board #11389

# Conflicts:
#	README.md
#	docs/diagrams/cli-terminal.svg
#	docs/diagrams/promise-pillars.svg

AGENTS.md

@@ -46,7 +46,7 @@ Repository map and Reference Documentation sections below.
 
 ## Project at a Glance
 
-**OmniRoute** — unified AI proxy/router. One endpoint, 350 LLM providers, auto-fallback.
+**OmniRoute** — unified AI proxy/router. One endpoint, 351 LLM providers, auto-fallback.
 
 | Layer         | Location                | Purpose                                                                                                                                                                   |
 | ------------- | ----------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |

Dockerfile.bun

@@ -1,5 +1,5 @@
 # ── Multi-stage Dockerfile for Native Bun Runtime (web-latest-bun) ───────────
-FROM oven/bun:1.3.14-slim AS base
+FROM oven/bun:1.4.0-slim AS base
 WORKDIR /app
 
 RUN apt-get update \
@@ -50,7 +50,7 @@ ENV NODE_ENV=production
 RUN bun run --quiet build
 
 # ── Runner Base stage (100% Bun Native Production Runtime) ──────────────────
-FROM oven/bun:1.3.14-slim AS runner-base
+FROM oven/bun:1.4.0-slim AS runner-base
 
 LABEL org.opencontainers.image.title="omniroute" \
   org.opencontainers.image.description="Unified AI proxy — route any LLM through one endpoint (Bun Native)" \

README.md

@@ -7,7 +7,7 @@
 
 # 🚀 OmniRoute — The Free AI Gateway
 
-<img src="./docs/diagrams/readme-hero.svg" width="100%" alt="OmniRoute — Never stop coding. Every AI tool → 350 providers — 90+ free — through one endpoint. Claude Code, Codex, Cursor, Cline, Copilot & Antigravity into FREE Claude / GPT / Gemini with auto-fallback. RTK + Caveman stacked compression saves 15–95% tokens (~89% avg) — never hit limits. 350 AI providers · 90+ free tiers · ~1.51B free tokens/mo · 19 routing strategies · $0 to start."/>
+<img src="./docs/diagrams/readme-hero.svg" width="100%" alt="OmniRoute — Never stop coding. Every AI tool → 351 providers — 90+ free — through one endpoint. Claude Code, Codex, Cursor, Cline, Copilot & Antigravity into FREE Claude / GPT / Gemini with auto-fallback. RTK + Caveman stacked compression saves 15–95% tokens (~89% avg) — never hit limits. 351 AI providers · 90+ free tiers · ~1.51B free tokens/mo · 19 routing strategies · $0 to start."/>
 
 </div>
 
@@ -210,7 +210,7 @@ curl http://localhost:20128/v1/chat/completions \
 
 </div>
 
-<img src="./docs/diagrams/promise-pillars.svg" width="100%" alt="The Promise — One endpoint and 350 providers. Automatic fallback keeps routing while another healthy target is available. Six pillars: resilient fallback across 350 providers · up to 95% token savings on eligible workloads · $0 to start with 90+ free tiers and 56 recurring/keyless free-forever providers · 35 CLI/agent integrations through one config · OpenAI, Claude, Gemini and Responses API compatibility at /v1 · production controls including circuit breakers, TLS stealth, MCP 110 tools, A2A, memory, guardrails, evals and 39,000+ static test declarations across 5,100+ tracked test files."/>
+<img src="./docs/diagrams/promise-pillars.svg" width="100%" alt="The Promise — One endpoint and 351 providers. Automatic fallback keeps routing while another healthy target is available. Six pillars: resilient fallback across 351 providers · up to 95% token savings on eligible workloads · $0 to start with 90+ free tiers and 56 recurring/keyless free-forever providers · 35 CLI/agent integrations through one config · OpenAI, Claude, Gemini and Responses API compatibility at /v1 · production controls including circuit breakers, TLS stealth, MCP 110 tools, A2A, memory, guardrails, evals and 39,000+ static test declarations across 5,100+ tracked test files."/>
 
 <br/>
 <br/>
@@ -461,7 +461,7 @@ All **19** strategies — mix & match per combo step:
 
 </div>
 
-<img src="./docs/diagrams/comparison-table.svg" width="100%" alt="What sets OmniRoute apart — a dated feature snapshot vs 9router, OpenRouter, CLIProxyAPI and LiteLLM across 13 capabilities. OmniRoute: 350 providers, 90+ free tiers built in, 19 routing strategies, 12-engine token compression, built-in MCP server with 110 tools, A2A agent protocol, persistent memory, guardrails, cloud agents, TLS fingerprint stealth, Desktop/Termux/PWA and 43 i18n UI locales. OmniRoute is MIT-licensed and self-hostable. Competitor capabilities and counts may change; see the linked methodology."/>
+<img src="./docs/diagrams/comparison-table.svg" width="100%" alt="What sets OmniRoute apart — a dated feature snapshot vs 9router, OpenRouter, CLIProxyAPI and LiteLLM across 13 capabilities. OmniRoute: 351 providers, 90+ free tiers built in, 19 routing strategies, 12-engine token compression, built-in MCP server with 110 tools, A2A agent protocol, persistent memory, guardrails, cloud agents, TLS fingerprint stealth, Desktop/Termux/PWA and 43 i18n UI locales. OmniRoute is MIT-licensed and self-hostable. Competitor capabilities and counts may change; see the linked methodology."/>
 
 <sub>📊 Full methodology &amp; per-feature detail vs 9router, OpenRouter, CLIProxyAPI &amp; LiteLLM → [`docs/comparison/OMNIROUTE_VS_ALTERNATIVES.md`](docs/comparison/OMNIROUTE_VS_ALTERNATIVES.md)</sub>
 
@@ -559,7 +559,7 @@ the current catalog at **[radar.omniroute.online/planos](https://radar.omniroute
 - **🖼️ New endpoints** — `/v1/ocr` (Mistral OCR) and `/v1/audio/translations` (Whisper-style) round out the media surface. → [API Reference](docs/reference/API_REFERENCE.md)
 - **🎨 Image / video / audio generation** — one API for media: xAI Grok Imagine & Novita AI video, ComfyUI, Freepik, Adobe Firefly, Microsoft Designer, Segmind, EdgeTTS. → [API Reference](docs/reference/API_REFERENCE.md)
 - **🌍 Deployment & ops** — reverse-proxy `basePath`, browser-language auto-detect, per-key device tracking, root-less MITM trust, zh-TW localization. → [Environment](docs/reference/ENVIRONMENT.md)
-- **🤝 More providers & agents** — Cursor Cloud Agent, Grok Build (xAI) with browser + OAuth login, Ollama first-class card, Claude Opus 5 & Sonnet 5, Kimi official partnership (Code/Web/Moonshot), Zed, Requesty, SenseNova, Yuanbao, Agnes AI… and a refreshed **350-provider catalog**. → [Providers](docs/reference/PROVIDER_REFERENCE.md)
+- **🤝 More providers & agents** — Cursor Cloud Agent, Grok Build (xAI) with browser + OAuth login, Ollama first-class card, Claude Opus 5 & Sonnet 5, Kimi official partnership (Code/Web/Moonshot), Zed, Requesty, SenseNova, Yuanbao, Agnes AI… and a refreshed **351-provider catalog**. → [Providers](docs/reference/PROVIDER_REFERENCE.md)
 - **📡 Routing transparency** — every response carries an `X-OmniRoute-Decision` header naming the strategy/provider/latency that served it, a new `cache-optimized` combo strategy + Auto-Combo `cacheAffinity` factor route repeat requests back to the connection holding the cached prefix, and a read-only `/v1/auto-combo/{channel}/candidates` endpoint exposes an `auto/*` channel's live candidate pool. → [Auto-Combo](docs/routing/AUTO-COMBO.md)
 - **⚡ Local performance & infra** — one-click local Redis, Cloudflare Workers / Deno Deploy relay deployers, Bifrost & Mux as supervised embedded services. → [Embedded Services](docs/frameworks/EMBEDDED-SERVICES.md)
 
@@ -642,11 +642,11 @@ of your shell history. → [CLI Integrations](docs/guides/CLI-INTEGRATIONS.md)
 
 <div align="center">
 
-## 🌐 350 AI Providers — 154 Catalog-Marked Free
+## 🌐 351 AI Providers — 154 Catalog-Marked Free
 
 </div>
 
-> **350 registered providers** across the canonical chat, media, search, local, cloud-agent and system collections, including **154 carrying `hasFree: true` discovery metadata**. The chat model registry covers **268 providers / 2,566 distinct provider-model pairs / 1,312 raw model IDs**; the separate free-budget catalog has **455 per-model rows**, **40 recurring pools** and **56 recurring/keyless free-forever providers**. These are different denominators by design; definitions and pool-deduped calculations live in the [Provider Reference](docs/reference/PROVIDER_REFERENCE.md) and [Free Tiers](docs/reference/FREE_TIERS.md).
+> **351 registered providers** across the canonical chat, media, search, local, cloud-agent and system collections, including **154 carrying `hasFree: true` discovery metadata**. The chat model registry covers **268 providers / 2,566 distinct provider-model pairs / 1,312 raw model IDs**; the separate free-budget catalog has **455 per-model rows**, **40 recurring pools** and **56 recurring/keyless free-forever providers**. These are different denominators by design; definitions and pool-deduped calculations live in the [Provider Reference](docs/reference/PROVIDER_REFERENCE.md) and [Free Tiers](docs/reference/FREE_TIERS.md).
 
 <div align="center">
 

changelog.d/features/11340-web-session-contract.md

@@ -0,0 +1 @@
+- **feat(providers):** publish a management-authenticated versioned web-session credential contract from OmniRoute's canonical browser credential metadata ([#11340](https://github.com/diegosouzapw/OmniRoute/pull/11340)) — thanks @Zartharas

changelog.d/features/11370-xquik-search-provider.md

@@ -0,0 +1 @@
+- **feat(search):** Add Xquik X search with typed results, credential validation, REST routing, and MCP selection ([#11370](https://github.com/diegosouzapw/OmniRoute/pull/11370)) — thanks @kriptoburak

changelog.d/features/11389-exclusive-managed-session-dashboard.md

@@ -0,0 +1 @@
+- **feat(dashboard):** surface durable exclusive managed leases in the existing Sessions view, keeping leased clients visible across idle gaps while marking connections with in-flight work as active ([#11389](https://github.com/diegosouzapw/OmniRoute/pull/11389)) — thanks @KaspaPulse

changelog.d/fixes/11441-discontinued-free-models.md

@@ -0,0 +1 @@
+- **fix(free-models):** the shared `isFreeModel()` predicate no longer reports catalog entries marked `freeType: "discontinued"` as free, so `hidePaidModels` can't route to Pollinations' seven premium models that now require a paid key ([#11441](https://github.com/diegosouzapw/OmniRoute/pull/11441))

changelog.d/fixes/7592-windows-packaged-smoke.md

@@ -0,0 +1 @@
+- **fix(electron):** the packaged Windows desktop build now passes the #7592 cold-restart smoke end to end. Five defects were found and fixed on the way: (1) optional-pack staging failed on any machine whose `tar` is GNU tar — it reads the drive letter in an absolute `-f C:\...` archive path as a remote rsh host (`Cannot connect to C:`), so staging now passes a bare filename with `cwd` at the tarball directory and surfaces tar stderr on failure; (2) the Electron `files` allowlist was missing `lib/loginHeaderCapture.js`, so the packaged main process crashed at startup with `Cannot find module './lib/loginHeaderCapture'`; (3) electron-builder ≥26 injects an `!**/node_modules/**` ignore into every extraResources pattern list that no positive filter can override, which silently dropped the staged runtime `node_modules` (including the better-sqlite3 N-API prebuild) from `resources/app` — a fresh v3.8.50 desktop build would have shipped with no native SQLite driver and reproduced the #7592 sql.js fallback on every machine; an `afterPack` hook now restores the staged `node_modules` after packing; (4) the packaged-app smoke harness redirected `USERPROFILE` into its temp DATA_DIR, but Electron resolves the Roaming profile from `%USERPROFILE%\AppData\Roaming\<name>` (USERPROFILE takes precedence over the APPDATA env var) and the path service throws instead of creating it, so `requestSingleInstanceLock()` returned false and the app exited(0) silently before `app.whenReady()` — the harness now pre-creates the derived tree, and `ensureSmokeEnvDirs` is exported and unit-tested; (5) the #7592 driver assertion parsed a `[DB] Driver: ...` line that the server's primary DB path never printed (only the unused `openDatabaseAsync()` did) — `getDbInstance()` now emits the same line on open so the guard can actually assert the native driver. Also: the smoke env-allowlist unit test hardcoded POSIX paths and could only pass on Linux/macOS; expectations are now host-agnostic, with new regression tests for the USERPROFILE-derived tree and `tarPack` under absolute Windows-style paths

changelog.d/fixes/memory-embedding-quick-select-providers.md

@@ -0,0 +1 @@
+- **fix(memory):** Embedding Model Quick select, Embedding Source remote dropdown, and Rerank selector now list every configured provider with embedding/rerank support instead of only chat-catalog text matches plus OpenRouter live discovery; a generic OpenAI-compatible `/embeddings` + Cohere-compatible `/rerank` runtime fallback resolves any configured chat provider's embedding/rerank endpoint, so unlisted providers no longer fail with "Unknown embedding provider"; both memory selectors gained a free-text model override

config/quality/dependency-allowlist.json

@@ -12,6 +12,7 @@
     "@dnd-kit/core",
     "@dnd-kit/sortable",
     "@dnd-kit/utilities",
+    "@eslint/compat",
     "@huggingface/transformers",
     "@lobehub/icons",
     "@modelcontextprotocol/sdk",
@@ -60,6 +61,7 @@
     "eslint",
     "eslint-config-next",
     "eslint-plugin-sonarjs",
+    "espree",
     "express",
     "fast-check",
     "fetch-socks",

config/quality/eslint-suppressions.json

@@ -2372,16 +2372,6 @@
       "count": 2
     }
   },
-  "src/lib/guardrails/videoBridgeContactSheet.ts": {
-    "@typescript-eslint/no-unused-vars": {
-      "count": 1
-    }
-  },
-  "src/lib/guardrails/videoBridgeRuntime.ts": {
-    "@typescript-eslint/no-unused-vars": {
-      "count": 1
-    }
-  },
   "src/lib/guardrails/visionBridgeHelpers.ts": {
     "@typescript-eslint/no-unused-vars": {
       "count": 1
@@ -2422,11 +2412,6 @@
       "count": 1
     }
   },
-  "src/lib/memory/embedding/index.ts": {
-    "no-restricted-imports": {
-      "count": 1
-    }
-  },
   "src/lib/memory/genericBackend.ts": {
     "@typescript-eslint/no-unused-vars": {
       "count": 1

docs/changelog/fragments/7778.md

@@ -0,0 +1 @@
+feat(resilience): atomically enforce cumulative global, provider, and account concurrency limits

docs/design/EXTERNAL-DB-BACKEND.md

@@ -0,0 +1,98 @@
+# Design: Pluggable External Durable-State DB Backend (PostgreSQL / MySQL)
+
+Status: PROPOSAL · Discussion issue: [#8075](https://github.com/diegosouzapw/OmniRoute/issues/8075) · Author: @oyi77
+
+Grounded in the tree at `release/v3.8.51` (`3192eb88d`). Every number below was measured on that tree, not estimated.
+
+---
+
+## 0. Why this document exists
+
+#8075 asks five questions (backends & order, abstraction contract, what stays SQLite-only, config surface, migration path). This document answers them with a census of how deeply the current data layer is coupled to SQLite, and proposes a repository-contract seam that lets external backends be adopted per domain instead of via a big-bang driver swap.
+
+## 1. Measured coupling census (`release/v3.8.51`)
+
+| Surface | Count | Implication for external backends |
+|---|---|---|
+| `src/lib/db/*.ts` modules calling `.prepare()` / `.transaction()` | **126 files** | The synchronous SQLite call shape is the de-facto data API; every module is a translation candidate |
+| `.pragma()` call sites | **36** | No equivalent in PG/MySQL — each needs an audit (many are performance tuning that becomes server config) |
+| SQLite migrations on disk | **159** | A second dialect needs its own migration lineage, not a translator |
+| `sqlite-vec` / `vec0` referencing files | **9** | Vector search must move to an external store (Qdrant proposed in #8075) |
+| FTS5 referencing files | **8** | Maps to Postgres `tsvector` / MySQL `FULLTEXT` — dialect work |
+| Existing driver adapters | **4** (`better-sqlite3`, `node:sqlite`, `bun:sqlite`, `sql.js`) behind `SqliteAdapter` (`src/lib/db/adapters/types.ts`) | Proves a seam exists at the DRIVER level — but all four are SQLite; the seam OmniRoute actually needs is one level up |
+
+The decisive fact: **the adapter interface is synchronous** (`prepare().run/get/all`, sync closure transactions). PostgreSQL and MySQL drivers are inherently async. A faithful external backend behind `SqliteAdapter` is therefore impossible without faking sync over async — which is exactly the trap #8037/#8073 already flagged ("not a drop-in ORM/driver swap"). The contract must live ABOVE the adapter layer.
+
+## 2. Proposed contract: durable-state repositories, not a driver swap
+
+Introduce a `DurableStateStore` port with per-domain repositories, defined once and implemented per dialect:
+
+```ts
+// src/lib/db/ports/ — dialect-agnostic, async, transactional by unit-of-work
+interface UnitOfWork {
+  connections: ConnectionRepository;   // provider connections + credentials
+  apiKeys: ApiKeyRepository;           // keys, groups, usage-limit fields
+  combos: ComboRepository;             // combo defs + routing policies
+  quotaState: QuotaStateRepository;    // quota/account state, cooldowns
+  sessionAffinity: AffinityRepository; // sticky routing state
+  audit: AuditRepository;              // config audit log
+}
+interface DurableStateStore {
+  readonly driver: "sqlite" | "postgres" | "mysql";
+  withUnitOfWork<T>(fn: (uow: UnitOfWork) => Promise<T>): Promise<T>;
+  healthCheck(): Promise<{ ok: boolean; latencyMs: number }>;
+}
+```
+
+Deliberate exclusions from v1 (stay SQLite-local, matching #8075 §3): WAL/VACUUM/PRAGMA lifecycle, file backup/import/export, `sqlite-vec` vector storage (external backends delegate vectors to Qdrant), FTS (dialect-native fulltext later).
+
+### What makes each repository shippable
+
+The single-writer pattern OmniRoute already uses internally (targeted raw-SQL stat bumps like `touchConnectionLastUsed`, single-writer caches) maps cleanly onto `UnitOfWork`. Repositories are adopted **per domain**: a deployment can run `postgres` for connections+keys while hot-path ephemeral state stays local — because the port is injected at the repository boundary, not under 126 modules.
+
+## 3. Sequencing: PostgreSQL first, MySQL second
+
+1. **Postgres-first** (`pg` + `pg-pool`): richer native JSONB (several modules persist JSON blobs), `tsvector`, `INSERT ... ON CONFLICT` semantics closest to `INSERT OR REPLACE` usage.
+2. **MySQL 8+ after**, gated by a shared **dialect-conformance suite**: the same behavioral test pack runs against both implementations (schema bootstrap, UoW rollback, credential round-trip incl. encryption-at-rest fields, pagination/ordering contracts). A backend is "supported" only when the suite is green.
+
+Rationale: one conformance suite prevents the MySQL implementation from drifting into a shim that passes smoke tests but corrupts ordering/collation edge cases.
+
+## 4. Config surface
+
+```
+DATABASE_DRIVER=sqlite            # sqlite (default) | postgres | mysql
+DATABASE_URL=postgres://...       # required when driver != sqlite
+DATABASE_POOL_MAX=10
+DATABASE_TLS_REJECT_UNAUTHORIZED=1
+DATABASE_MIGRATIONS_TABLE=_omniroute_migrations
+```
+
+- Default stays `sqlite` — zero-config npm/Electron/Termux behavior is non-negotiable (#8075 §6).
+- Health check surfaces in the existing `/api/monitoring/health` payload next to `inflightRequests`.
+- Migrations: per-dialect lineage directories (`migrations-postgres/`, `migrations-mysql/`) with their own runner sharing the safety checks (mass-migration abort, pre-migration backup → pg_dump/mysqldump equivalent) already proven in `migrationRunner.ts`.
+
+## 5. Migration path (one-way export tool)
+
+`omniroute db export --to postgres --dsn ...`:
+1. Snapshot `storage.sqlite` (reuse `createPreMigrationBackup`).
+2. Replay schema from the target dialect's lineage to head.
+3. Copy tables row-by-row through the new repositories (not raw SQL translation), preserving IDs; re-encrypt nothing (credential ciphertexts are dialect-neutral strings today).
+4. Verify with row-count + checksum comparison per table; write a receipt file.
+5. Refuse to continue on any mismatch (fail-closed).
+
+## 6. Phased delivery plan
+
+| Phase | Deliverable | Risk gate |
+|---|---|---|
+| P1 | `ports/` contract + SQLite implementation of `ConnectionRepository` + conformance-suite skeleton running against SQLite itself | Suite green on the reference driver |
+| P2 | Postgres implementation of the same repo + suite green on PG | Suite green on PG |
+| P3 | Remaining domains (apiKeys, combos, quotaState, affinity, audit) on PG | Full-suite parity vs SQLite run |
+| P4 | MySQL via conformance suite | Suite green on MySQL |
+| P5 | Export tool + docs | Round-trip receipt on a 1GB-scale fixture |
+
+P1–P2 alone deliver value: a clustered deployment can share connection/key state while everything else stays local.
+
+## 7. Explicitly out of scope
+
+- Translating the remaining ~120 SQLite-shaped modules verbatim ("just make prepare() hit Postgres") — rejected: it fakes sync over async and preserves dialect bugs forever.
+- Replacing `sqlite-vec` with pgvector — #8075 names Qdrant; keeping vector search out-of-process also keeps the lean proxy core lean (4.0 modular direction).

