apikey/<family>.ts). Locks: the public surface (every catalog
-// + helpers still exported), the spread-merge integrity (197 APIKEY entries, no loss/dup), and that
+// + helpers still exported), the spread-merge integrity (198 APIKEY entries, no loss/dup), and that
 // load-time Zod validation still runs. Pure-data move → behavior must be identical.
 // Count was 171 before obsolete provider removals (PR #6675: glhf/kluster/cablyai/inclusionai etc.,
 // 171->167) plus #6126 (ClinePass dual-auth): the API-key-only APIKEY_PROVIDERS_GATEWAYS entry was
@@ -17,7 +17,7 @@
 // sarvam+plamo in regional) to 193, then #8170 (inception/typhoon — inception in frontier-labs,
 // typhoon in regional) to 195, then Firecrawl dual search+fetch under SEARCH_PROVIDERS.firecrawl
 // (removed specialty-media duplicate) to 194, #8861 (Xiaomi MiMo Token Plan, regional) to 195, and
-// the Cheaper Inference gateway (OSS-sponsor reseller, gateways family) to 197 (UnoRouter, #9009).
+// the Cheaper Inference gateway (OSS-sponsor reseller, gateways family) to 198 (UnoRouter #9009, Raycast Pro #8895).
 import { test } from "node:test";
 import assert from "node:assert/strict";
 
@@ -46,12 +46,12 @@ test("barrel still exports every catalog + key helpers", () => {
   }
 });
 
-test("APIKEY_PROVIDERS merges the 6 family files into 197 entries (no loss / no dup)", async () => {
+test("APIKEY_PROVIDERS merges the 6 family files into 198 entries (no loss / no dup)", async () => {
   const keys = Object.keys((P as Record<string, object>).APIKEY_PROVIDERS);
-  assert.equal(keys.length, 197);
-  assert.equal(new Set(keys).size, 197, "duplicate keys after spread-merge");
+  assert.equal(keys.length, 198);
+  assert.equal(new Set(keys).size, 198, "duplicate keys after spread-merge");
   // the merged object's entry-count equals the sum of the 6 semantic family files; families are a
-  // strict partition (every provider in exactly one), so the sum must be exactly 197.
+  // strict partition (every provider in exactly one), so the sum must be exactly 198.
   const families: [string, string][] = [
     ["gateways", "APIKEY_PROVIDERS_GATEWAYS"],
     ["frontier-labs", "APIKEY_PROVIDERS_FRONTIER"],
@@ -71,7 +71,7 @@ test("APIKEY_PROVIDERS merges the 6 family files into 197 entries (no loss / no
       seen.add(k);
     }
   }
-  assert.equal(famTotal, 197, "families must partition all 197 providers");
+  assert.equal(famTotal, 198, "families must partition all 198 providers");
 });
 
 test("AI_PROVIDERS Proxy aggregates all sections; lookups resolve", () => {

tests/unit/ratelimit-reservoir-refresh.test.ts

@@ -85,15 +85,20 @@ test("reservoir keeps refreshing after updateSettings() touches an already-heart
   // number of event-loop ticks: Bottleneck's own updateSettings() goes through
   // at least one real setTimeout(0) (yieldLoop) before storeOptions reflects the
   // new value.
+  // #9604 replaced Bottleneck's fixed-window reservoir with the rolling lease gate
+  // (open-sse/services/rollingRpmGate.ts), so `reservoir` is null now and pinning it
+  // would assert a mechanism that no longer exists. What must still hold — and what
+  // the Bottleneck heartbeat bug actually broke — is that the limiter SURVIVES the
+  // header-learned updateSettings() and keeps admitting work (steps 3 and 4 below).
   const pollDeadline = Date.now() + 2000;
   let state = await rateLimitManager.__getLimiterStateForTests(PROVIDER, CONNECTION_ID, null);
-  while (state?.reservoir !== 2 && Date.now() < pollDeadline) {
+  while (!state && Date.now() < pollDeadline) {
     await wait(10);
     state = await rateLimitManager.__getLimiterStateForTests(PROVIDER, CONNECTION_ID, null);
   }
-  assert.equal(state?.reservoir, 2, "reservoir must land at 2 before the slots below are consumed");
+  assert.ok(state, "the limiter must still exist after the header-learned update");
 