docs/diagrams/cli-terminal.svg

@@ -1,4 +1,4 @@
-<svg viewBox="0 0 1200 350" xmlns="http://www.w3.org/2000/svg" role="img" aria-label="Animated terminal demoing the OmniRoute CLI: omniroute providers list (350 providers registered, anthropic, codex, glm, kimi shown active), omniroute combo list (always-on priority, cost-saver, fusion-panel, context-relay) and omniroute health (healthy, 18412 requests in 24h, p95 412ms, circuit breakers 24 closed, 1 half-open, 0 open), cycling over 85 top-level commands: providers, oauth, keys, combo, nodes, models, cache, compression, cost, usage, quota, health, resilience, telemetry, logs, audit, mcp, a2a, cloud, memory, skills, eval, doctor, repl, tunnel, backup, sync, webhooks, policy, pricing, translator, simulate and more.">
+<svg viewBox="0 0 1200 350" xmlns="http://www.w3.org/2000/svg" role="img" aria-label="Animated terminal demoing the OmniRoute CLI: omniroute providers list (351 providers registered, anthropic, codex, glm, kimi shown active), omniroute combo list (always-on priority, cost-saver, fusion-panel, context-relay) and omniroute health (healthy, 18412 requests in 24h, p95 412ms, circuit breakers 24 closed, 1 half-open, 0 open), cycling over 85 top-level commands: providers, oauth, keys, combo, nodes, models, cache, compression, cost, usage, quota, health, resilience, telemetry, logs, audit, mcp, a2a, cloud, memory, skills, eval, doctor, repl, tunnel, backup, sync, webhooks, policy, pricing, translator, simulate and more.">
 <desc>Compact animated terminal cycling three real OmniRoute CLI commands with a typewriter effect and a scrolling subcommand ticker; the first frame shows the completed providers-list screen.</desc>
 <defs><clipPath id="tickerClip"><rect x="12" y="304" width="1176" height="40"/></clipPath><clipPath id="tw0"><rect x="64" y="46" height="26" width="0"><animate attributeName="width" calcMode="discrete" values="0;31;61;92;122;153;184;214;245;245" keyTimes="0;0.012;0.018;0.024;0.030;0.036;0.042;0.048;0.054;1" dur="18s" repeatCount="indefinite"/></rect></clipPath><clipPath id="tw1"><rect x="64" y="46" height="26" width="0"><animate attributeName="width" calcMode="discrete" values="0;26;51;76;102;128;153;178;204;204" keyTimes="0;0.348;0.351;0.357;0.363;0.369;0.375;0.381;0.387;1" dur="18s" repeatCount="indefinite"/></rect></clipPath><clipPath id="tw2"><rect x="64" y="46" height="26" width="0"><animate attributeName="width" calcMode="discrete" values="0;20;41;61;82;102;122;143;163;163" keyTimes="0;0.678;0.684;0.690;0.696;0.702;0.708;0.714;0.720;1" dur="18s" repeatCount="indefinite"/></rect></clipPath></defs>
 <rect width="1200" height="350" fill="#0d1117"/>
@@ -6,7 +6,7 @@
 <path d="M 0 34 L 1200 34" stroke="#ffffff" stroke-opacity="0.08" stroke-width="1"/>
 <circle cx="24" cy="17" r="6" fill="#ff5f56"/><circle cx="46" cy="17" r="6" fill="#ffbd2e"/><circle cx="68" cy="17" r="6" fill="#27c93f"/>
 <text x="600" y="22" text-anchor="middle" font-family="Consolas, 'Courier New', monospace" font-size="13" fill="#71717a">omniroute &#8212; 85 top-level commands</text>
-<g font-family="Consolas, 'Courier New', monospace" font-size="17"><animate attributeName="opacity" values="1;0;0" keyTimes="0;0.006;1" dur="18s" repeatCount="indefinite"/><text x="64" y="66" fill="#F7F6FC">omniroute providers list</text><text x="40" y="100" font-weight="700" fill="#38bdf8">OmniRoute Providers</text><text x="40" y="128" fill="#a1a1aa">1f3a9c2e&#160;&#160;anthropic&#160;&#160;&#160;Claude Max 20x&#160;&#160;&#160;&#160;<tspan fill='#22c55e'>active</tspan></text><text x="40" y="154" fill="#a1a1aa">8c2d5b1a&#160;&#160;codex&#160;&#160;&#160;&#160;&#160;&#160;&#160;Codex Pro (team)&#160;&#160;<tspan fill='#22c55e'>active</tspan></text><text x="40" y="180" fill="#a1a1aa">f4e0a97b&#160;&#160;glm&#160;&#160;&#160;&#160;&#160;&#160;&#160;&#160;&#160;GLM Coding Plan&#160;&#160;&#160;<tspan fill='#22c55e'>active</tspan></text><text x="40" y="206" fill="#a1a1aa">03bd6e5f&#160;&#160;kimi&#160;&#160;&#160;&#160;&#160;&#160;&#160;&#160;Kimi K2 free&#160;&#160;&#160;&#160;&#160;&#160;<tspan fill='#22c55e'>active</tspan></text><text x="40" y="232" fill="#71717a">&#8230; 346 more providers</text></g><g opacity="1" font-family="Consolas, 'Courier New', monospace" font-size="17">
+<g font-family="Consolas, 'Courier New', monospace" font-size="17"><animate attributeName="opacity" values="1;0;0" keyTimes="0;0.006;1" dur="18s" repeatCount="indefinite"/><text x="64" y="66" fill="#F7F6FC">omniroute providers list</text><text x="40" y="100" font-weight="700" fill="#38bdf8">OmniRoute Providers</text><text x="40" y="128" fill="#a1a1aa">1f3a9c2e&#160;&#160;anthropic&#160;&#160;&#160;Claude Max 20x&#160;&#160;&#160;&#160;<tspan fill='#22c55e'>active</tspan></text><text x="40" y="154" fill="#a1a1aa">8c2d5b1a&#160;&#160;codex&#160;&#160;&#160;&#160;&#160;&#160;&#160;Codex Pro (team)&#160;&#160;<tspan fill='#22c55e'>active</tspan></text><text x="40" y="180" fill="#a1a1aa">f4e0a97b&#160;&#160;glm&#160;&#160;&#160;&#160;&#160;&#160;&#160;&#160;&#160;GLM Coding Plan&#160;&#160;&#160;<tspan fill='#22c55e'>active</tspan></text><text x="40" y="206" fill="#a1a1aa">03bd6e5f&#160;&#160;kimi&#160;&#160;&#160;&#160;&#160;&#160;&#160;&#160;Kimi K2 free&#160;&#160;&#160;&#160;&#160;&#160;<tspan fill='#22c55e'>active</tspan></text><text x="40" y="232" fill="#71717a">&#8230; 347 more providers</text></g><g opacity="1" font-family="Consolas, 'Courier New', monospace" font-size="17">
 <animate attributeName="opacity" values="1;1;0;0" keyTimes="0;0.315;0.33;1" dur="18s" repeatCount="indefinite"/>
 <text x="40" y="66" fill="#22c55e">$</text>
 <g clip-path="url(#tw0)"><text x="64" y="66" fill="#F7F6FC">omniroute providers list</text></g>
@@ -14,7 +14,7 @@
   <animate attributeName="x" calcMode="discrete" values="64;95;125;156;186;217;248;278;309;309" keyTimes="0.000;0.012;0.018;0.024;0.030;0.036;0.042;0.048;0.054;1" dur="18s" repeatCount="indefinite"/>
   <animate attributeName="opacity" values="0;0;1;0.2;1;0.2;1;0;0" keyTimes="0;0.011;0.012;0.022;0.032;0.042;0.052;0.074;1" dur="18s" repeatCount="indefinite"/>
 </rect>
-<text x="40" y="100" font-weight="700" fill="#38bdf8" opacity="0"><animate attributeName="opacity" calcMode="discrete" values="0;0;1" keyTimes="0;0.045;0.047" dur="18s" repeatCount="indefinite"/>OmniRoute Providers</text><text x="40" y="128" fill="#a1a1aa" opacity="0"><animate attributeName="opacity" calcMode="discrete" values="0;0;1" keyTimes="0;0.053;0.055" dur="18s" repeatCount="indefinite"/>1f3a9c2e&#160;&#160;anthropic&#160;&#160;&#160;Claude Max 20x&#160;&#160;&#160;&#160;<tspan fill='#22c55e'>active</tspan></text><text x="40" y="154" fill="#a1a1aa" opacity="0"><animate attributeName="opacity" calcMode="discrete" values="0;0;1" keyTimes="0;0.061;0.063" dur="18s" repeatCount="indefinite"/>8c2d5b1a&#160;&#160;codex&#160;&#160;&#160;&#160;&#160;&#160;&#160;Codex Pro (team)&#160;&#160;<tspan fill='#22c55e'>active</tspan></text><text x="40" y="180" fill="#a1a1aa" opacity="0"><animate attributeName="opacity" calcMode="discrete" values="0;0;1" keyTimes="0;0.069;0.07100000000000001" dur="18s" repeatCount="indefinite"/>f4e0a97b&#160;&#160;glm&#160;&#160;&#160;&#160;&#160;&#160;&#160;&#160;&#160;GLM Coding Plan&#160;&#160;&#160;<tspan fill='#22c55e'>active</tspan></text><text x="40" y="206" fill="#a1a1aa" opacity="0"><animate attributeName="opacity" calcMode="discrete" values="0;0;1" keyTimes="0;0.077;0.079" dur="18s" repeatCount="indefinite"/>03bd6e5f&#160;&#160;kimi&#160;&#160;&#160;&#160;&#160;&#160;&#160;&#160;Kimi K2 free&#160;&#160;&#160;&#160;&#160;&#160;<tspan fill='#22c55e'>active</tspan></text><text x="40" y="232" fill="#71717a" opacity="0"><animate attributeName="opacity" calcMode="discrete" values="0;0;1" keyTimes="0;0.085;0.08700000000000001" dur="18s" repeatCount="indefinite"/>&#8230; 346 more providers</text>
+<text x="40" y="100" font-weight="700" fill="#38bdf8" opacity="0"><animate attributeName="opacity" calcMode="discrete" values="0;0;1" keyTimes="0;0.045;0.047" dur="18s" repeatCount="indefinite"/>OmniRoute Providers</text><text x="40" y="128" fill="#a1a1aa" opacity="0"><animate attributeName="opacity" calcMode="discrete" values="0;0;1" keyTimes="0;0.053;0.055" dur="18s" repeatCount="indefinite"/>1f3a9c2e&#160;&#160;anthropic&#160;&#160;&#160;Claude Max 20x&#160;&#160;&#160;&#160;<tspan fill='#22c55e'>active</tspan></text><text x="40" y="154" fill="#a1a1aa" opacity="0"><animate attributeName="opacity" calcMode="discrete" values="0;0;1" keyTimes="0;0.061;0.063" dur="18s" repeatCount="indefinite"/>8c2d5b1a&#160;&#160;codex&#160;&#160;&#160;&#160;&#160;&#160;&#160;Codex Pro (team)&#160;&#160;<tspan fill='#22c55e'>active</tspan></text><text x="40" y="180" fill="#a1a1aa" opacity="0"><animate attributeName="opacity" calcMode="discrete" values="0;0;1" keyTimes="0;0.069;0.07100000000000001" dur="18s" repeatCount="indefinite"/>f4e0a97b&#160;&#160;glm&#160;&#160;&#160;&#160;&#160;&#160;&#160;&#160;&#160;GLM Coding Plan&#160;&#160;&#160;<tspan fill='#22c55e'>active</tspan></text><text x="40" y="206" fill="#a1a1aa" opacity="0"><animate attributeName="opacity" calcMode="discrete" values="0;0;1" keyTimes="0;0.077;0.079" dur="18s" repeatCount="indefinite"/>03bd6e5f&#160;&#160;kimi&#160;&#160;&#160;&#160;&#160;&#160;&#160;&#160;Kimi K2 free&#160;&#160;&#160;&#160;&#160;&#160;<tspan fill='#22c55e'>active</tspan></text><text x="40" y="232" fill="#71717a" opacity="0"><animate attributeName="opacity" calcMode="discrete" values="0;0;1" keyTimes="0;0.085;0.08700000000000001" dur="18s" repeatCount="indefinite"/>&#8230; 347 more providers</text>
 </g><g opacity="0" font-family="Consolas, 'Courier New', monospace" font-size="17">
 <animate attributeName="opacity" values="0;0;1;1;0;0" keyTimes="0;0.333;0.34800000000000003;0.648;0.663;1" dur="18s" repeatCount="indefinite"/>
 <text x="40" y="66" fill="#22c55e">$</text>

docs/diagrams/comparison-table.svg

@@ -1,4 +1,4 @@
-<svg viewBox="0 0 1200 780" xmlns="http://www.w3.org/2000/svg" role="img" aria-label="Comparison table: OmniRoute versus 9router, OpenRouter, CLIProxyAPI and LiteLLM across 13 capabilities. OmniRoute is the only one with the full set: 350 providers, 90+ free providers built-in, 19 routing strategies, 12-engine token compression, a built-in MCP server with 110 tools, A2A protocol, persistent memory, guardrails, cloud agents, TLS fingerprint stealth, desktop/Termux/PWA, 43 UI locales and 100% MIT self-hosted. 9router has free providers, RTK compression and translation but no MCP, A2A, memory, guardrails, cloud agents or stealth. OpenRouter is a hosted SaaS with 400+ models, guardrails and a hosted MCP but is not self-hosted and lacks A2A, memory, cloud agents and stealth. CLIProxyAPI is a light OAuth proxy with two routing strategies. LiteLLM has 100+ providers, A2A and extensive guardrails but no memory, compression, free tier, stealth or cloud agents.">
+<svg viewBox="0 0 1200 780" xmlns="http://www.w3.org/2000/svg" role="img" aria-label="Comparison table: OmniRoute versus 9router, OpenRouter, CLIProxyAPI and LiteLLM across 13 capabilities. OmniRoute is the only one with the full set: 351 providers, 90+ free providers built-in, 19 routing strategies, 12-engine token compression, a built-in MCP server with 110 tools, A2A protocol, persistent memory, guardrails, cloud agents, TLS fingerprint stealth, desktop/Termux/PWA, 43 UI locales and 100% MIT self-hosted. 9router has free providers, RTK compression and translation but no MCP, A2A, memory, guardrails, cloud agents or stealth. OpenRouter is a hosted SaaS with 400+ models, guardrails and a hosted MCP but is not self-hosted and lacks A2A, memory, cloud agents and stealth. CLIProxyAPI is a light OAuth proxy with two routing strategies. LiteLLM has 100+ providers, A2A and extensive guardrails but no memory, compression, free tier, stealth or cloud agents.">
   <desc>Static-header comparison table where each capability row fades in top to bottom; the OmniRoute column is highlighted and shows a check or a leading value in every row, while competitors show a mix of checks, partials and crosses.</desc>
   <defs>
     <pattern id="gC" width="32" height="32" patternUnits="userSpaceOnUse"><path d="M 32 0 L 0 0 0 32" fill="none" stroke="#ffffff" stroke-opacity="0.05" stroke-width="1"/></pattern>

docs/diagrams/promise-pillars.svg

@@ -1,4 +1,4 @@
-<svg viewBox="0 0 1200 540" xmlns="http://www.w3.org/2000/svg" role="img" aria-label="The OmniRoute promise: one endpoint and 350 providers. Six pillars. Resilient fallback: automatic routing continues while another healthy target is available. Save up to 95 percent of eligible tokens: RTK plus Caveman stacked compression averages about 89 percent on tool-heavy sessions. Zero dollars to start: 90+ providers with a free tier and 56 recurring or keyless free-forever providers. Every tool works: 35 CLI and agent integration records, including Claude Code, Codex, Cursor, Cline, Copilot and Antigravity, through one config. One endpoint: OpenAI, Claude, Gemini and Responses API translation at /v1. Production controls: circuit breakers, TLS stealth, MCP with 110 tools, A2A, memory, guardrails, evals, and 39,000+ static test declarations across 5,100+ tracked test files.">
+<svg viewBox="0 0 1200 540" xmlns="http://www.w3.org/2000/svg" role="img" aria-label="The OmniRoute promise: one endpoint and 351 providers. Six pillars. Resilient fallback: automatic routing continues while another healthy target is available. Save up to 95 percent of eligible tokens: RTK plus Caveman stacked compression averages about 89 percent on tool-heavy sessions. Zero dollars to start: 90+ providers with a free tier and 56 recurring or keyless free-forever providers. Every tool works: 35 CLI and agent integration records, including Claude Code, Codex, Cursor, Cline, Copilot and Antigravity, through one config. One endpoint: OpenAI, Claude, Gemini and Responses API translation at /v1. Production controls: circuit breakers, TLS stealth, MCP with 110 tools, A2A, memory, guardrails, evals, and 39,000+ static test declarations across 5,100+ tracked test files.">
   <desc>Animated promise card: six pillar tiles fade in in reading order, then a soft colored border highlight sweeps from tile to tile in a continuous cycle.</desc>
   <defs>
     <pattern id="gridPaperP" width="32" height="32" patternUnits="userSpaceOnUse">
@@ -21,7 +21,7 @@
     <line x1="150" y1="53" x2="1160" y2="53" stroke="#232b38" stroke-width="1.5"/>
   </g>
   <g>
-    <text x="40" y="100" font-family="Inter, 'Segoe UI', Arial, Helvetica, system-ui, sans-serif" font-size="23" font-weight="600" fill="#c9d1d9">One endpoint. <tspan fill="#a78bfa" font-weight="800">350 providers.</tspan> Never stop building — OmniRoute picks <tspan fill="#7ee787" font-weight="700">the cheapest one that works</tspan>.</text>
+    <text x="40" y="100" font-family="Inter, 'Segoe UI', Arial, Helvetica, system-ui, sans-serif" font-size="23" font-weight="600" fill="#c9d1d9">One endpoint. <tspan fill="#a78bfa" font-weight="800">351 providers.</tspan> Never stop building — OmniRoute picks <tspan fill="#7ee787" font-weight="700">the cheapest one that works</tspan>.</text>
   </g>
 
   <g font-family="Inter, 'Segoe UI', Arial, Helvetica, system-ui, sans-serif">
@@ -38,7 +38,7 @@
         <line x1="3.9" y1="3.9" x2="18.1" y2="18.1"/>
       </g>
       <text x="102" y="170" font-size="18" font-weight="800" fill="#74b9ff">Never hit limits</text>
-      <text x="66" y="204" font-size="13.5" fill="#a1a1aa">Auto-fallback across 350 providers in</text>
+      <text x="66" y="204" font-size="13.5" fill="#a1a1aa">Auto-fallback across 351 providers in</text>
       <text x="66" y="226" font-size="13.5" fill="#a1a1aa">milliseconds. Quota out? The next provider</text>
       <text x="66" y="248" font-size="13.5" fill="#a1a1aa">takes over while a healthy target remains.</text>
     </g>

docs/diagrams/readme-hero.svg

@@ -1,4 +1,4 @@
-<svg viewBox="0 0 1200 548" xmlns="http://www.w3.org/2000/svg" role="img" aria-label="OmniRoute hero: Never stop coding. Every AI tool to 350 providers — 90+ free — through one endpoint. Claude Code, Codex, Cursor, Cline, Copilot and Antigravity into free Claude, GPT and Gemini with auto-fallback. RTK + Caveman stacked compression saves 15 to 95 percent of tokens — about 89 percent average on tool-heavy sessions — so you never hit limits. Stats: 350 AI providers, 90+ free tiers, about 1.51B free tokens per month, 15 to 95 percent token savings, 19 routing strategies, zero dollars to start.">
+<svg viewBox="0 0 1200 548" xmlns="http://www.w3.org/2000/svg" role="img" aria-label="OmniRoute hero: Never stop coding. Every AI tool to 351 providers — 90+ free — through one endpoint. Claude Code, Codex, Cursor, Cline, Copilot and Antigravity into free Claude, GPT and Gemini with auto-fallback. RTK + Caveman stacked compression saves 15 to 95 percent of tokens — about 89 percent average on tool-heavy sessions — so you never hit limits. Stats: 351 AI providers, 90+ free tiers, about 1.51B free tokens per month, 15 to 95 percent token savings, 19 routing strategies, zero dollars to start.">
   <desc>Animated hero card: a pulse travels the divider line and a compression bar demo repeatedly shrinks a prompt by up to 95 percent; all headline content is static and readable on the first frame.</desc>
   <defs>
     <pattern id="gridPaperH" width="32" height="32" patternUnits="userSpaceOnUse">
@@ -28,7 +28,7 @@
   <text x="48" y="138" font-family="Inter, 'Segoe UI', Arial, Helvetica, system-ui, sans-serif" font-size="60" font-weight="800" fill="#e9edf3">Never stop coding<tspan fill="#a855f7">.</tspan></text>
 
   <!-- subheadline -->
-  <text x="48" y="184" font-family="Inter, 'Segoe UI', Arial, Helvetica, system-ui, sans-serif" font-size="25" font-weight="600" fill="#c9d1d9">Every AI tool → <tspan fill="#a78bfa" font-weight="800">350 providers</tspan> — <tspan fill="#7ee787" font-weight="800">90+ free</tspan> — through one endpoint.</text>
+  <text x="48" y="184" font-family="Inter, 'Segoe UI', Arial, Helvetica, system-ui, sans-serif" font-size="25" font-weight="600" fill="#c9d1d9">Every AI tool → <tspan fill="#a78bfa" font-weight="800">351 providers</tspan> — <tspan fill="#7ee787" font-weight="800">90+ free</tspan> — through one endpoint.</text>
 
   <!-- plug line -->
   <text x="48" y="222" font-family="Inter, 'Segoe UI', Arial, Helvetica, system-ui, sans-serif" font-size="16.5" fill="#a1a1aa">Claude Code · Codex · Cursor · Cline · Copilot · Antigravity&#160;&#160;→&#160;&#160;<tspan fill="#7ee787" font-weight="700">FREE</tspan> Claude / GPT / Gemini · auto-fallback</text>

docs/frameworks/MCP-SERVER.md

@@ -66,22 +66,22 @@ Cursor, Cline, and compatible MCP client setup.
 
 ## Essential Tools (13) — Phase 1
 
-| Tool                            | Scopes                | Description                                                   |
-| :------------------------------ | :-------------------- | :------------------------------------------------------------ |
-| `omniroute_get_health`          | `read:health`         | Uptime, memory, circuit breakers, rate limits, cache stats    |
-| `omniroute_list_combos`         | `read:combos`         | All configured combos with strategies (optional metrics)      |
-| `omniroute_get_combo_metrics`   | `read:combos`         | Performance metrics for a specific combo                      |
-| `omniroute_switch_combo`        | `write:combos`        | Activate or deactivate a combo                                |
-| `omniroute_create_combo`        | `write:combos`        | Create a validated combo through the existing combo API       |
-| `omniroute_check_quota`         | `read:quota`          | Quota used/total, percent remaining, reset time, token health |
-| `omniroute_route_request`       | `execute:completions` | Send a chat completion through OmniRoute routing              |
-| `omniroute_cost_report`         | `read:usage`          | Cost report by period (session/day/week/month)                |
-| `omniroute_list_models_catalog` | `read:models`         | Full model catalog with capabilities, status, pricing         |
-| `omniroute_radar_catalog`       | `read:radar`          | Local signed Radar catalog; optional provider/family filters  |
-| `omniroute_tool_search`         | `read:tools`          | Discover tools from the registered MCP catalog                |
-| `omniroute_web_search`          | `execute:search`      | Web search through the configured search providers. Not X/Twitter. |
-| `omniroute_x_search`            | `execute:search`      | Search X (Twitter) through SuperGrok / xAI server-side `x_search`. Requires `xai-oauth` or an xAI API key. Not the X Developer Platform MCP. |
-| `omniroute_web_fetch`           | `execute:search`      | Fetch web content through the configured fetch providers      |
+| Tool                            | Scopes                | Description                                                                                                                    |
+| :------------------------------ | :-------------------- | :----------------------------------------------------------------------------------------------------------------------------- |
+| `omniroute_get_health`          | `read:health`         | Uptime, memory, circuit breakers, rate limits, cache stats                                                                     |
+| `omniroute_list_combos`         | `read:combos`         | All configured combos with strategies (optional metrics)                                                                       |
+| `omniroute_get_combo_metrics`   | `read:combos`         | Performance metrics for a specific combo                                                                                       |
+| `omniroute_switch_combo`        | `write:combos`        | Activate or deactivate a combo                                                                                                 |
+| `omniroute_create_combo`        | `write:combos`        | Create a validated combo through the existing combo API                                                                        |
+| `omniroute_check_quota`         | `read:quota`          | Quota used/total, percent remaining, reset time, token health                                                                  |
+| `omniroute_route_request`       | `execute:completions` | Send a chat completion through OmniRoute routing                                                                               |
+| `omniroute_cost_report`         | `read:usage`          | Cost report by period (session/day/week/month)                                                                                 |
+| `omniroute_list_models_catalog` | `read:models`         | Full model catalog with capabilities, status, pricing                                                                          |
+| `omniroute_radar_catalog`       | `read:radar`          | Local signed Radar catalog; optional provider/family filters                                                                   |
+| `omniroute_tool_search`         | `read:tools`          | Discover tools from the registered MCP catalog                                                                                 |
+| `omniroute_web_search`          | `execute:search`      | Web search through the configured search providers. Not X/Twitter.                                                             |
+| `omniroute_x_search`            | `execute:search`      | Search X through xAI/SuperGrok, or choose `xquik-search` for Xquik API results. Requires credentials for the selected backend. |
+| `omniroute_web_fetch`           | `execute:search`      | Fetch web content through the configured fetch providers                                                                       |
 
 ## Advanced Tools (11) — Phase 2
 

docs/i18n/ar/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/az/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/bg/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/bn/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/cs/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/da/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/de/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/es/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/fa/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/fi/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/fr/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/gu/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/he/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/hi/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/hu/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/id/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/in/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/it/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/ja/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/ko/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/mr/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/ms/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/nl/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/no/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/phi/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/pl/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/pt-BR/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/pt/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/ro/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/ru/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/sk/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/sv/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/sw/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/ta/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/te/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/th/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/tr/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/uk-UA/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/ur/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/vi/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/zh-CN/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/zh-TW/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/openapi.yaml

@@ -43,6 +43,8 @@ tags:
     description: Content moderation
   - name: Rerank
     description: Document reranking
+  - name: Search
+    description: Unified web, news, and X search
   - name: Models
     description: Available model listing
   - name: Providers
@@ -1153,6 +1155,199 @@ paths:
           $ref: "#/components/responses/Unauthorized"
   # ─── Proxy Endpoints ──────────────────────────────────────────
 
+  /api/v1/search:
+    get:
+      tags: [Search]
+      summary: List search providers
+      description: Lists configured search providers and their supported search types.
+      responses:
+        "200":
+          description: Search provider catalog
+          content:
+            application/json:
+              schema:
+                type: object
+                required: [object, data]
+                properties:
+                  object:
+                    type: string
+                    const: list
+                  data:
+                    type: array
+                    items:
+                      type: object
+                      required: [id, object, created, name, search_types]
+                      properties:
+                        id:
+                          type: string
+                        object:
+                          type: string
+                          const: search_provider
+                        created:
+                          type: integer
+                        name:
+                          type: string
+                        search_types:
+                          type: array
+                          items:
+                            type: string
+                            enum: [web, news, x]
+    post:
+      tags: [Search]
+      summary: Run a unified search
+      description: >-
+        Searches the web, news, or X through a configured provider. Set `provider`
+        to `xquik-search` to use Xquik for X search. The aliases `xquik` and
+        `xquik_search` resolve to the same provider.
+      security:
+        - BearerAuth: []
+      requestBody:
+        required: true
+        content:
+          application/json:
+            schema:
+              type: object
+              required: [query]
+              properties:
+                query:
+                  type: string
+                  minLength: 1
+                  maxLength: 500
+                provider:
+                  type: string
+                  minLength: 1
+                  description: A search provider id or registered alias.
+                  examples: [xquik-search]
+                max_results:
+                  type: integer
+                  minimum: 1
+                  maximum: 100
+                  default: 5
+                search_type:
+                  type: string
+                  enum: [web, news, x]
+                  default: web
+                offset:
+                  type: integer
+                  minimum: 0
+                  default: 0
+                country:
+                  type: string
+                  maxLength: 2
+                language:
+                  type: string
+                  minLength: 2
+                  maxLength: 5
+                time_range:
+                  type: string
+                  enum: [any, hour, day, week, month, year]
+                content:
+                  type: object
+                  properties:
+                    snippet: { type: boolean, default: true }
+                    full_page: { type: boolean, default: false }
+                    format: { type: string, enum: [text, markdown], default: text }
+                    max_characters: { type: integer, minimum: 100, maximum: 100000 }
+                filters:
+                  type: object
+                  properties:
+                    include_domains:
+                      type: array
+                      maxItems: 20
+                      items: { type: string, maxLength: 253 }
+                    exclude_domains:
+                      type: array
+                      maxItems: 20
+                      items: { type: string, maxLength: 253 }
+                    safe_search: { type: string, enum: [off, moderate, strict] }
+                provider_options:
+                  type: object
+                  additionalProperties: true
+                strict_filters:
+                  type: boolean
+                  default: false
+              additionalProperties: true
+      responses:
+        "200":
+          description: Normalized search results
+          content:
+            application/json:
+              schema:
+                type: object
+                required: [id, provider, query, results, answer, usage, metrics, errors, cached]
+                properties:
+                  id:
+                    type: string
+                    pattern: ^search-
+                  provider:
+                    type: string
+                  query:
+                    type: string
+                  cached:
+                    type: boolean
+                  results:
+                    type: array
+                    items:
+                      type: object
+                      required: [title, url, snippet, position, citation]
+                      properties:
+                        title: { type: string }
+                        url: { type: string, format: uri }
+                        display_url: { type: string }
+                        snippet: { type: string }
+                        position: { type: integer, minimum: 1 }
+                        score:
+                          type: [number, "null"]
+                          minimum: 0
+                          maximum: 1
+                        published_at: { type: [string, "null"] }
+                        favicon_url: { type: [string, "null"], format: uri }
+                        citation:
+                          type: object
+                          required: [provider, retrieved_at, rank]
+                          properties:
+                            provider: { type: string }
+                            retrieved_at: { type: string, format: date-time }
+                            rank: { type: integer, minimum: 1 }
+                  answer:
+                    type: [object, "null"]
+                  usage:
+                    type: object
+                    required: [queries_used, search_cost_usd]
+                    properties:
+                      queries_used: { type: integer, minimum: 0 }
+                      search_cost_usd: { type: number, minimum: 0 }
+                      llm_tokens: { type: integer, minimum: 0 }
+                  metrics:
+                    type: object
+                    required: [response_time_ms, upstream_latency_ms, total_results_available]
+                    properties:
+                      response_time_ms: { type: number, minimum: 0 }
+                      upstream_latency_ms: { type: number, minimum: 0 }
+                      gateway_latency_ms: { type: number, minimum: 0 }
+                      total_results_available: { type: [integer, "null"], minimum: 0 }
+                  errors:
+                    type: array
+                    items:
+                      type: object
+                      required: [provider, code, message]
+                      properties:
+                        provider: { type: string }
+                        code: { type: string }
+                        message: { type: string }
+        "400":
+          description: Invalid request, provider, credentials, or search type
+        "401":
+          $ref: "#/components/responses/Unauthorized"
+        "403":
+          description: Search provider blocked by API key or security policy
+        "429":
+          description: Every eligible provider credential is rate limited
+        "500":
+          $ref: "#/components/responses/InternalError"
+        "502":
+          description: Search provider failed
+
   /api/v1/chat/completions:
     post:
       tags: [Chat]

docs/plans/7778-hierarchical-admission.md

@@ -0,0 +1,23 @@
+# #7778 hierarchical admission cleanup plan
+
+1. Lock the existing single-key semaphore contract and the new atomic multi-key
+   contract with focused tests: no partial reservations, FIFO queueing, abort,
+   timeout, queue-full, idempotent release, stats, and cleanup.
+2. Generalize the existing account semaphore in place. Keep `acquire()` as a
+   compatibility wrapper around `acquireMany()`; do not add a second scheduler
+   or a dependency.
+3. Replace the account-only acquisition in `chatCore` with one cumulative
+   global/provider/account acquisition immediately before `withRateLimit`.
+   Reacquire the whole set whenever account rotation changes the connection,
+   and retain the release through streaming completion.
+4. Extend the existing resilience settings pipeline (types, defaults,
+   normalization, schema, API response, UI, and translations) with the global
+   and provider caps. Relabel the old Bottleneck concurrency control as
+   connection/quota-scope concurrency so its real scope is explicit.
+5. Run focused tests, lint, typecheck, static checks, and the full test suite;
+   document the behavioral change in the changelog.
+
+Behavior intentionally preserved: zero/null concurrency bypasses a gate,
+account-only callers keep using `acquire()`, blocked-account controls retain
+their key format and API, and provider rate-limit queue behavior remains
+unchanged.

docs/reference/PROVIDER_REFERENCE.md

@@ -1,16 +1,16 @@
 ---
 title: "Provider Reference"
 version: 3.8.50
-lastUpdated: 2026-08-23
+lastUpdated: 2026-08-24
 ---
 
 # Provider Reference
 
 > **Auto-generated** from `src/shared/constants/providers.ts` — do not edit by hand.
 > Regenerate with: `npm run gen:provider-reference`
-> **Last generated:** 2026-08-23
+> **Last generated:** 2026-08-24
 
-Total providers: **350**. See category breakdown below.
+Total providers: **351**. See category breakdown below.
 
 ## Categories
 
@@ -377,7 +377,7 @@ Use the dashboard at `/dashboard/providers` to enable, configure, and test each
 | `vllm` | `vllm` | vLLM | Local, self-hosted | [link](https://github.com/vllm-project/vllm) | API key optional. Configure the local vLLM OpenAI-compatible base URL (default: http://localhost:8000/v1). |
 | `xinference` | `xinference` | XInference | Local, self-hosted | [link](https://inference.readthedocs.io) | API key optional. Configure the local XInference OpenAI-compatible base URL (default: http://localhost:9997/v1). |
 
-## Search Providers (14)
+## Search Providers (15)
 
 | ID | Alias | Name | Tags | Website | Notes |
 |----|-------|------|------|---------|-------|
@@ -394,6 +394,7 @@ Use the dashboard at `/dashboard/providers` to enable, configure, and test each
 | `serper-search` | `serper-search` | Serper Search | Search | [link](https://serper.dev) | API key from serper.dev dashboard |
 | `tavily-search` | `tavily-search` | Tavily Search | Search | [link](https://tavily.com) | API key from app.tavily.com (format: tvly-...) |
 | `x-search` | `x_search` | X Search (Grok) | Search | [link](https://docs.x.ai/developers/tools/x-search) | SuperGrok OAuth (xai-oauth) or xAI API key. This is Grok X Search, not the X Developer MCP. |
+| `xquik-search` | `xquik` | Xquik X Search | Search | [link](https://docs.xquik.com) | Xquik API key (xq_...). Search is metered per returned post; the catalog estimate uses 5 results. |
 | `youcom-search` | `youcom-search` | You.com Search | Search | [link](https://you.com/business/api/) | X-API-Key from the You.com platform dashboard |
 
 ## Audio-only Providers (12)
@@ -438,7 +439,7 @@ Use the dashboard at `/dashboard/providers` to enable, configure, and test each
 
 - Catalog: [`src/shared/constants/providers.ts`](../../src/shared/constants/providers.ts)
 - Registry (per-model details): [`open-sse/config/providerRegistry.ts`](../../open-sse/config/providerRegistry.ts)
-- Executors: [`open-sse/executors/`](../../open-sse/executors/) (108 implementations)
+- Executors: [`open-sse/executors/`](../../open-sse/executors/) (109 implementations)
 - Translators: [`open-sse/translator/`](../../open-sse/translator/)
 
 ## See Also

electron/package.json

@@ -45,6 +45,7 @@
     "copyright": "Copyright © 2025 OmniRoute",
     "buildDependenciesFromSource": false,
     "npmRebuild": false,
+    "afterPack": "../scripts/build/afterpack-copy-node-modules.mjs",
     "directories": {
       "output": "dist-electron",
       "buildResources": "assets"
@@ -58,6 +59,7 @@
       "main.js",
       "preload.js",
       "loginManager.js",
+      "lib/loginHeaderCapture.js",
       "processTree.js",
       "sqlite-inspection.js",
       "remoteServerPromptPreload.js",

eslint.config.mjs

@@ -1,4 +1,6 @@
+import { fixupConfigRules } from "@eslint/compat";
 import nextVitals from "eslint-config-next/core-web-vitals";
+import * as espree from "espree";
 import tseslint from "typescript-eslint";
 