-  // 3. Consume both reservoir slots.
+  // 3. Consume the learned capacity.
   assert.equal(
     await rateLimitManager.withRateLimit(PROVIDER, CONNECTION_ID, null, async () => "slot-1"),
     "slot-1"
@@ -103,10 +108,9 @@ test("reservoir keeps refreshing after updateSettings() touches an already-heart
     "slot-2"
   );
 
-  // 4. Reservoir is now 0. A healthy Bottleneck heartbeat refills it ~1s later
-  // from the reservoirRefreshInterval/reservoirRefreshAmount configured above.
-  // Race a 3rd request against a 5s timer: if the heartbeat died (unfixed bug),
-  // the request stays QUEUED forever and the timer wins instead.
+  // 4. Capacity is spent. Race a 3rd request against a 5s timer: if the limiter
+  // stopped pacing after updateSettings() (the original bug) the request stays
+  // queued forever and the timer wins instead.
   const RACE_TIMEOUT_MS = 5000;
   let timeoutHandle: ReturnType<typeof setTimeout> | undefined;
   const timeout = new Promise<"timed-out">((resolve) => {
@@ -125,8 +129,8 @@ test("reservoir keeps refreshing after updateSettings() touches an already-heart
   assert.equal(
     result,
     "slot-3",
-    'reservoir must refresh ~1s after being exhausted; "timed-out" means the Bottleneck ' +
-      "heartbeat died after updateSettings() and the reservoir never refilled " +
-      "(node_modules/bottleneck/lib/LocalDatastore.js _startHeartbeat clearInterval-without-null bug)"
+    'capacity must recover after being exhausted; "timed-out" means the limiter stopped ' +
+      "admitting work after the header-learned updateSettings() — the failure shape of the " +
+      "original Bottleneck heartbeat bug (LocalDatastore _startHeartbeat clearInterval-without-null)"
   );
 });

tests/unit/responses-commentary-passthrough-6199.test.ts

@@ -260,7 +260,10 @@ test("response.completed normalizes usage when lifecycle echoes are stripped", a
 
   const completed = JSON.parse(completedLine.slice(5).trim());
   assert.equal("instructions" in completed.response, false);
-  assert.equal("tools" in completed.response, false);
+  // #8990 (commit c996dc93c2) deliberately stopped stripping `tools` from the
+  // TERMINAL snapshot — Codex CLI rebuilds its tool list from response.completed.
+  // stripResponsesLifecycleEcho still strips it on created/in_progress.
+  assert.deepEqual(completed.response.tools, [{ type: "function", name: "echoed_tool" }]);
   assert.equal(completed.response.usage.total_tokens, 91);
 });
 

tests/unit/translator-openai-to-gemini.test.ts

@@ -582,7 +582,10 @@ test("OpenAI -> Antigravity wraps Gemini requests in a Cloud Code envelope", ()
     "model",
     "userAgent",
     "requestType",
+    // #9568: identity entries are emitted too (Gemini lowercases tool names in responses).
+    "_toolNameMap",
   ]);
+  assert.equal((result._toolNameMap as Map<string, string>).get("weather"), "weather");
   assert.equal(result.userAgent, "antigravity");
   assert.equal(result.requestType, "agent");
   assert.match(result.requestId, /^agent\/\d+\/[0-9a-f]{8}$/);

tests/unit/vertex-functioncall-id-3440.test.ts

@@ -20,6 +20,11 @@ const { buildGeminiThoughtSignatureKey, storeGeminiThoughtSignature } =
 
 type UnknownRecord = Record<string, unknown>;
 
+/** Shape of the translated Gemini request the three finders below walk. */
+interface GeminiRequestLike {
+  contents?: Array<{ parts?: UnknownRecord[] }>;
+}
+
 const CLAUDE_SIGNATURE_NAMESPACE = "regression-3440";
 
 function seedClaudeThoughtSignature() {
@@ -29,7 +34,7 @@ function seedClaudeThoughtSignature() {
   );
 }
 
-function findFunctionCall(result: any): UnknownRecord | undefined {
+function findFunctionCall(result: GeminiRequestLike): UnknownRecord | undefined {
   for (const content of result.contents ?? []) {
     for (const part of content.parts ?? []) {
       if (part?.functionCall) return part.functionCall as UnknownRecord;
@@ -38,7 +43,7 @@ function findFunctionCall(result: any): UnknownRecord | undefined {
   return undefined;
 }
 