 // #7879: bar NEW local `toNumber` definitions outside the canonical helper.
@@ -37,7 +39,22 @@ const IMPORT_BOUNDARY_RESTRICTIONS = {
 
 /** @type {import("eslint").Linter.Config[]} */
 const eslintConfig = [
-  ...nextVitals,
+  ...fixupConfigRules(nextVitals),
+  // eslint-config-next's Babel parser has not adopted ESLint 10's ScopeManager
+  // finalize contract yet. Plain JS/JSX does not need that parser, so use the
+  // ESLint-native parser while retaining Next's plugins and rules.
+  {
+    files: ["**/*.{js,jsx,mjs,cjs}"],
+    languageOptions: {
+      parser: espree,
+    },
+  },
+  {
+    files: ["**/*.{mts,cts}"],
+    languageOptions: {
+      parser: tseslint.parser,
+    },
+  },
   // Pacote 4 (plano mestre testes+CI, 2026-07-04) — zero-warning policy: TODA regra roda
   // como "error" e a dívida pré-existente vive congelada por arquivo+regra em
   // config/quality/eslint-suppressions.json (ESLint bulk suppressions nativo). Violação
@@ -133,28 +150,45 @@ const eslintConfig = [
       "no-restricted-syntax": "off",
     },
   },
-  // Relaxed rules for open-sse and tests (incremental adoption)
+  // Relaxed rules for TypeScript in open-sse and tests (incremental adoption).
+  // eslint-config-next already registers @typescript-eslint for every TS file;
+  // registering a second plugin object is rejected by ESLint 10.
   {
-    files: ["open-sse/**/*.ts", "tests/**/*.mjs", "tests/**/*.ts"],
-    plugins: {
-      "@typescript-eslint": tseslint.plugin,
-    },
+    files: ["open-sse/**/*.ts", "tests/**/*.ts"],
     rules: {
       "@typescript-eslint/no-explicit-any": "error",
       "@next/next/no-assign-module-variable": "off",
       "react-hooks/rules-of-hooks": "off",
     },
   },
+  {
+    files: ["tests/**/*.mjs"],
+    plugins: {
+      "@typescript-eslint": tseslint.plugin,
+    },
+    rules: {
+      "@next/next/no-assign-module-variable": "off",
+      "react-hooks/rules-of-hooks": "off",
+    },
+  },
+  // JS/JSX files do not match eslint-config-next's TypeScript block. Register
+  // the plugin only for that disjoint scope so the shared unused-vars ratchet
+  // works without redefining the plugin for TS/TSX under ESLint 10.
+  {
+    files: ["src/**/*.{js,jsx}"],
+    plugins: {
+      "@typescript-eslint": tseslint.plugin,
+    },
+  },
   // Ratchet: bar NEW unused vars/args/catches outside the `_` escape hatch.
   // Pre-existing violations are frozen via config/quality/eslint-suppressions.json
   // (same pattern as #7879 toNumber); only genuinely NEW unused bindings fail
   // lint. `args: "all"` (not `after-used`) so a leading unused param is never
   // silently skipped, e.g. `function handle(req, _opts, next)` must flag `req`.
+  // eslint-config-next already registers @typescript-eslint; registering it again
+  // in this block is rejected by ESLint 10.
   {
     files: ["src/**/*.{ts,tsx,js,jsx}", "open-sse/**/*.ts", "tests/**/*.{ts,tsx,mjs}"],
-    plugins: {
-      "@typescript-eslint": tseslint.plugin,
-    },
     rules: {
       "@typescript-eslint/no-unused-vars": [
         "error",

llm.txt

@@ -1,6 +1,6 @@
 # OmniRoute
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 350 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -277,7 +277,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **350 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -475,7 +475,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **350-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

open-sse/config/embeddingRegistry.ts

@@ -408,6 +408,7 @@ export const EMBEDDING_PROVIDERS: Record<string, EmbeddingProvider> = {
       },
     ],
   },
+
 };
 
 const EMBEDDING_PROVIDER_ALIASES: Record<string, string> = {
@@ -470,6 +471,38 @@ export function getEmbeddingProvider(providerId: string): EmbeddingProvider | nu
   return EMBEDDING_PROVIDERS[resolveEmbeddingProviderId(providerId)] || null;
 }
 
+/**
+ * Derive an OpenAI-compatible embeddings config for a chat provider that has NO
+ * curated EMBEDDING_PROVIDERS entry. Works for any registry provider whose base
+ * URL ends in /chat/completions by swapping that suffix for /embeddings (groq,
+ * mistral, together, upstage, fireworks, nvidia, vercel-ai-gateway, ...).
+ * Dynamic-URL providers (no usable static base) derive to
+ * null — they need bespoke URL handling, not a bogus endpoint.
+ *
+ * This is a FALLBACK only: callers must check getEmbeddingProvider() first so
+ * curated entries keep their specialized configuration.
+ */
+export function deriveEmbeddingProviderForChatProvider(
+  providerId: string,
+  chatEntry: { id?: string; baseUrl?: string | string[] } | null | undefined
+): EmbeddingProvider | null {
+  if (!chatEntry) return null;
+  const rawBase = Array.isArray(chatEntry.baseUrl)
+    ? chatEntry.baseUrl[0]
+    : chatEntry.baseUrl;
+  if (!rawBase || typeof rawBase !== "string") return null;
+  // stripTrailingSlashes-equivalent without importing open-sse utils here:
+  const base = rawBase.replace(/\/+$/, "");
+  if (!base.endsWith("/chat/completions")) return null;
+  return {
+    id: providerId,
+    baseUrl: `${base.slice(0, -"/chat/completions".length)}/embeddings`,
+    authType: "apikey",
+    authHeader: "bearer",
+    models: [],
+  };
+}
+
 /**
  * Parse embedding model string (format: "provider/model" or just "model")
  * Returns { provider, model }
@@ -485,6 +518,18 @@ export function parseEmbeddingModel(
   const slashIdx = modelStr.indexOf("/");
   if (slashIdx > 0) {
     const rawProvider = modelStr.slice(0, slashIdx);
+
+    // A configured provider_node whose prefix exactly equals the requested
+    // provider segment always wins — even when that segment is also an alias
+    // of a curated provider (a local node must not be hijacked by a registry
+    // alias). Same exact-match precedence documented for
+    // EMBEDDING_MODEL_ALIASES above.
+    const dynamicExact =
+      dynamicProviders && dynamicProviders.find((dp) => dp.id === rawProvider);
+    if (dynamicExact) {
+      return { provider: rawProvider, model: modelStr.slice(slashIdx + 1) };
+    }
+
     const resolvedProvider = resolveEmbeddingProviderId(rawProvider);
 
     if (EMBEDDING_PROVIDERS[resolvedProvider]) {

open-sse/config/freeModelCatalog.ts

@@ -69,6 +69,31 @@ export interface FreeModelTotals {
 
 const RECURRING = new Set<FreeModelFreeType>(["recurring-daily", "recurring-monthly", "keyless"]);
 
+/**
+ * What each free-tier regime engages for "can I route here without paying?".
+ * Exhaustive by construction: adding a member to `FreeModelFreeType` will not
+ * compile until it is classified here. `discontinued` is the one regime a
+ * provider uses to retire a free tier behind a paid key — it does NOT grant
+ * free access, and the shared predicate (`isFreeModel`) must read this instead
+ * of treating every catalogued id as free. `RECURRING` (above) answers a
+ * different question (which regimes feed the headline token totals) and is left
+ * independent on purpose — deriving it from this table would silently change
+ * the homepage totals.
+ */
+const FREE_REGIME_TRAITS = {
+  "recurring-daily": { grantsFreeAccess: true },
+  "recurring-monthly": { grantsFreeAccess: true },
+  "recurring-credit": { grantsFreeAccess: true },
+  "recurring-uncapped": { grantsFreeAccess: true },
+  "one-time-initial": { grantsFreeAccess: true },
+  keyless: { grantsFreeAccess: true },
+  discontinued: { grantsFreeAccess: false },
+} satisfies Record<FreeModelFreeType, { grantsFreeAccess: boolean }>;
+
+export function grantsFreeAccess(freeType: FreeModelFreeType): boolean {
+  return FREE_REGIME_TRAITS[freeType].grantsFreeAccess;
+}
+
 /**
  * Deposit-unlock boosts: a one-time small top-up that permanently raises a
  * provider's recurring free quota. Kept OUT of the steady headline and surfaced

open-sse/config/opencodeZenGoSharedModels.ts

@@ -12,5 +12,15 @@
 export const OPENCODE_ZEN_GO_SHARED_MODELS = Object.freeze([
   { id: "kimi-k2.7-code", name: "Kimi K2.7 Code" },
   { id: "qwen3.5-plus", name: "Qwen3.5 Plus", targetFormat: "claude", supportsVision: false },
-  { id: "qwen3.6-plus", name: "Qwen3.6 Plus", targetFormat: "claude", supportsVision: false },
+  {
+    id: "qwen3.6-plus",
+    name: "Qwen3.6 Plus",
+    targetFormat: "claude",
+    supportsVision: false,
+    // #10788: effort-tier aliases exist as explicit registry rows; declare the
+    // vocabulary on the shared base row so variant synthesis and the sanitizer
+    // agree on it for both opencode-go and opencode-zen.
+    supportsReasoning: true,
+    supportedThinkingEfforts: ["high", "max"],
+  },
 ]);

open-sse/config/providers/registry/nvidia/index.ts

@@ -18,7 +18,17 @@ export const nvidiaProvider: RegistryEntry = {
   passthroughModels: true,
   models: [
     // #6108: z-ai/glm-5.1 EOL'd 2026-07-02 (direct probe returns 410) — dropped.
-    { id: "z-ai/glm-5.2", name: "GLM 5.2" },
+    // #10788: NVIDIA's hosted GLM-5.2 exposes a BINARY thinking switch
+    // (chat_template_kwargs.enable_thinking), not effort tiers — see
+    // mapNvidiaGlm52ReasoningParams. Declaring an empty tier list keeps the
+    // catalog from synthesizing unresolvable -low/-high/-max variant ids while
+    // still marking the model reasoning-capable.
+    {
+      id: "z-ai/glm-5.2",
+      name: "GLM 5.2",
+      supportsReasoning: true,
+      supportedThinkingEfforts: [],
+    },
     // #3329/#6108: minimaxai/minimax-m3 stays excluded from the nvidia tier — it
     // still 404s here for most callers; the single 200 probe in #6108 was not
     // reproducible enough to override the #3329 guard. Re-add only once NVIDIA

open-sse/config/providers/registry/opencode/go/index.ts

@@ -19,9 +19,15 @@ export const opencode_goProvider: RegistryEntry = {
     // `kimi-k2.7-code` (the live API rejects the plain `kimi-k2.7` alias for
     // `/chat/completions`, even though the docs config example uses it).
     // GLM-5.2 — base model + effort-tier aliases (#6922).
-    // OpencodeExecutor rewrites the alias to the canonical id and injects
-    // reasoning_effort, mirroring the deepseek-v4-pro-* pattern.
-    { id: "glm-5.2", name: "GLM-5.2", supportsReasoning: true },
+    // #10788: the tier vocabulary is declared on the base row so the catalog's
+    // variant synthesis (#9485) and the effort sanitizer share one source of
+    // truth with OpencodeExecutor's EFFORT_TIERS.
+    {
+      id: "glm-5.2",
+      name: "GLM-5.2",
+      supportsReasoning: true,
+      supportedThinkingEfforts: ["high", "max"],
+    },
     { id: "glm-5.2-high", name: "GLM-5.2 (high effort)", supportsReasoning: true },
     { id: "glm-5.2-max", name: "GLM-5.2 (max effort)", supportsReasoning: true },
 
@@ -34,11 +40,16 @@ export const opencode_goProvider: RegistryEntry = {
     { id: "kimi-k2.6", name: "Kimi K2.6" },
     { id: "kimi-k2.5", name: "Kimi K2.5" },
     // #8353: Kimi K3 base + max-effort alias from the OpenCode Go registry.
-    { id: "kimi-k3", name: "Kimi K3", supportsReasoning: true },
+    { id: "kimi-k3", name: "Kimi K3", supportsReasoning: true, supportedThinkingEfforts: ["max"] },
     { id: "kimi-k3-max", name: "Kimi K3 (max effort)", supportsReasoning: true },
     // MiMo-V2.5 — base model + effort-tier aliases (#6922).
     { id: "mimo-v2.5-pro", name: "MiMo-V2.5-Pro", supportsReasoning: true },
-    { id: "mimo-v2.5", name: "MiMo-V2.5", supportsReasoning: true },
+    {
+      id: "mimo-v2.5",
+      name: "MiMo-V2.5",
+      supportsReasoning: true,
+      supportedThinkingEfforts: ["high", "max"],
+    },
     { id: "mimo-v2.5-high", name: "MiMo-V2.5 (high effort)", supportsReasoning: true },
     { id: "mimo-v2.5-max", name: "MiMo-V2.5 (max effort)", supportsReasoning: true },
     // #3110: MiniMax M3 via OpenCode Go tier
@@ -59,7 +70,14 @@ export const opencode_goProvider: RegistryEntry = {
     // so combo routing skips them when the request contains image blocks,
     // preventing image content from reaching a vision-incapable upstream.
     // #8353: effort-tier aliases from the OpenCode Go registry.
-    { id: "qwen3.7-max", name: "Qwen3.7 Max", targetFormat: "claude", supportsVision: false },
+    {
+      id: "qwen3.7-max",
+      name: "Qwen3.7 Max",
+      targetFormat: "claude",
+      supportsVision: false,
+      supportsReasoning: true,
+      supportedThinkingEfforts: ["high", "max"],
+    },
     {
       id: "qwen3.7-max-high",
       name: "Qwen3.7 Max (high effort)",
@@ -79,6 +97,8 @@ export const opencode_goProvider: RegistryEntry = {
       name: "Qwen3.7 Plus",
       targetFormat: "claude",
       supportsVision: false,
+      supportsReasoning: true,
+      supportedThinkingEfforts: ["high", "max"],
     },
     {
       id: "qwen3.7-plus-high",
@@ -111,7 +131,13 @@ export const opencode_goProvider: RegistryEntry = {
       supportsReasoning: true,
     },
     // #8353: hy3 is the Go-tier base id (distinct from hy3-preview / hy3-free).
-    { id: "hy3", name: "Hunyuan3", contextLength: 256000, supportsReasoning: true },
+    {
+      id: "hy3",
+      name: "Hunyuan3",
+      contextLength: 256000,
+      supportsReasoning: true,
+      supportedThinkingEfforts: ["none", "low", "high"],
+    },
     {
       id: "hy3-none",
       name: "Hunyuan3 (none effort)",
@@ -201,7 +227,12 @@ export const opencode_goProvider: RegistryEntry = {
       targetFormat: "openai-responses",
     },
     // #8353: Grok 4.5 + effort tiers from the OpenCode Go registry.
-    { id: "grok-4.5", name: "Grok 4.5", supportsReasoning: true },
+    {
+      id: "grok-4.5",
+      name: "Grok 4.5",
+      supportsReasoning: true,
+      supportedThinkingEfforts: ["low", "medium", "high"],
+    },
     { id: "grok-4.5-low", name: "Grok 4.5 (low effort)", supportsReasoning: true },
     { id: "grok-4.5-medium", name: "Grok 4.5 (medium effort)", supportsReasoning: true },
     { id: "grok-4.5-high", name: "Grok 4.5 (high effort)", supportsReasoning: true },

open-sse/config/providers/registry/opencode/zen/index.ts

@@ -79,18 +79,35 @@ export const opencode_zenProvider: RegistryEntry = {
     },
 
     // ── DeepSeek ────────────────────────────────────────────────
-    { id: "deepseek-v4-pro", name: "DeepSeek V4 Pro" },
-    { id: "deepseek-v4-flash", name: "DeepSeek V4 Flash" },
+    // #10788: same tier vocabulary as opencode-go's DeepSeek rows — the Zen
+    // upstream accepts the identical effort set on these models.
+    {
+      id: "deepseek-v4-pro",
+      name: "DeepSeek V4 Pro",
+      supportsReasoning: true,
+      supportedThinkingEfforts: ["none", "low", "high", "max"],
+    },
+    {
+      id: "deepseek-v4-flash",
+      name: "DeepSeek V4 Flash",
+      supportsReasoning: true,
+      supportedThinkingEfforts: ["none", "low", "high", "max"],
+    },
 
     // ── GLM / Z.AI ─────────────────────────────────────────────
-    { id: "glm-5.2", name: "GLM-5.2" },
+    {
+      id: "glm-5.2",
+      name: "GLM-5.2",
+      supportsReasoning: true,
+      supportedThinkingEfforts: ["high", "max"],
+    },
 
     // ── MiniMax ────────────────────────────────────────────────
     // #3110: MiniMax M3 — frontier coding model with 1M context
     { id: "minimax-m3", name: "MiniMax M3", contextLength: 1048576, supportsVision: true },
 
     // ── Kimi / Moonshot ────────────────────────────────────────
-    { id: "kimi-k3", name: "Kimi K3" },
+    { id: "kimi-k3", name: "Kimi K3", supportsReasoning: true, supportedThinkingEfforts: ["max"] },
     // kimi-k2.7-code declared identically on opencode-go — see OPENCODE_ZEN_GO_SHARED_MODELS.
 
     // ── Qwen ───────────────────────────────────────────────────

open-sse/config/rerankRegistry.ts

@@ -218,3 +218,29 @@ export function getAllRerankModels() {
   }
   return models;
 }
+
+/**
+ * Derive a Cohere-compatible rerank config for a chat provider that has NO
+ * curated RERANK_PROVIDERS entry. Works for any registry provider whose base
+ * URL ends in /chat/completions by swapping that suffix for /rerank (groq,
+ * mistral, vercel-ai-gateway, ...). Dynamic-URL providers (no usable static
+ * base, e.g. dynamic account-scoped hosts) derive to null — they need bespoke
+ * URL handling.
+ *
+ * This is a FALLBACK only: callers must check getRerankProvider() first so
+ * curated entries keep their specialized configuration and format adapters.
+ */
+export function deriveRerankProviderForChatProvider(providerId, chatEntry) {
+  if (!chatEntry) return null;
+  const rawBase = Array.isArray(chatEntry.baseUrl) ? chatEntry.baseUrl[0] : chatEntry.baseUrl;
+  if (!rawBase || typeof rawBase !== "string") return null;
+  const base = rawBase.replace(/\/+$/, "");
+  if (!base.endsWith("/chat/completions")) return null;
+  return {
+    id: providerId,
+    baseUrl: `${base.slice(0, -"/chat/completions".length)}/rerank`,
+    authType: "apikey",
+    authHeader: "bearer",
+    models: [],
+  };
+}

open-sse/config/searchRegistry.ts

@@ -330,6 +330,25 @@ export const SEARCH_PROVIDERS: Record<string, SearchProviderConfig> = {
     timeoutMs: 60_000,
     cacheTTLMs: 5 * 60 * 1000,
   },
+
+  // Direct X API search through Xquik. Keep it fallback-only so the existing
+  // SuperGrok provider remains the default for search_type "x".
+  "xquik-search": {
+    id: "xquik-search",
+    name: "Xquik X Search",
+    baseUrl: "https://xquik.com/api/v1/x/tweets/search",
+    method: "GET",
+    authType: "apikey",
+    authHeader: "x-api-key",
+    costPerQuery: 0.00075,
+    freeMonthlyQuota: 0,
+    searchTypes: ["x"],
+    defaultMaxResults: 5,
+    maxMaxResults: 20,
+    timeoutMs: 15_000,
+    cacheTTLMs: 5 * 60 * 1000,
+    fallbackOnly: true,
+  },
 };
 
 /**
@@ -377,6 +396,8 @@ export const SEARCH_PROVIDER_ALIASES: Record<string, string> = {
   c7: "context7",
   x_search: "x-search",
   x: "x-search",
+  xquik: "xquik-search",
+  xquik_search: "xquik-search",
 };
 
 export function resolveSearchProviderId(providerId: string): string {

open-sse/executors/index.ts

@@ -1,246 +1,218 @@
 import { SEARCH_PROVIDERS } from "../config/searchRegistry.ts";
-import { registerExecutor, getRegisteredExecutor, hasRegisteredExecutor } from "./registry.ts";
+import {
+  registerLazyExecutor,
+  loadRegisteredExecutor,
+  hasRegisteredExecutor,
+} from "./registry.ts";
+// Type-only: pulls no runtime code, keeps DefaultExecutor the only eager class.
 import type { BaseExecutor } from "./base.ts";
-import { AntigravityExecutor } from "./antigravity.ts";
-import { GithubExecutor } from "./github.ts";
-import { GheCopilotExecutor } from "./ghe-copilot.ts";
-import { QoderExecutor } from "./qoder.ts";
-import { KiroExecutor } from "./kiro.ts";
-import { CodexExecutor } from "./codex.ts";
-import { CodexAppServerExecutor } from "./codex-app-server.ts";
-import { CursorExecutor } from "./cursor.ts";
-import { TraeExecutor } from "./trae.ts";
 import { DefaultExecutor } from "./default.ts";
-import { BedrockExecutor } from "./bedrock.ts";
-import { GlmExecutor } from "./glm.ts";
-import { PollinationsExecutor } from "./pollinations.ts";
-import { CloudflareAIExecutor } from "./cloudflare-ai.ts";
-import { FreebuffExecutor } from "./freebuff.ts";
-import { OpencodeExecutor } from "./opencode.ts";
-import { VertexExecutor } from "./vertex.ts";
-import { CliproxyapiExecutor } from "./cliproxyapi.ts";
-import { DarioExecutor } from "./dario.ts";
-import { NineRouterExecutor } from "./ninerouter.ts";
-import { PerplexityWebExecutor } from "./perplexity-web.ts";
-import { GrokWebExecutor } from "./grok-web.ts";
-import { GeminiWebExecutor } from "./gemini-web.ts";
-import { TencentAIStudioWebExecutor } from "./tencent-aistudio-web.ts";
-import { GeminiBusinessExecutor } from "./gemini-business.ts";
-import { ChatGptWebExecutor } from "./chatgpt-web.ts";
-import { ChatGptWebCodexExecutor } from "./chatgpt-web-codex.ts";
-import { BlackboxWebExecutor } from "./blackbox-web.ts";
-import { MuseSparkWebExecutor } from "./muse-spark-web.ts";
-import { AzureOpenAIExecutor } from "./azure-openai.ts";
-import { AzureAiExecutor } from "./azure-ai.ts";
-import { CommandCodeExecutor } from "./commandCode.ts";
-import { GitlabExecutor } from "./gitlab.ts";
-import { NlpCloudExecutor } from "./nlpcloud.ts";
-import { DevinDesktopExecutor } from "./devin-desktop.ts";
-import { ZedHostedExecutor } from "./zed-hosted.ts";
-import { DevinCliExecutor } from "./devin-cli.ts";
-import { ZcodeExecutor } from "./zcode.ts";
-import { DevinCliAgenticExecutor } from "./devin-cli-agentic.ts";
-import { AuggieExecutor } from "./auggie.ts";
-import { DeepSeekWebExecutor } from "./deepseek-web.ts";
-import { DeepSeekWebWithAutoRefreshExecutor } from "./deepseek-web-with-auto-refresh.ts";
-import { AdaptaWebExecutor } from "./adapta-web.ts";
-import { CopilotWebExecutor } from "./copilot-web.ts";
-import { CopilotM365WebExecutor } from "./copilot-m365-web.ts";
-import { MicrosoftDesignerWebExecutor } from "./microsoft-designer-web.ts";
-import { AdobeFireflyExecutor } from "./adobe-firefly.ts";
-import { VeoAIFreeWebExecutor } from "./veoaifree-web.ts";
-import { DuckDuckGoWebExecutor } from "./duckduckgo-web.ts";
-import { FeloWebExecutor } from "./felo-web.ts";
-import { T3ChatWebExecutor } from "./t3-chat-web.ts";
-import { ClaudeWebExecutor } from "./claude-web.ts";
-import { InnerAiExecutor } from "./inner-ai.ts";
-import { HuggingChatExecutor } from "./huggingchat.ts";
-import { YuanbaoWebExecutor } from "./yuanbao-web.ts";
-import { PoeWebExecutor } from "./poe-web.ts";
-import { VeniceWebExecutor } from "./venice-web.ts";
-import { NotionWebExecutor } from "./notion-web.ts";
-import { V0VercelWebExecutor } from "./v0-vercel-web.ts";
-import { CheaperInferenceExecutor } from "./cheaperinference.ts";
-import { KimiWebExecutor } from "./kimi-web.ts";
-import { DoubaoWebExecutor } from "./doubao-web.ts";
-import { QwenWebExecutor } from "./qwen-web.ts";
-import { RaycastExecutor } from "./raycast.ts";
-import { HailuoWebExecutor } from "./hailuo-web.ts";
-import { ZaiWebExecutor } from "./zai-web.ts";
-import { KimiExecutor } from "./kimi.ts";
-import { MoonshotExecutor } from "./moonshot.ts";
-import { TheOldLlmExecutor } from "./theoldllm.ts";
-import { ChipotleExecutor } from "./chipotle.ts";
-import { LMArenaExecutor } from "./lmarena.ts";
-import { GrokCliExecutor } from "./grok-cli.ts";
-import { CodeBuddyCnExecutor } from "./codebuddy-cn.ts";
-import { ZenmuxFreeExecutor } from "./zenmux-free.ts";
-import { CloudflarePlaygroundExecutor } from "./cloudflare-playground.ts";
-import { TinyCmsExecutor } from "./tinycms.ts";
-import { HyperAgentExecutor } from "./hyperagent.ts";
-import { XaiExecutor } from "./xai.ts";
-import { PromptQlExecutor } from "./promptql.ts";
-import { ConolWebExecutor } from "./conol-web.ts";
 
-// R0.3 — declarative built-in table. The object literal stays as the single
-// place built-ins are declared (compile-time duplicate-key safety; the
-// check:known-symbols gate parses this literal from source), but lookup goes
-// through the ExecutorRegistry (./registry.ts): every entry is registered at
-// module load below, and getExecutor()/hasSpecializedExecutor() consult the
-// registry — the literal is never read at request time.
-const executors = {
-  antigravity: new AntigravityExecutor(),
-  agy: new AntigravityExecutor(),
-  github: new GithubExecutor(),
-  "ghe-copilot": new GheCopilotExecutor(),
-  qoder: new QoderExecutor(),
-  kiro: new KiroExecutor(),
-  "amazon-q": new KiroExecutor("amazon-q"),
-  bedrock: new BedrockExecutor(),
-  codex: new CodexExecutor(),
-  "codex-app-server": new CodexAppServerExecutor({}, "codex-app-server"),
-  "chatgpt-web-codex": new ChatGptWebCodexExecutor(),
-  "cgpt-codex": new ChatGptWebCodexExecutor(),
-  cursor: new CursorExecutor(),
-  trae: new TraeExecutor(),
-  glm: new GlmExecutor("glm"),
-  "glm-cn": new GlmExecutor("glm-cn"),
-  glmt: new GlmExecutor("glmt"),
-  cu: new CursorExecutor(), // Alias for cursor
-  "cursor-api": new CursorExecutor("cursor-api"),
-  cua: new CursorExecutor("cursor-api"),
-  "azure-openai": new AzureOpenAIExecutor(),
-  "azure-ai": new AzureAiExecutor(),
-  "command-code": new CommandCodeExecutor(),
-  cmd: new CommandCodeExecutor(), // Alias
-  gitlab: new GitlabExecutor(),
-  "gitlab-duo": new GitlabExecutor("gitlab-duo"),
-  nlpcloud: new NlpCloudExecutor(),
-  pollinations: new PollinationsExecutor(),
-  pol: new PollinationsExecutor(), // Alias
-  "cloudflare-ai": new CloudflareAIExecutor(),
-  cf: new CloudflareAIExecutor(), // Alias
-  freebuff: new FreebuffExecutor(),
-  fb: new FreebuffExecutor(), // Alias
-  "opencode-zen": new OpencodeExecutor("opencode-zen"),
-  "opencode-go": new OpencodeExecutor("opencode-go"),
-  opencode: new OpencodeExecutor("opencode-zen"), // Alias for opencode-zen
-  vertex: new VertexExecutor(),
-  "vertex-partner": new VertexExecutor(),
-  cliproxyapi: new CliproxyapiExecutor(),
-  cpa: new CliproxyapiExecutor(), // Alias
-  dario: new DarioExecutor(),
-  dr: new DarioExecutor(), // Alias
-  "9router": new NineRouterExecutor(),
-  nr: new NineRouterExecutor(), // Alias
-  "perplexity-web": new PerplexityWebExecutor(),
-  "pplx-web": new PerplexityWebExecutor(), // Alias
-  "grok-web": new GrokWebExecutor(),
-  "claude-web": new ClaudeWebExecutor(),
-  "cw-web": new ClaudeWebExecutor(), // Alias
-  "gemini-web": new GeminiWebExecutor(),
-  gweb: new GeminiWebExecutor(), // Alias
-  "gemini-business": new GeminiBusinessExecutor(),
-  gembiz: new GeminiBusinessExecutor(), // Alias
-  "chatgpt-web": new ChatGptWebExecutor(),
-  "cgpt-web": new ChatGptWebExecutor(), // Alias
-  "blackbox-web": new BlackboxWebExecutor(),
-  "bb-web": new BlackboxWebExecutor(), // Alias
-  "muse-spark-web": new MuseSparkWebExecutor(),
-  "ms-web": new MuseSparkWebExecutor(), // Alias
-  "devin-desktop": new DevinDesktopExecutor(),
-  "zed-hosted": new ZedHostedExecutor(),
-  "devin-cli": new DevinCliExecutor(),
-  zcode: new ZcodeExecutor(),
-  zc: new ZcodeExecutor(), // Alias
-  "devin-cli-agentic": new DevinCliAgenticExecutor(),
-  devin: new DevinCliExecutor(), // Alias
-  "deepseek-web": new DeepSeekWebWithAutoRefreshExecutor(),
-  "ds-web": new DeepSeekWebWithAutoRefreshExecutor(), // Alias
-  "adapta-web": new AdaptaWebExecutor(),
-  "adp-web": new AdaptaWebExecutor(), // Alias
-  "copilot-web": new CopilotWebExecutor(),
-  "copilot-m365-web": new CopilotM365WebExecutor(),
-  copilot: new CopilotWebExecutor(), // Alias
-  "microsoft-designer-web": new MicrosoftDesignerWebExecutor(),
-  msdesigner: new MicrosoftDesignerWebExecutor(), // Alias
-  "adobe-firefly": new AdobeFireflyExecutor(),
-  firefly: new AdobeFireflyExecutor(), // Alias
-  "veoaifree-web": new VeoAIFreeWebExecutor(),
-  "veo-free": new VeoAIFreeWebExecutor(), // Alias
-  "duckduckgo-web": new DuckDuckGoWebExecutor(),
-  ddgw: new DuckDuckGoWebExecutor(), // Alias
-  "felo-web": new FeloWebExecutor(),
-  felo: new FeloWebExecutor(), // Alias
-  "t3-web": new T3ChatWebExecutor(),
-  t3chat: new T3ChatWebExecutor(), // Alias
-  "inner-ai": new InnerAiExecutor(),
-  "in-ai": new InnerAiExecutor(), // Alias
-  huggingchat: new HuggingChatExecutor(),
-  hc: new HuggingChatExecutor(), // Alias
-  "yuanbao-web": new YuanbaoWebExecutor(),
-  "tencent-aistudio-web": new TencentAIStudioWebExecutor(),
-  tasw: new TencentAIStudioWebExecutor(),
-  ybw: new YuanbaoWebExecutor(), // Alias
-  "poe-web": new PoeWebExecutor(),
+// R0.3 — declarative built-in table, made LAZY by #11220.
+//
+// The object literal below stays as the single place built-ins are declared
+// (compile-time duplicate-key safety; the check:known-symbols gate parses this
+// literal from source), but its values are now deferred loaders instead of
+// eagerly-constructed instances. At module load every ALIAS is registered in
+// declaration order; the class import + construction happen on first use via
+// loadRegisteredExecutor() and are cached in the same registry a static
+// registration would have populated.
+//
+// Why: importing this barrel previously pulled all ~100 executor modules and
+// constructed every instance at boot — ~0.7–1.2s and ~35MB of heap measured on
+// top of the minimal set — even for deployments that use a handful of
+// providers. Bundlers split the dynamic imports into on-demand chunks.
+//
+// Contract preserved (pinned by tests/unit/executor-map-golden.test.ts):
+//   - keys and their ORDER are byte-stable
+//   - each alias still gets its OWN instance (aliases never share)
+//   - ctor arguments are unchanged
+const lazyExecutors: Record<string, () => Promise<BaseExecutor>> = {
+  antigravity: () => import("./antigravity.ts").then((m) => new m.AntigravityExecutor()),
+  agy: () => import("./antigravity.ts").then((m) => new m.AntigravityExecutor()),
+  github: () => import("./github.ts").then((m) => new m.GithubExecutor()),
+  "ghe-copilot": () => import("./ghe-copilot.ts").then((m) => new m.GheCopilotExecutor()),
+  qoder: () => import("./qoder.ts").then((m) => new m.QoderExecutor()),
+  kiro: () => import("./kiro.ts").then((m) => new m.KiroExecutor()),
+  "amazon-q": () => import("./kiro.ts").then((m) => new m.KiroExecutor("amazon-q")),
+  bedrock: () => import("./bedrock.ts").then((m) => new m.BedrockExecutor()),
+  codex: () => import("./codex.ts").then((m) => new m.CodexExecutor()),
+  "codex-app-server": () =>
+    import("./codex-app-server.ts").then(
+      (m) => new m.CodexAppServerExecutor({}, "codex-app-server")
+    ),
+  "chatgpt-web-codex": () =>
+    import("./chatgpt-web-codex.ts").then((m) => new m.ChatGptWebCodexExecutor()),
+  "cgpt-codex": () =>
+    import("./chatgpt-web-codex.ts").then((m) => new m.ChatGptWebCodexExecutor()),
+  cursor: () => import("./cursor.ts").then((m) => new m.CursorExecutor()),
+  trae: () => import("./trae.ts").then((m) => new m.TraeExecutor()),
+  glm: () => import("./glm.ts").then((m) => new m.GlmExecutor("glm")),
+  "glm-cn": () => import("./glm.ts").then((m) => new m.GlmExecutor("glm-cn")),
+  glmt: () => import("./glm.ts").then((m) => new m.GlmExecutor("glmt")),
+  cu: () => import("./cursor.ts").then((m) => new m.CursorExecutor()), // Alias for cursor
+  "cursor-api": () => import("./cursor.ts").then((m) => new m.CursorExecutor("cursor-api")),
+  cua: () => import("./cursor.ts").then((m) => new m.CursorExecutor("cursor-api")),
+  "azure-openai": () => import("./azure-openai.ts").then((m) => new m.AzureOpenAIExecutor()),
+  "azure-ai": () => import("./azure-ai.ts").then((m) => new m.AzureAiExecutor()),
+  "command-code": () => import("./commandCode.ts").then((m) => new m.CommandCodeExecutor()),
+  cmd: () => import("./commandCode.ts").then((m) => new m.CommandCodeExecutor()), // Alias
+  gitlab: () => import("./gitlab.ts").then((m) => new m.GitlabExecutor()),
+  "gitlab-duo": () => import("./gitlab.ts").then((m) => new m.GitlabExecutor("gitlab-duo")),
+  nlpcloud: () => import("./nlpcloud.ts").then((m) => new m.NlpCloudExecutor()),
+  pollinations: () => import("./pollinations.ts").then((m) => new m.PollinationsExecutor()),
+  pol: () => import("./pollinations.ts").then((m) => new m.PollinationsExecutor()), // Alias
+  "cloudflare-ai": () => import("./cloudflare-ai.ts").then((m) => new m.CloudflareAIExecutor()),
+  cf: () => import("./cloudflare-ai.ts").then((m) => new m.CloudflareAIExecutor()), // Alias
+  freebuff: () => import("./freebuff.ts").then((m) => new m.FreebuffExecutor()),
+  fb: () => import("./freebuff.ts").then((m) => new m.FreebuffExecutor()), // Alias
+  "opencode-zen": () =>
+    import("./opencode.ts").then((m) => new m.OpencodeExecutor("opencode-zen")),
+  "opencode-go": () =>
+    import("./opencode.ts").then((m) => new m.OpencodeExecutor("opencode-go")),
+  opencode: () =>
+    import("./opencode.ts").then((m) => new m.OpencodeExecutor("opencode-zen")), // Alias for opencode-zen
+  vertex: () => import("./vertex.ts").then((m) => new m.VertexExecutor()),
+  "vertex-partner": () => import("./vertex.ts").then((m) => new m.VertexExecutor()),
+  cliproxyapi: () => import("./cliproxyapi.ts").then((m) => new m.CliproxyapiExecutor()),
+  cpa: () => import("./cliproxyapi.ts").then((m) => new m.CliproxyapiExecutor()), // Alias
+  dario: () => import("./dario.ts").then((m) => new m.DarioExecutor()),
+  dr: () => import("./dario.ts").then((m) => new m.DarioExecutor()), // Alias
+  "9router": () => import("./ninerouter.ts").then((m) => new m.NineRouterExecutor()),
+  nr: () => import("./ninerouter.ts").then((m) => new m.NineRouterExecutor()), // Alias
+  "perplexity-web": () =>
+    import("./perplexity-web.ts").then((m) => new m.PerplexityWebExecutor()),
+  "pplx-web": () =>
+    import("./perplexity-web.ts").then((m) => new m.PerplexityWebExecutor()), // Alias
+  "grok-web": () => import("./grok-web.ts").then((m) => new m.GrokWebExecutor()),
+  "claude-web": () => import("./claude-web.ts").then((m) => new m.ClaudeWebExecutor()),
+  "cw-web": () => import("./claude-web.ts").then((m) => new m.ClaudeWebExecutor()), // Alias
+  "gemini-web": () => import("./gemini-web.ts").then((m) => new m.GeminiWebExecutor()),
+  gweb: () => import("./gemini-web.ts").then((m) => new m.GeminiWebExecutor()), // Alias
+  "gemini-business": () =>
+    import("./gemini-business.ts").then((m) => new m.GeminiBusinessExecutor()),
+  gembiz: () =>
+    import("./gemini-business.ts").then((m) => new m.GeminiBusinessExecutor()), // Alias
+  "chatgpt-web": () => import("./chatgpt-web.ts").then((m) => new m.ChatGptWebExecutor()),
+  "cgpt-web": () => import("./chatgpt-web.ts").then((m) => new m.ChatGptWebExecutor()), // Alias
+  "blackbox-web": () => import("./blackbox-web.ts").then((m) => new m.BlackboxWebExecutor()),
+  "bb-web": () => import("./blackbox-web.ts").then((m) => new m.BlackboxWebExecutor()), // Alias
+  "muse-spark-web": () =>
+    import("./muse-spark-web.ts").then((m) => new m.MuseSparkWebExecutor()),
+  "ms-web": () => import("./muse-spark-web.ts").then((m) => new m.MuseSparkWebExecutor()), // Alias
+  "devin-desktop": () => import("./devin-desktop.ts").then((m) => new m.DevinDesktopExecutor()),
+  "zed-hosted": () => import("./zed-hosted.ts").then((m) => new m.ZedHostedExecutor()),
+  "devin-cli": () => import("./devin-cli.ts").then((m) => new m.DevinCliExecutor()),
+  zcode: () => import("./zcode.ts").then((m) => new m.ZcodeExecutor()),
+  zc: () => import("./zcode.ts").then((m) => new m.ZcodeExecutor()), // Alias
+  "devin-cli-agentic": () =>
+    import("./devin-cli-agentic.ts").then((m) => new m.DevinCliAgenticExecutor()),
+  devin: () => import("./devin-cli.ts").then((m) => new m.DevinCliExecutor()), // Alias
+  "deepseek-web": () =>
+    import("./deepseek-web-with-auto-refresh.ts").then(
+      (m) => new m.DeepSeekWebWithAutoRefreshExecutor()
+    ),
+  "ds-web": () =>
+    import("./deepseek-web-with-auto-refresh.ts").then(
+      (m) => new m.DeepSeekWebWithAutoRefreshExecutor()
+    ), // Alias
+  "adapta-web": () => import("./adapta-web.ts").then((m) => new m.AdaptaWebExecutor()),
+  "adp-web": () => import("./adapta-web.ts").then((m) => new m.AdaptaWebExecutor()), // Alias
+  "copilot-web": () => import("./copilot-web.ts").then((m) => new m.CopilotWebExecutor()),
+  "copilot-m365-web": () =>
+    import("./copilot-m365-web.ts").then((m) => new m.CopilotM365WebExecutor()),
+  copilot: () => import("./copilot-web.ts").then((m) => new m.CopilotWebExecutor()), // Alias
+  "microsoft-designer-web": () =>
+    import("./microsoft-designer-web.ts").then((m) => new m.MicrosoftDesignerWebExecutor()),
+  msdesigner: () =>
+    import("./microsoft-designer-web.ts").then((m) => new m.MicrosoftDesignerWebExecutor()), // Alias
+  "adobe-firefly": () => import("./adobe-firefly.ts").then((m) => new m.AdobeFireflyExecutor()),
+  firefly: () => import("./adobe-firefly.ts").then((m) => new m.AdobeFireflyExecutor()), // Alias
+  "veoaifree-web": () => import("./veoaifree-web.ts").then((m) => new m.VeoAIFreeWebExecutor()),
+  "veo-free": () => import("./veoaifree-web.ts").then((m) => new m.VeoAIFreeWebExecutor()), // Alias
+  "duckduckgo-web": () =>
+    import("./duckduckgo-web.ts").then((m) => new m.DuckDuckGoWebExecutor()),
+  ddgw: () => import("./duckduckgo-web.ts").then((m) => new m.DuckDuckGoWebExecutor()), // Alias
+  "felo-web": () => import("./felo-web.ts").then((m) => new m.FeloWebExecutor()),
+  felo: () => import("./felo-web.ts").then((m) => new m.FeloWebExecutor()), // Alias
+  "t3-web": () => import("./t3-chat-web.ts").then((m) => new m.T3ChatWebExecutor()),
+  t3chat: () => import("./t3-chat-web.ts").then((m) => new m.T3ChatWebExecutor()), // Alias
+  "inner-ai": () => import("./inner-ai.ts").then((m) => new m.InnerAiExecutor()),
+  "in-ai": () => import("./inner-ai.ts").then((m) => new m.InnerAiExecutor()), // Alias
+  huggingchat: () => import("./huggingchat.ts").then((m) => new m.HuggingChatExecutor()),
+  hc: () => import("./huggingchat.ts").then((m) => new m.HuggingChatExecutor()), // Alias
+  "yuanbao-web": () => import("./yuanbao-web.ts").then((m) => new m.YuanbaoWebExecutor()),
+  "tencent-aistudio-web": () =>
+    import("./tencent-aistudio-web.ts").then((m) => new m.TencentAIStudioWebExecutor()),
+  tasw: () =>
+    import("./tencent-aistudio-web.ts").then((m) => new m.TencentAIStudioWebExecutor()), // Alias
+  ybw: () => import("./yuanbao-web.ts").then((m) => new m.YuanbaoWebExecutor()), // Alias
+  "poe-web": () => import("./poe-web.ts").then((m) => new m.PoeWebExecutor()),
   // #8969: do NOT alias canonical `poe` (API-key / api.poe.com) to PoeWebExecutor.
   // Registry declares executor:"default"; the hard-coded map previously won and
   // routed API-key traffic to GraphQL /api/gql_POST → HTTP 405.
-  "venice-web": new VeniceWebExecutor(),
-  ven: new VeniceWebExecutor(), // Alias
-  "notion-web": new NotionWebExecutor(),
-  nw: new NotionWebExecutor(), // Alias
-  promptql: new PromptQlExecutor(),
-  pql: new PromptQlExecutor(), // Alias
-  "v0-vercel-web": new V0VercelWebExecutor(),
-  v0: new V0VercelWebExecutor(), // Alias
-  "kimi-web": new KimiWebExecutor(),
-  "kimi-coding-apikey": new KimiExecutor("kimi-coding-apikey"), // Legacy alias
-  "kimi-coding": new KimiExecutor(), // Alias
-  moonshot: new MoonshotExecutor(),
-  kimi: new MoonshotExecutor("kimi"), // Hidden legacy Moonshot provider id
-  cheaperinference: new CheaperInferenceExecutor(),
-  cinf: new CheaperInferenceExecutor("cheaperinference"), // Alias
-  "doubao-web": new DoubaoWebExecutor(),
-  db: new DoubaoWebExecutor(), // Alias
-  "qwen-web": new QwenWebExecutor(),
-  raycast: new RaycastExecutor(),
-  rc: new RaycastExecutor(), // Alias
-  "hailuo-web": new HailuoWebExecutor(),
-  "zai-web": new ZaiWebExecutor(),
-  zw: new ZaiWebExecutor(), // Alias
-  theoldllm: new TheOldLlmExecutor(),
-  tllm: new TheOldLlmExecutor(), // Alias
-  chipotle: new ChipotleExecutor(),
-  pepper: new ChipotleExecutor(), // Alias
-  lmarena: new LMArenaExecutor(),
-  lma: new LMArenaExecutor(), // Alias
-  "grok-cli": new GrokCliExecutor(),
-  gc: new GrokCliExecutor(), // Alias
-  "codebuddy-cn": new CodeBuddyCnExecutor(),
-  cbcn: new CodeBuddyCnExecutor(), // Alias for codebuddy-cn
-  "zenmux-free": new ZenmuxFreeExecutor(),
-  "cloudflare-playground": new CloudflarePlaygroundExecutor(),
-  cfp: new CloudflarePlaygroundExecutor(), // Alias for cloudflare-playground
-  "tinycms-web": new TinyCmsExecutor(),
-  tcw: new TinyCmsExecutor(), // Alias
-  hyperagent: new HyperAgentExecutor(),
-  ha: new HyperAgentExecutor(), // Alias
-  zmf: new ZenmuxFreeExecutor(), // Alias for zenmux-free
-  auggie: new AuggieExecutor(),
-  xai: new XaiExecutor(),
-  "xai-oauth": new XaiExecutor("xai-oauth"),
-  xao: new XaiExecutor("xai-oauth"),
-  qw: new QwenWebExecutor(), // Alias
-  "conol-web": new ConolWebExecutor(),
-  cnl: new ConolWebExecutor(), // Alias
+  "venice-web": () => import("./venice-web.ts").then((m) => new m.VeniceWebExecutor()),
+  ven: () => import("./venice-web.ts").then((m) => new m.VeniceWebExecutor()), // Alias
+  "notion-web": () => import("./notion-web.ts").then((m) => new m.NotionWebExecutor()),
+  nw: () => import("./notion-web.ts").then((m) => new m.NotionWebExecutor()), // Alias
+  promptql: () => import("./promptql.ts").then((m) => new m.PromptQlExecutor()),
+  pql: () => import("./promptql.ts").then((m) => new m.PromptQlExecutor()), // Alias
+  "v0-vercel-web": () => import("./v0-vercel-web.ts").then((m) => new m.V0VercelWebExecutor()),
+  v0: () => import("./v0-vercel-web.ts").then((m) => new m.V0VercelWebExecutor()), // Alias
+  "kimi-web": () => import("./kimi-web.ts").then((m) => new m.KimiWebExecutor()),
+  "kimi-coding-apikey": () =>
+    import("./kimi.ts").then((m) => new m.KimiExecutor("kimi-coding-apikey")), // Legacy alias
+  "kimi-coding": () => import("./kimi.ts").then((m) => new m.KimiExecutor()), // Alias
+  moonshot: () => import("./moonshot.ts").then((m) => new m.MoonshotExecutor()),
+  kimi: () => import("./moonshot.ts").then((m) => new m.MoonshotExecutor("kimi")), // Hidden legacy Moonshot provider id
+  cheaperinference: () =>
+    import("./cheaperinference.ts").then((m) => new m.CheaperInferenceExecutor()),
+  cinf: () =>
+    import("./cheaperinference.ts").then(
+      (m) => new m.CheaperInferenceExecutor("cheaperinference")
+    ), // Alias
+  "doubao-web": () => import("./doubao-web.ts").then((m) => new m.DoubaoWebExecutor()),
+  db: () => import("./doubao-web.ts").then((m) => new m.DoubaoWebExecutor()), // Alias
+  "qwen-web": () => import("./qwen-web.ts").then((m) => new m.QwenWebExecutor()),
+  raycast: () => import("./raycast.ts").then((m) => new m.RaycastExecutor()),
+  rc: () => import("./raycast.ts").then((m) => new m.RaycastExecutor()), // Alias
+  "hailuo-web": () => import("./hailuo-web.ts").then((m) => new m.HailuoWebExecutor()),
+  "zai-web": () => import("./zai-web.ts").then((m) => new m.ZaiWebExecutor()),
+  zw: () => import("./zai-web.ts").then((m) => new m.ZaiWebExecutor()), // Alias
+  theoldllm: () => import("./theoldllm.ts").then((m) => new m.TheOldLlmExecutor()),
+  tllm: () => import("./theoldllm.ts").then((m) => new m.TheOldLlmExecutor()), // Alias
+  chipotle: () => import("./chipotle.ts").then((m) => new m.ChipotleExecutor()),
+  pepper: () => import("./chipotle.ts").then((m) => new m.ChipotleExecutor()), // Alias
+  lmarena: () => import("./lmarena.ts").then((m) => new m.LMArenaExecutor()),
+  lma: () => import("./lmarena.ts").then((m) => new m.LMArenaExecutor()), // Alias
+  "grok-cli": () => import("./grok-cli.ts").then((m) => new m.GrokCliExecutor()),
+  gc: () => import("./grok-cli.ts").then((m) => new m.GrokCliExecutor()), // Alias
+  "codebuddy-cn": () => import("./codebuddy-cn.ts").then((m) => new m.CodeBuddyCnExecutor()),
+  cbcn: () => import("./codebuddy-cn.ts").then((m) => new m.CodeBuddyCnExecutor()), // Alias for codebuddy-cn
+  "zenmux-free": () => import("./zenmux-free.ts").then((m) => new m.ZenmuxFreeExecutor()),
+  "cloudflare-playground": () =>
+    import("./cloudflare-playground.ts").then((m) => new m.CloudflarePlaygroundExecutor()),
+  cfp: () =>
+    import("./cloudflare-playground.ts").then((m) => new m.CloudflarePlaygroundExecutor()), // Alias for cloudflare-playground
+  "tinycms-web": () => import("./tinycms.ts").then((m) => new m.TinyCmsExecutor()),
+  tcw: () => import("./tinycms.ts").then((m) => new m.TinyCmsExecutor()), // Alias
+  hyperagent: () => import("./hyperagent.ts").then((m) => new m.HyperAgentExecutor()),
+  ha: () => import("./hyperagent.ts").then((m) => new m.HyperAgentExecutor()), // Alias
+  zmf: () => import("./zenmux-free.ts").then((m) => new m.ZenmuxFreeExecutor()), // Alias for zenmux-free
+  auggie: () => import("./auggie.ts").then((m) => new m.AuggieExecutor()),
+  xai: () => import("./xai.ts").then((m) => new m.XaiExecutor()),
+  "xai-oauth": () => import("./xai.ts").then((m) => new m.XaiExecutor("xai-oauth")),
+  xao: () => import("./xai.ts").then((m) => new m.XaiExecutor("xai-oauth")),
+  qw: () => import("./qwen-web.ts").then((m) => new m.QwenWebExecutor()), // Alias
+  "conol-web": () => import("./conol-web.ts").then((m) => new m.ConolWebExecutor()),
+  cnl: () => import("./conol-web.ts").then((m) => new m.ConolWebExecutor()), // Alias
 };
 