-function findFunctionCallPart(result: any): UnknownRecord | undefined {
+function findFunctionCallPart(result: GeminiRequestLike): UnknownRecord | undefined {
   for (const content of result.contents ?? []) {
     for (const part of content.parts ?? []) {
       if (part?.functionCall) return part as UnknownRecord;
@@ -47,7 +52,7 @@ function findFunctionCallPart(result: any): UnknownRecord | undefined {
   return undefined;
 }
 
-function findFunctionResponse(result: any): UnknownRecord | undefined {
+function findFunctionResponse(result: GeminiRequestLike): UnknownRecord | undefined {
   for (const content of result.contents ?? []) {
     for (const part of content.parts ?? []) {
       if (part?.functionResponse) return part.functionResponse as UnknownRecord;

tests/unit/vision-compression-authoritative-capability-7237.test.ts

@@ -36,17 +36,21 @@ function imageBody() {
 }
 
 describe("#7237 vision-capable models keep their images through compression", () => {
-  it("documents the drift: the conservative id-fragment heuristic disagrees with the authoritative spec for gpt-5.5", () => {
-    assert.equal(
-      isVisionModelId("gpt-5.5"),
-      false,
-      "the fragment-list heuristic has no gpt-5.x entry — it is a deliberately conservative fallback, not the source of truth"
-    );
+  it("the id-fragment heuristic now agrees with the authoritative spec for gpt-5.5", () => {
+    // Originally this case documented a DRIFT: the fragment list had no gpt-5.x
+    // entry, so the heuristic said false while modelSpecs said true. Commit
+    // 68cb678780 (enable vision flags for CC models) added the "gpt-5" fragment
+    // and closed that gap, so the two sources now converge. The invariant worth
+    // guarding is the AGREEMENT plus the fact that modelSpecs stays authoritative.
+    assert.equal(isVisionModelId("gpt-5.5"), true);
     assert.equal(
       getResolvedModelCapabilities({ model: "gpt-5.5" }).supportsVision,
       true,
       "modelSpecs.ts registers gpt-5.5 with supportsVision:true — this is the authoritative source chatCore must use"
     );
+    // The heuristic stays deliberately conservative for ids it does not know;
+    // chatCore must still read the authoritative capability, never this fallback.
+    assert.equal(isVisionModelId("some-unknown-text-only-model"), false);
   });
 
   it("replaceImageUrls preserves the image when fed the authoritative capability (the fixed chatCore.ts:1330 behavior)", () => {
@@ -59,13 +63,16 @@ describe("#7237 vision-capable models keep their images through compression", ()
     assert.equal(content[0].type, "image_url", "the block must remain a real image_url block");
   });
 
-  it("regresses the pre-fix bug: feeding the raw heuristic value strips the image for gpt-5.5", () => {
-    const buggyValue = isVisionModelId("gpt-5.5"); // false — the pre-fix chatCore.ts:1330 input
-    const result = replaceImageUrls(imageBody(), { supportsVision: buggyValue });
+  it("reproduces the bug SHAPE: a false supportsVision strips the image to a placeholder", () => {
+    // The original case derived the wrong value from isVisionModelId("gpt-5.5").
+    // That no longer returns false (68cb678780), so the false is now supplied
+    // directly — what this guards is the stripping behaviour itself, which is
+    // exactly why chatCore must pass the authoritative capability and not a guess.
+    const result = replaceImageUrls(imageBody(), { supportsVision: false });
     assert.equal(
       result.applied,
       true,
-      "sanity check: this reproduces the bug shape when fed the wrong (heuristic) value"
+      "sanity check: this reproduces the bug shape when fed the wrong value"
     );
   });
 

vitest.mcp.config.ts

@@ -14,6 +14,9 @@ export default defineConfig({
       "open-sse/services/autoCombo/__tests__/**/*.test.ts",
       "open-sse/services/combo/__tests__/**/*.test.ts",
       "open-sse/services/__tests__/antigravity-quota-family.test.ts",
+      // #8890 shipped this suite into a directory no runner collects, so it had
+      // never executed once (check:test-discovery flags it as a NEW orphan).
+      "open-sse/services/__tests__/fail-fast-concurrency-gate.test.ts",
       "src/lib/memory/__tests__/generic-backend.test.ts",
       "tests/unit/autoCombo/**/*.test.ts",
       "tests/unit/encryption.spec.ts",