-// Bootstrap: register every built-in in the ExecutorRegistry. registerExecutor
-// throws on duplicates, so an alias collision fails at module load, exactly as
-// loudly as a duplicate object key would have failed at lint time.
-for (const [alias, executor] of Object.entries(executors) as [string, BaseExecutor][]) {
-  registerExecutor(alias, executor);
+// Bootstrap: declare every built-in alias in the ExecutorRegistry. Duplicate
+// aliases fail at module load, exactly as loudly as a duplicate object key
+// would have failed at lint time. Instances materialize on first use (#11220).
+for (const [alias, load] of Object.entries(lazyExecutors)) {
+  registerLazyExecutor(alias, load);
 }
 
 const defaultCache = new Map();
@@ -267,9 +239,9 @@ const CHAT_UNSUPPORTED_CLOUD_AGENT_PROVIDERS = new Set(["jules"]);
 // providers must be executed through /v1/search, never the chat-completions path.
 const CHAT_UNSUPPORTED_SEARCH_PROVIDERS = new Set(Object.keys(SEARCH_PROVIDERS));
 
-export function getExecutor(provider) {
-  const registered = getRegisteredExecutor(provider);
-  if (registered) return registered;
+export async function getExecutor(provider: string): Promise<BaseExecutor> {
+  const loaded = await loadRegisteredExecutor(provider);
+  if (loaded) return loaded;
   if (CHAT_UNSUPPORTED_CLOUD_AGENT_PROVIDERS.has(provider)) {
     const err = new Error(
       `Provider "${provider}" is a cloud-agent provider and does not support direct chat completions; use the Cloud Agents task API instead.`
@@ -285,77 +257,19 @@ export function getExecutor(provider) {
     throw err;
   }
   if (!defaultCache.has(provider)) defaultCache.set(provider, new DefaultExecutor(provider));
-  return defaultCache.get(provider);
+  return defaultCache.get(provider)!;
 }
 
-export function hasSpecializedExecutor(provider) {
+export function hasSpecializedExecutor(provider: string): boolean {
   return hasRegisteredExecutor(provider);
 }
 
-export { registerExecutor, listExecutorAliases } from "./registry.ts";
-
+export {
+  registerExecutor,
+  registerLazyExecutor,
+  listExecutorAliases,
+} from "./registry.ts";
+// Value re-export: base.ts is already eager (DefaultExecutor extends it), and
+// scripts/check/check-known-symbols.ts reads this export from the module.
 export { BaseExecutor } from "./base.ts";
-export { AntigravityExecutor } from "./antigravity.ts";
-export { GithubExecutor } from "./github.ts";
-export { QoderExecutor } from "./qoder.ts";
-export { KiroExecutor } from "./kiro.ts";
-export { CodexExecutor } from "./codex.ts";
-export { CursorExecutor } from "./cursor.ts";
-export { TraeExecutor } from "./trae.ts";
 export { DefaultExecutor } from "./default.ts";
-export { BedrockExecutor } from "./bedrock.ts";
-export { GlmExecutor } from "./glm.ts";
-export { PollinationsExecutor } from "./pollinations.ts";
-export { CloudflareAIExecutor } from "./cloudflare-ai.ts";
-export { OpencodeExecutor } from "./opencode.ts";
-export { CliproxyapiExecutor } from "./cliproxyapi.ts";
-export { DarioExecutor } from "./dario.ts";
-export { NineRouterExecutor } from "./ninerouter.ts";
-export { VertexExecutor } from "./vertex.ts";
-export { PerplexityWebExecutor } from "./perplexity-web.ts";
-export { GrokWebExecutor } from "./grok-web.ts";
-export { GeminiWebExecutor } from "./gemini-web.ts";
-export { KieExecutor } from "./kie.ts";
-export { ChatGptWebExecutor } from "./chatgpt-web.ts";
-export { BlackboxWebExecutor } from "./blackbox-web.ts";
-export { MuseSparkWebExecutor } from "./muse-spark-web.ts";
-export { AzureOpenAIExecutor } from "./azure-openai.ts";
-export { AzureAiExecutor } from "./azure-ai.ts";
-export { CommandCodeExecutor } from "./commandCode.ts";
-export { GitlabExecutor } from "./gitlab.ts";
-export { NlpCloudExecutor } from "./nlpcloud.ts";
-export { DevinDesktopExecutor } from "./devin-desktop.ts";
-export { ZedHostedExecutor } from "./zed-hosted.ts";
-export { DevinCliExecutor } from "./devin-cli.ts";
-export { DevinCliAgenticExecutor } from "./devin-cli-agentic.ts";
-export { AuggieExecutor } from "./auggie.ts";
-export { CopilotWebExecutor } from "./copilot-web.ts";
-export { CopilotM365WebExecutor } from "./copilot-m365-web.ts";
-export { MicrosoftDesignerWebExecutor } from "./microsoft-designer-web.ts";
-export { AdobeFireflyExecutor } from "./adobe-firefly.ts";
-export { VeoAIFreeWebExecutor } from "./veoaifree-web.ts";
-export { DuckDuckGoWebExecutor } from "./duckduckgo-web.ts";
-export { FeloWebExecutor } from "./felo-web.ts";
-export { ClaudeWebExecutor } from "./claude-web.ts";
-export { DeepSeekWebExecutor } from "./deepseek-web.ts";
-export { DeepSeekWebWithAutoRefreshExecutor } from "./deepseek-web-with-auto-refresh.ts";
-export { AdaptaWebExecutor } from "./adapta-web.ts";
-export { YuanbaoWebExecutor } from "./yuanbao-web.ts";
-export { T3ChatWebExecutor } from "./t3-chat-web.ts";
-export { InnerAiExecutor } from "./inner-ai.ts";
-export { QwenWebExecutor } from "./qwen-web.ts";
-export { HailuoWebExecutor } from "./hailuo-web.ts";
-export { TheOldLlmExecutor } from "./theoldllm.ts";
-export { ChipotleExecutor } from "./chipotle.ts";
-export { LMArenaExecutor } from "./lmarena.ts";
-export { GrokCliExecutor } from "./grok-cli.ts";
-export { CodeBuddyCnExecutor } from "./codebuddy-cn.ts";
-export { ZenmuxFreeExecutor } from "./zenmux-free.ts";
-export { CloudflarePlaygroundExecutor } from "./cloudflare-playground.ts";
-export { TinyCmsExecutor } from "./tinycms.ts";
-export { HyperAgentExecutor } from "./hyperagent.ts";
-export { XaiExecutor } from "./xai.ts";
-export { MoonshotExecutor } from "./moonshot.ts";
-export { CheaperInferenceExecutor } from "./cheaperinference.ts";
-export { PromptQlExecutor } from "./promptql.ts";
-export { ConolWebExecutor } from "./conol-web.ts";

open-sse/executors/opencode.ts

@@ -895,10 +895,21 @@ export class OpencodeExecutor extends BaseExecutor {
       const mb = modifiedBody as Record<string, unknown>;
       const parsed = parseEffortLevel(model);
       if (parsed) {
-        mb.model = parsed.baseModel;
-        if (mb.reasoning_effort === undefined) {
-          mb.reasoning_effort = parsed.effort;
+        const deepseekFamily =
+          parsed.baseModel === "deepseek-v4-pro" || parsed.baseModel === "deepseek-v4-flash";
+        if (deepseekFamily) {
+          // DeepSeek via opencode-go proxies the native DeepSeek contract, which
+          // accepts a flat reasoning_effort field (#4647).
+          mb.model = parsed.baseModel;
+          if (mb.reasoning_effort === undefined) {
+            mb.reasoning_effort = parsed.effort;
+          }
         }
+        // #10788: every other family's ONLY native effort mechanism is the
+        // -<tier> suffix in the model id itself (the ids `opencode models
+        // opencode-go --verbose` lists). The opencode-go ChatCompletionRequest
+        // carries no flat reasoning_effort field, so rewriting to the base id
+        // silently dropped the tier — forward the aliased id verbatim instead.
       }
     }
     // #1543 / upstream PR #1099: thinking-mode upstreams routed through OpenCode

open-sse/executors/registry.ts

@@ -28,11 +28,46 @@ export function getRegisteredExecutor(alias: string): BaseExecutor | undefined {
   return registry.get(alias);
 }
 
-export function hasRegisteredExecutor(alias: string): boolean {
-  return registry.has(alias);
+// ── #11220: lazy registration ───────────────────────────────────────────────
+// Aliases may register a deferred loader instead of an instance. The alias and
+// its registration ORDER are declared eagerly — hasRegisteredExecutor() and
+// listExecutorAliases() stay synchronous and the golden snapshot keeps its
+// shape — while the class import + construction happen on first use. A
+// completed load caches into `registry`, so later resolution is identical to a
+// static registration.
+const lazyLoaders = new Map<string, () => Promise<BaseExecutor>>();
+const lazyInFlight = new Map<string, Promise<BaseExecutor>>();
+
+export function registerLazyExecutor(alias: string, load: () => Promise<BaseExecutor>): void {
+  if (registry.has(alias) || lazyLoaders.has(alias)) {
+    throw new Error(`executor alias already registered: "${alias}"`);
+  }
+  lazyLoaders.set(alias, load);
 }
 
-/** All registered aliases, in registration order. */
+export function loadRegisteredExecutor(alias: string): Promise<BaseExecutor> | undefined {
+  const cached = registry.get(alias);
+  if (cached) return Promise.resolve(cached);
+  const load = lazyLoaders.get(alias);
+  if (!load) return undefined;
+  let inFlight = lazyInFlight.get(alias);
+  if (!inFlight) {
+    inFlight = load().then((executor) => {
+      registerExecutor(alias, executor);
+      lazyLoaders.delete(alias);
+      lazyInFlight.delete(alias);
+      return executor;
+    });
+    lazyInFlight.set(alias, inFlight);
+  }
+  return inFlight;
+}
+
+export function hasRegisteredExecutor(alias: string): boolean {
+  return registry.has(alias) || lazyLoaders.has(alias);
+}
+
+/** All registered aliases — static and lazy — in registration order. */
 export function listExecutorAliases(): string[] {
-  return [...registry.keys()];
+  return [...registry.keys(), ...lazyLoaders.keys()];
 }

open-sse/handlers/chatCore.ts

@@ -396,7 +396,7 @@ import {
 } from "../services/rateLimitManager.ts";
 import * as localLimiterErrors from "../services/rateLimitManager/errors.ts";
 import {
-  acquire as acquireAccountSemaphore,
+  acquireMany as acquireConcurrencyGates,
   markBlocked as markAccountSemaphoreBlocked,
 } from "../services/accountSemaphore.ts";
 import { lockModel, lockModelIfPerModelQuota } from "../services/accountFallback.ts";
@@ -447,6 +447,7 @@ import { extractFacts } from "@/lib/memory/extraction";
 import { handleToolCallExecution } from "@/lib/skills/interception";
 import { MEMORY_BUILTIN_TOOL_NAMES } from "@/lib/skills/memoryBuiltins";
 import { OMNIROUTE_RESPONSE_HEADERS } from "@/shared/constants/headers";
+import { resolveProviderId } from "@/shared/constants/providers";
 import { getClaudeCodeCompatibleRequestDefaults } from "@/lib/providers/requestDefaults";
 import {
   buildClaudeCodeCompatibleRequest,
@@ -525,6 +526,7 @@ export async function handleChatCore({
   managedLease = null,
 }) {
   let { provider, model, extendedContext } = modelInfo;
+  const resilienceSettings = resolveResilienceSettings(cachedSettings);
   if (!skipResourcePressureGuard) {
     try {
       const pressureGuard = checkResourcePressureGuard();
@@ -3051,6 +3053,10 @@ export async function handleChatCore({
               connectionId: attemptConnectionId,
               credentials: execCreds,
             });
+            const canonicalProviderKey = resolveProviderId(String(provider).trim().toLowerCase());
+            const providerConcurrency =
+              resilienceSettings.providerQuotaOverrides[canonicalProviderKey]
+                ?.providerConcurrency ?? 0;
 
             trace("pre_semaphore", {
               semaphoreKey: accountSemaphoreKey,
@@ -3061,13 +3067,27 @@ export async function handleChatCore({
                 stage: "waiting_account_slot",
               });
             }
-            const releaseAccountSemaphore =
-              accountSemaphoreKey && accountSemaphoreMaxConcurrency != null
-                ? await acquireAccountSemaphore(accountSemaphoreKey, {
-                    maxConcurrency: accountSemaphoreMaxConcurrency,
-                    signal: streamController.signal,
-                  })
-                : () => {};
+            const releaseAccountSemaphore = await acquireConcurrencyGates(
+              [
+                {
+                  key: "global",
+                  maxConcurrency: resilienceSettings.requestQueue.globalConcurrentRequests,
+                },
+                {
+                  key: `provider:${canonicalProviderKey}`,
+                  maxConcurrency: providerConcurrency,
+                },
+                {
+                  key: accountSemaphoreKey || "",
+                  maxConcurrency: accountSemaphoreKey ? accountSemaphoreMaxConcurrency : null,
+                },
+              ],
+              {
+                timeoutMs: resilienceSettings.requestQueue.maxWaitMs,
+                maxQueueSize: resilienceSettings.requestQueue.maxQueueDepth,
+                signal: streamController.signal,
+              }
+            );
             trace("post_semaphore");
             updatePendingScope(pendingScope, {
               stage: "waiting_rate_limit",
@@ -3328,6 +3348,7 @@ export async function handleChatCore({
                       "ANTIGRAVITY_BYOP_ROTATION",
                       `BYOP 422 on connection ${String(byopFailedId).slice(0, 8)} → rotating to ${String(byopNextCreds.connectionId).slice(0, 8)}`
                     );
+                    releaseAccountSemaphore();
                     Object.assign(credentials, byopNextCreds);
                     antigravityByopRotationPending = true;
                     continue;

open-sse/handlers/chatCore/cliproxyModelMapping.ts

@@ -14,12 +14,12 @@
 type ExecutorInput = {
   model: string;
   body: unknown;
-  [key: string]: unknown;
 };
 
+// No index signature: executors (BaseExecutor subclasses) must satisfy this
+// structurally, and class instances don't carry index signatures.
 type ExecutorLike = {
   execute: (input: ExecutorInput) => Promise<unknown>;
-  [key: string]: unknown;
 };
 
 export type CliproxyapiModelMapping = Record<string, unknown> | null | undefined;

open-sse/handlers/chatCore/cliproxyapiCredentials.ts

@@ -19,12 +19,12 @@ import type { ProviderCredentials } from "../../executors/base.ts";
 
 type ExecutorInput = {
   credentials: ProviderCredentials;
-  [key: string]: unknown;
 };
 
+// No index signature: executors (BaseExecutor subclasses) must satisfy this
+// structurally, and class instances don't carry index signatures.
 type ExecutorLike = {
   execute: (input: ExecutorInput) => Promise<unknown>;
-  [key: string]: unknown;
 };
 
 /**

open-sse/handlers/chatCore/executorProxy.ts

@@ -76,12 +76,15 @@ async function loadCliproxyapiSettings(): Promise<{
  * dedicated-credential wrappers applied. Used by the direct `cliproxyapi` leg
  * and the CLIProxyAPI branch of `fallback`.
  */
-function resolveCliproxyapiExecutor(
+async function resolveCliproxyapiExecutor(
   cliproxyapiModelMapping: Record<string, unknown> | null,
   dedicatedApiKey: string | null
 ) {
   return wrapExecutorWithCliproxyapiCredentials(
-    wrapExecutorWithCliproxyapiModelMapping(getExecutor("cliproxyapi"), cliproxyapiModelMapping),
+    wrapExecutorWithCliproxyapiModelMapping(
+      await getExecutor("cliproxyapi"),
+      cliproxyapiModelMapping
+    ),
     dedicatedApiKey
   );
 }
@@ -138,7 +141,7 @@ export async function resolveExecutorWithProxy(
   // backend on specific failures. The backend defaults to CLIProxyAPI so every
   // pre-existing fallback config behaves exactly as before; fallbackBackend
   // === "dario" opts the retry leg over to Dario instead.
-  const nativeExec = getExecutor(prov);
+  const nativeExec = await getExecutor(prov);
   const fallbackBackend: FallbackBackend = cfg.fallbackBackend;
   const { fallbackCodes, dedicatedApiKey } = await loadCliproxyapiSettings();
 
@@ -146,8 +149,8 @@ export async function resolveExecutorWithProxy(
   // the native leg must keep seeing the original, unmapped model.
   const proxyExec =
     fallbackBackend === "dario"
-      ? getExecutor("dario")
-      : resolveCliproxyapiExecutor(cfg.cliproxyapiModelMapping, dedicatedApiKey);
+      ? await getExecutor("dario")
+      : await resolveCliproxyapiExecutor(cfg.cliproxyapiModelMapping, dedicatedApiKey);
   const backendLabel = fallbackBackend === "dario" ? "Dario" : "CLIProxyAPI";
   const isRetryableStatus = (s: number) => fallbackCodes.includes(s) || s === 0;
 

open-sse/handlers/rerank.ts

@@ -201,6 +201,7 @@ export async function handleRerank({
   connectionId = null,
   apiKeyId = null,
   apiKeyName = null,
+  resolvedProvider = null,
 }) {
   const startTime = Date.now();
   if (!model) return errorResponse(400, "model is required");
@@ -210,7 +211,8 @@ export async function handleRerank({
   }
 
   const { provider: providerId, model: modelId } = parseRerankModel(model);
-  const providerConfig = providerId ? getRerankProvider(providerId) : null;
+  const providerConfig =
+    resolvedProvider || (providerId ? getRerankProvider(providerId) : null);
 
   if (!providerConfig) {
     const availableProviders = Object.keys(RERANK_PROVIDERS).join(", ");
@@ -219,10 +221,13 @@ export async function handleRerank({
       `No rerank provider found for model "${model}". Available: ${availableProviders}`
     );
   }
+  // When a derived/generic provider is injected, its id is authoritative for
+  // logging and cost attribution even though parseRerankModel returned null.
+  const effectiveProviderId = providerConfig.id || providerId;
 
   const token = credentials?.apiKey || credentials?.accessToken;
   if (!token) {
-    return errorResponse(401, `No credentials for rerank provider: ${providerId}`);
+    return errorResponse(401, `No credentials for rerank provider: ${effectiveProviderId}`);
   }
 
   const requestBody = transformRequestForProvider(providerConfig, {
@@ -275,8 +280,8 @@ export async function handleRerank({
         method: "POST",
         path: "/v1/rerank",
         status: res.status,
-        model: `${providerId}/${modelId}`,
-        provider: providerId,
+        model: `${effectiveProviderId}/${modelId}`,
+        provider: effectiveProviderId,
         connectionId: connectionId || undefined,
         duration: Date.now() - startTime,
         requestBody,
@@ -296,14 +301,14 @@ export async function handleRerank({
     });
 
     const searchUnits = Number(result?.meta?.billed_units?.search_units) || 0;
-    const costUsd = await calculateModalCost("rerank", providerId, modelId, { searchUnits });
+    const costUsd = await calculateModalCost("rerank", effectiveProviderId, modelId, { searchUnits });
 
     saveCallLog({
       method: "POST",
       path: "/v1/rerank",
       status: 200,
-      model: `${providerId}/${modelId}`,
-      provider: providerId,
+      model: `${effectiveProviderId}/${modelId}`,
+      provider: effectiveProviderId,
       connectionId: connectionId || undefined,
       duration: Date.now() - startTime,
       tokens: { prompt_tokens: 0, completion_tokens: 0 },
@@ -315,7 +320,7 @@ export async function handleRerank({
 
     const headers = new Headers({ ...CORS_HEADERS, "Content-Type": "application/json" });
     attachOmniRouteMetaHeaders(headers, {
-      provider: providerId,
+      provider: effectiveProviderId,
       model: modelId,
       costUsd,
       latencyMs: Date.now() - startTime,

open-sse/handlers/search.ts

@@ -8,6 +8,7 @@ import { randomUUID } from "crypto";
  *   firecrawl, google-pse-search, linkup-search, searchapi-search,
  *   youcom-search, searxng-search, ollama-search, zai-search, jina-search,
  *   duckduckgo-free, x-search (Grok / SuperGrok X Search — explicit or search_type "x")
+ *   and xquik-search (direct X API search — explicit or credentialed fallback)
  *
  * Request format:
  * {
@@ -28,6 +29,7 @@ import * as fcSearch from "./search/firecrawlSearch.ts";
 import { type FirecrawlSearchEnvelope } from "./search/firecrawlSearch.ts";
 import { buildJinaSearchRequest, extractJinaSearchItems } from "./search/jinaSearch.ts";
 import * as xSearch from "./search/xSearch.ts";
+import * as xquikSearch from "./search/xquikSearch.ts";
 import { freeWebSearch } from "../services/freeWebSearch.ts";
 import { saveCallLog } from "@/lib/usageDb";
 import { safeOutboundFetch } from "@/shared/network/safeOutboundFetch";
@@ -714,6 +716,7 @@ const requestBuilders: Record<string, SearchRequestBuilder> = {
   "ollama-search": buildOllamaRequest,
   "jina-search": buildJinaSearchRequest,
   "x-search": xSearch.buildXSearchRequest,
+  "xquik-search": xquikSearch.buildXquikSearchRequest,
 };
 
 function buildRequest(
@@ -1290,6 +1293,7 @@ const responseNormalizers: Record<string, SearchResponseNormalizer> = {
   "ollama-search": normalizeOllamaResponse,
   "jina-search": normalizeJinaSearchResponse,
   "x-search": normalizeXSearchResponse,
+  "xquik-search": (data) => xquikSearch.normalizeXquikSearchResponse(data, makeResult),
 };
 
 function normalizeResponse(

open-sse/handlers/search/xquikSearch.ts

@@ -0,0 +1,158 @@
+/** Xquik-backed X search for the unified search gateway. */
+
+import { z } from "zod";
+import type { SearchProviderConfig } from "../../config/searchRegistry.ts";
+import type { SearchResult } from "../search.ts";
+
+export const XQUIK_SEARCH_PROVIDER_ID = "xquik-search";
+
+export interface XquikSearchParams {
+  query: string;
+  maxResults: number;
+  token?: string;
+  timeRange?: string;
+  providerOptions?: Record<string, unknown>;
+  providerSpecificData?: Record<string, unknown>;
+}
+
+export interface XquikSearchHit {
+  title: string;
+  url: string;
+  snippet: string;
+  author?: string;
+  publishedAt?: string;
+}
+
+type MakeResult = (
+  providerId: string,
+  item: {
+    title?: string;
+    url?: string;
+    snippet?: string;
+    published_at?: string;
+    author?: string;
+    source_type?: string;
+  },
+  index: number,
+  now: string
+) => SearchResult;
+
+const X_HANDLE_RE = /^[A-Za-z0-9_]{1,15}$/;
+const TWEET_ID_RE = /^\d+$/;
+
+const XquikTweetSchema = z
+  .object({
+    id: z.string().regex(TWEET_ID_RE),
+    text: z.string(),
+    createdAt: z.string().optional(),
+    author: z
+      .object({
+        username: z.string().regex(X_HANDLE_RE),
+        name: z.string().optional(),
+      })
+      .passthrough()
+      .optional(),
+  })
+  .passthrough();
+
+const XquikSearchEnvelopeSchema = z
+  .object({
+    tweets: z.array(z.unknown()).default([]),
+  })
+  .passthrough();
+
+function getProviderSettingString(
+  params: Pick<XquikSearchParams, "providerOptions" | "providerSpecificData">,
+  key: string
+): string | undefined {
+  const option = params.providerOptions?.[key];
+  if (typeof option === "string" && option.trim()) return option.trim();
+  const configured = params.providerSpecificData?.[key];
+  if (typeof configured === "string" && configured.trim()) return configured.trim();
+  return undefined;
+}
+
+function sinceTimeForRange(timeRange: string | undefined, now = Date.now()): string | undefined {
+  const hour = 60 * 60 * 1000;
+  const durations: Record<string, number> = {
+    hour,
+    day: 24 * hour,
+    week: 7 * 24 * hour,
+    month: 30 * 24 * hour,
+    year: 365 * 24 * hour,
+  };
+  const duration = timeRange ? durations[timeRange] : undefined;
+  return duration ? new Date(now - duration).toISOString() : undefined;
+}
+
+export function buildXquikSearchRequest(
+  config: SearchProviderConfig,
+  params: XquikSearchParams
+): { url: string; init: RequestInit } {
+  const queryType = getProviderSettingString(params, "queryType") === "Top" ? "Top" : "Latest";
+  const query = new URLSearchParams({
+    q: params.query,
+    queryType,
+    limit: String(params.maxResults),
+  });
+  const sinceTime = sinceTimeForRange(params.timeRange);
+  if (sinceTime) query.set("sinceTime", sinceTime);
+
+  return {
+    url: `${config.baseUrl.replace(/\/+$/, "")}?${query}`,
+    init: {
+      method: "GET",
+      headers: {
+        Accept: "application/json",
+        ...(params.token ? { "x-api-key": params.token } : {}),
+      },
+    },
+  };
+}
+
+export function extractXquikSearchHits(data: unknown, maxResults: number): XquikSearchHit[] {
+  const envelope = XquikSearchEnvelopeSchema.safeParse(data);
+  if (!envelope.success) return [];
+
+  const hits: XquikSearchHit[] = [];
+  for (const value of envelope.data.tweets) {
+    const parsed = XquikTweetSchema.safeParse(value);
+    if (!parsed.success) continue;
+    const tweet = parsed.data;
+    const author = tweet.author?.username;
+    hits.push({
+      title: author ? `@${author}` : "X post",
+      url: author
+        ? `https://x.com/${author}/status/${tweet.id}`
+        : `https://x.com/i/status/${tweet.id}`,
+      snippet: tweet.text,
+      author,
+      publishedAt: tweet.createdAt,
+    });
+    if (hits.length >= maxResults) break;
+  }
+  return hits;
+}
+
+export function normalizeXquikSearchResponse(
+  data: unknown,
+  makeResult: MakeResult
+): { results: SearchResult[]; totalResults: number } {
+  const now = new Date().toISOString();
+  const results = extractXquikSearchHits(data, 20).map((hit, index) =>
+    makeResult(
+      XQUIK_SEARCH_PROVIDER_ID,
+      {
+        title: hit.title,
+        url: hit.url,
+        snippet: hit.snippet,
+        published_at: hit.publishedAt,
+        author: hit.author,
+        source_type: "x",
+      },
+      index,
+      now
+    )
+  );
+  return { results, totalResults: results.length };
+}

open-sse/handlers/videoGeneration.ts

@@ -374,7 +374,7 @@ async function handleVertexVeoGeneration({ model, body, credentials, log }) {
  * Submits an AnimateDiff or SVD workflow, polls for completion, fetches output video
  */
 async function handleVeoAiFreeVideoGeneration({ model, provider, body, credentials, log }) {
-  const executor = getExecutor(provider);
+  const executor = await getExecutor(provider);
   if (!executor) {
     return { success: false, status: 400, error: `Unknown video provider: ${provider}` };
   }

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

@@ -358,6 +358,38 @@ describe("omniroute_x_search handler (via MCP dispatch)", () => {
     expect(body.search_type).toBe("x");
     expect(body.provider).toBe("x-search");
   });
+
+  it("should route an explicit Xquik search through xquik-search", async () => {
+    mockFetch.mockResolvedValueOnce({
+      ok: true,
+      json: async () => ({
+        id: "xs2",
+        provider: "xquik-search",
+        query: "agents sdk",
+        results: [
+          {
+            title: "@openai",
+            url: "https://x.com/openai/status/1912345678901234567",
+            snippet: "Agents SDK update",
+            position: 1,
+          },
+        ],
+        cached: false,
+        usage: { queries_used: 1, search_cost_usd: 0.00015 },
+      }),
+    });
+
+    const result = await client.callTool({
+      name: "omniroute_x_search",
+      arguments: { query: "agents sdk", max_results: 5, provider: "xquik-search" },
+    });
+
+    expect(result.isError).toBeFalsy();
+    const [, options] = mockFetch.mock.calls[0];
+    const body = JSON.parse(options.body as string);
+    expect(body.search_type).toBe("x");
+    expect(body.provider).toBe("xquik-search");
+  });
 });
 
 // ── omniroute_get_health: handler dispatch tests ──────────────────────────────

open-sse/mcp-server/schemas/tools.ts

@@ -531,12 +531,17 @@ export const xSearchInput = z.object({
     .max(20)
     .default(5)
     .describe("Maximum number of X results to return"),
+  provider: z
+    .enum(["x-search", "xquik-search"])
+    .optional()
+    .default("x-search")
+    .describe("X search backend: x-search uses xAI/SuperGrok; xquik-search uses Xquik"),
 });
 
 export const xSearchTool: McpToolDefinition<typeof xSearchInput, typeof webSearchOutput> = {
   name: "omniroute_x_search",
   description:
-    "Search X (Twitter) through OmniRoute using SuperGrok / xAI server-side x_search. Requires a connected xai-oauth (SuperGrok) or xAI API key. This is Grok X Search, not web search and not the X Developer Platform MCP.",
+    "Search X (Twitter) through OmniRoute. Uses SuperGrok / xAI server-side x_search by default, or Xquik when provider is xquik-search. Requires credentials for the selected backend. This is not web search.",
   inputSchema: xSearchInput,
   outputSchema: webSearchOutput,
   scopes: ["execute:search"],

open-sse/mcp-server/server.ts

@@ -666,7 +666,11 @@ async function handleWebSearch(args: {
   }
 }
 
-async function handleXSearch(args: { query: string; max_results?: number }) {
+async function handleXSearch(args: {
+  query: string;
+  max_results?: number;
+  provider?: "x-search" | "xquik-search";
+}) {
   const start = Date.now();
   try {
     const result = await omniRouteFetch("/v1/search", {
@@ -675,7 +679,7 @@ async function handleXSearch(args: { query: string; max_results?: number }) {
         query: args.query,
         max_results: args.max_results ?? 5,
         search_type: "x",
-        provider: "x-search",
+        provider: args.provider ?? "x-search",
       }),
       signal: AbortSignal.timeout(120000),
     });

open-sse/services/accountSemaphore.ts

@@ -1,20 +1,40 @@
 /**
- * Account Semaphore
+ * Hierarchical in-memory concurrency admission.
  *
- * In-memory provider/account concurrency limiter keyed by provider and account.
- * Requests beyond the configured concurrency cap wait in a FIFO queue until a slot opens,
- * the gate is unblocked, or the queue timeout expires.
+ * `acquire()` preserves the account-semaphore API. `acquireMany()` admits one
+ * request only when every applicable global/provider/account gate has room.
  */
-
 export interface AccountSemaphoreKeyParts {
   provider: string;
   accountKey: string;
 }
 
-interface QueuedAcquire {
+export interface AcquireAccountSemaphoreOptions {
+  maxConcurrency?: number | null;
+  timeoutMs?: number;
+  signal?: AbortSignal | null;
+  maxQueueSize?: number;
+}
+
+export interface SemaphoreRequirement {
+  key: string;
+  maxConcurrency?: number | null;
+}
+
+export type AcquireManyOptions = Omit<AcquireAccountSemaphoreOptions, "maxConcurrency">;
+
+interface AcquireRequest {
+  keys: string[];
   resolve: (release: () => void) => void;
   reject: (error: Error) => void;
-  timer: ReturnType<typeof setTimeout>;
+  timer: ReturnType<typeof setTimeout> | null;
+  signal: AbortSignal | null;
+  abortListener: (() => void) | null;
+  settled: boolean;
+}
+
+interface QueuedAcquire {
+  request: AcquireRequest;
 }
 
 interface AccountGate {
@@ -25,13 +45,6 @@ interface AccountGate {
   cleanupTimer: ReturnType<typeof setTimeout> | null;
 }
 
-export interface AcquireAccountSemaphoreOptions {
-  maxConcurrency?: number | null;
-  timeoutMs?: number;
-  signal?: AbortSignal | null;
-  maxQueueSize?: number;
-}
-
 export interface AccountSemaphoreStatsEntry {
   running: number;
   queued: number;
@@ -41,12 +54,9 @@ export interface AccountSemaphoreStatsEntry {
 
 const DEFAULT_TIMEOUT_MS = 30_000;
 const DEFAULT_MAX_QUEUE_SIZE = 20;
-
 const gates = new Map<string, AccountGate>();
+const queuedRequests = new Set<AcquireRequest>();
 
-/**
- * Build the canonical account semaphore key.
- */
 export function buildAccountSemaphoreKey({
   provider,
   accountKey,
@@ -55,25 +65,23 @@ export function buildAccountSemaphoreKey({
 }
 
 function isBypassed(maxConcurrency?: number | null): boolean {
-  return maxConcurrency == null || maxConcurrency <= 0;
+  return maxConcurrency == null || !Number.isFinite(maxConcurrency) || maxConcurrency <= 0;
 }
 
 function createNoopReleaseFn(): () => void {
   let released = false;
-
   return () => {
     if (released) return;
     released = true;
   };
 }
 
-function ensureGate(semaphoreKey: string, maxConcurrency: number): AccountGate {
-  const existing = gates.get(semaphoreKey);
+function ensureGate(key: string, maxConcurrency: number): AccountGate {
+  const existing = gates.get(key);
   if (existing) {
     existing.maxConcurrency = maxConcurrency;
     return existing;
   }
-
   const created: AccountGate = {
     running: 0,
     maxConcurrency,
@@ -81,7 +89,7 @@ function ensureGate(semaphoreKey: string, maxConcurrency: number): AccountGate {
     blockedUntil: null,
     cleanupTimer: null,
   };
-  gates.set(semaphoreKey, created);
+  gates.set(key, created);
   return created;
 }
 
@@ -100,91 +108,101 @@ function clearCleanupTimer(gate: AccountGate): void {
   gate.cleanupTimer = null;
 }
 
-function cleanupGateIfIdle(semaphoreKey: string): void {
-  const gate = gates.get(semaphoreKey);
-  if (!gate) return;
-  if (gate.running > 0 || gate.queue.length > 0 || isBlocked(gate)) return;
+function cleanupGateIfIdle(key: string): void {
+  const gate = gates.get(key);
+  if (!gate || gate.running > 0 || gate.queue.length > 0 || isBlocked(gate)) return;
   clearCleanupTimer(gate);
-  gates.delete(semaphoreKey);
+  gates.delete(key);
 }
 
-function scheduleCleanup(semaphoreKey: string): void {
-  const gate = gates.get(semaphoreKey);
+function scheduleCleanup(key: string): void {
+  const gate = gates.get(key);
   if (!gate) return;
   clearCleanupTimer(gate);
-
   gate.cleanupTimer = setTimeout(() => {
     gate.cleanupTimer = null;
-    cleanupGateIfIdle(semaphoreKey);
+    cleanupGateIfIdle(key);
   }, 0);
-
   gate.cleanupTimer.unref?.();
 }
 
-function drainQueue(semaphoreKey: string): void {
-  const gate = gates.get(semaphoreKey);
-  if (!gate) return;
-
-  while (gate.queue.length > 0 && gate.running < gate.maxConcurrency && !isBlocked(gate)) {
-    const next = gate.queue.shift();
-    if (!next) break;
-    clearTimeout(next.timer);
-    gate.running++;
-    next.resolve(createReleaseFn(semaphoreKey));
-  }
-
-  if (gate.running === 0 && gate.queue.length === 0) {
-    scheduleCleanup(semaphoreKey);
-  }
-}
-
-function createReleaseFn(semaphoreKey: string): () => void {
-  let released = false;
-
-  return () => {
-    if (released) return;
-    released = true;
-
-    const gate = gates.get(semaphoreKey);
-    if (!gate) return;
-    if (gate.running > 0) {
-      gate.running--;
-    }
-
-    if (gate.queue.length > 0) {
-      drainQueue(semaphoreKey);
-      return;
-    }
-
-    scheduleCleanup(semaphoreKey);
-  };
-}
-
-function createSemaphoreTimeoutError(
-  semaphoreKey: string,
-  timeoutMs: number
-): Error & { code: string } {
-  const error = new Error(`Semaphore timeout after ${timeoutMs}ms for ${semaphoreKey}`) as Error & {
-    code: string;
-  };
-  error.code = "SEMAPHORE_TIMEOUT";
+function makeAbortError(signal: AbortSignal): Error {
+  if (signal.reason instanceof Error) return signal.reason;
+  const error = new Error(
+    typeof signal.reason === "string" ? signal.reason : "The operation was aborted"
+  );
+  error.name = "AbortError";
   return error;
 }
 
-function makeAbortError(signal: AbortSignal): Error {
-  const reason = signal.reason;
-  if (reason instanceof Error) return reason;
-  const err = new Error(typeof reason === "string" ? reason : "The operation was aborted");
-  err.name = "AbortError";
-  return err;
+function createSemaphoreError(code: string, message: string): Error & { code: string } {
+  const error = new Error(message) as Error & { code: string };
+  error.code = code;
+  return error;
+}
+
+function removeRequest(request: AcquireRequest): void {
+  queuedRequests.delete(request);
+  if (request.timer) clearTimeout(request.timer);
+  if (request.abortListener && request.signal) {
+    request.signal.removeEventListener("abort", request.abortListener);
+  }
+  for (const key of request.keys) {
+    const gate = gates.get(key);
+    if (!gate) continue;
+    const index = gate.queue.findIndex((queued) => queued.request === request);
+    if (index >= 0) gate.queue.splice(index, 1);
+    if (gate.running === 0 && gate.queue.length === 0) scheduleCleanup(key);
+  }
+}
+
+function canAcquire(request: AcquireRequest): boolean {
+  return request.keys.every((key) => {
+    const gate = gates.get(key);
+    return (
+      gate != null &&
+      !isBlocked(gate) &&
+      gate.running < gate.maxConcurrency &&
+      gate.queue[0]?.request === request
+    );
+  });
+}
+
+function createCompositeReleaseFn(keys: string[]): () => void {
+  let released = false;
+  return () => {
+    if (released) return;
+    released = true;
+    for (const key of keys) {
+      const gate = gates.get(key);
+      if (gate && gate.running > 0) gate.running--;
+    }
+    drainQueues();
+    for (const key of keys) {
+      const gate = gates.get(key);
+      if (gate && gate.running === 0 && gate.queue.length === 0) scheduleCleanup(key);
+    }
+  };
+}
+
+function drainQueues(): void {
+  let progressed = true;
+  while (progressed) {
+    progressed = false;
+    for (const request of queuedRequests) {
+      if (request.settled || !canAcquire(request)) continue;
+      request.settled = true;
+      removeRequest(request);
+      for (const key of request.keys) gates.get(key)!.running++;
+      request.resolve(createCompositeReleaseFn(request.keys));
+      progressed = true;
+      break;
+    }
+  }
 }
 
-/**
- * Acquire a slot for a provider/model/account tuple.
- * Returns an idempotent release function that is safe to call in finally blocks.
- */
 export function acquire(
-  semaphoreKey: string,
+  key: string,
   {
     maxConcurrency = null,
     timeoutMs = DEFAULT_TIMEOUT_MS,
@@ -192,146 +210,118 @@ export function acquire(
     maxQueueSize = DEFAULT_MAX_QUEUE_SIZE,
   }: AcquireAccountSemaphoreOptions = {}
 ): Promise<() => void> {
-  if (isBypassed(maxConcurrency)) {
-    return Promise.resolve(createNoopReleaseFn());
+  return acquireMany([{ key, maxConcurrency }], { timeoutMs, signal, maxQueueSize });
+}
+
+/**
+ * Acquire all enabled requirements as one FIFO reservation.
+ *
+ * Waiting never increments any gate, preventing a saturated child gate from
+ * holding capacity in a parent gate.
+ */
+export function acquireMany(
+  requirements: SemaphoreRequirement[],
+  {
+    timeoutMs = DEFAULT_TIMEOUT_MS,
+    signal = null,
+    maxQueueSize = DEFAULT_MAX_QUEUE_SIZE,
+  }: AcquireManyOptions = {}
+): Promise<() => void> {
+  const enabled = new Map<string, number>();
+  for (const requirement of requirements) {
+    if (isBypassed(requirement.maxConcurrency)) continue;
+    const limit = Math.trunc(requirement.maxConcurrency as number);
+    enabled.set(requirement.key, Math.min(enabled.get(requirement.key) ?? limit, limit));
+  }
+  if (enabled.size === 0) return Promise.resolve(createNoopReleaseFn());
+  if (signal?.aborted) return Promise.reject(makeAbortError(signal));
+
+  const keys = [...enabled.keys()].sort();
+  for (const key of keys) {
+    const gate = ensureGate(key, enabled.get(key)!);
+    clearCleanupTimer(gate);
+    if (maxQueueSize > 0 && gate.queue.length >= maxQueueSize) {
+      return Promise.reject(
+        createSemaphoreError(
+          "SEMAPHORE_QUEUE_FULL",
+          `Semaphore queue full (${maxQueueSize}) for ${key}`
+        )
+      );
+    }
   }
 
-  if (signal?.aborted) {
-    return Promise.reject(makeAbortError(signal));
-  }
-
-  // isBypassed() above already excluded null/<=0 — ensureGate requires a plain
-  // number, but a boolean-returning helper isn't a type predicate TS can narrow on.
-  const gate = ensureGate(semaphoreKey, maxConcurrency as number);
-  clearCleanupTimer(gate);
-
-  if (gate.running < gate.maxConcurrency && !isBlocked(gate)) {
-    gate.running++;
-    return Promise.resolve(createReleaseFn(semaphoreKey));
-  }
-
-  if (gate.queue.length >= maxQueueSize) {
-    const err = new Error(`Semaphore queue full (${maxQueueSize}) for ${semaphoreKey}`) as Error & {
-      code: string;
-    };
-    err.code = "SEMAPHORE_QUEUE_FULL";
-    return Promise.reject(err);
+  if (
+    keys.every((key) => {
+      const gate = gates.get(key)!;
+      return gate.queue.length === 0 && gate.running < gate.maxConcurrency && !isBlocked(gate);
+    })
+  ) {
+    for (const key of keys) gates.get(key)!.running++;
+    return Promise.resolve(createCompositeReleaseFn(keys));
   }
 
   return new Promise((resolve, reject) => {
-    let abortListener: (() => void) | null = null;
-
-    const cleanup = () => {
-      if (abortListener && signal) {
-        signal.removeEventListener("abort", abortListener);
-      }
+    const request: AcquireRequest = {
+      keys,
+      resolve,
+      reject,
+      timer: null,
+      signal,
+      abortListener: null,
+      settled: false,
     };
-
-    const timer = setTimeout(() => {
-      cleanup();
-      const nextGate = gates.get(semaphoreKey);
-      if (!nextGate) {
-        reject(createSemaphoreTimeoutError(semaphoreKey, timeoutMs));
-        return;
-      }
-
-      const queueIndex = nextGate.queue.findIndex((item) => item.timer === timer);
-      if (queueIndex !== -1) {
-        nextGate.queue.splice(queueIndex, 1);
-      }
-
-      if (nextGate.running === 0 && nextGate.queue.length === 0) {
-        scheduleCleanup(semaphoreKey);
-      }
-
-      reject(createSemaphoreTimeoutError(semaphoreKey, timeoutMs));
+    request.timer = setTimeout(() => {
+      if (request.settled) return;
+      request.settled = true;
+      removeRequest(request);
+      reject(
+        createSemaphoreError(
+          "SEMAPHORE_TIMEOUT",
+          `Semaphore timeout after ${timeoutMs}ms for ${keys.join(",")}`
+        )
+      );
+      drainQueues();
     }, timeoutMs);
-
-    timer.unref?.();
-
-    const queueItem: QueuedAcquire = {
-      resolve: (release) => {
-        cleanup();
-        resolve(release);
-      },
-      reject: (error) => {
-        cleanup();
-        reject(error);
-      },
-      timer,
-    };
-
-    gate.queue.push(queueItem);
-
+    request.timer.unref?.();
     if (signal) {
-      abortListener = () => {
-        cleanup();
-        clearTimeout(timer);
-
-        const nextGate = gates.get(semaphoreKey);
-        if (!nextGate) {
-          reject(makeAbortError(signal));
-          return;
-        }
-
-        const queueIndex = nextGate.queue.findIndex((item) => item.timer === timer);
-        if (queueIndex !== -1) {
-          nextGate.queue.splice(queueIndex, 1);
-        }
-
-        if (nextGate.running === 0 && nextGate.queue.length === 0) {
-          scheduleCleanup(semaphoreKey);
-        }
-
+      request.abortListener = () => {
+        if (request.settled) return;
+        request.settled = true;
+        removeRequest(request);
         reject(makeAbortError(signal));
+        drainQueues();
       };
-      if (signal.aborted) {
-        abortListener();
-      } else {
-        signal.addEventListener("abort", abortListener);
-      }
+      signal.addEventListener("abort", request.abortListener, { once: true });
     }
+    queuedRequests.add(request);
+    for (const key of keys) gates.get(key)!.queue.push({ request });
+    drainQueues();
   });
 }
 
-/**
- * Temporarily block new acquisitions for a key while allowing in-flight requests to finish.
- */
-export function markBlocked(semaphoreKey: string, cooldownMs: number): void {
-  const safeCooldownMs = Number.isFinite(cooldownMs) && cooldownMs > 0 ? cooldownMs : 0;
-  if (safeCooldownMs <= 0) {
-    const gate = gates.get(semaphoreKey);
-    if (!gate) return;
-    gate.blockedUntil = null;
-    drainQueue(semaphoreKey);
-    return;
-  }
-
-  const gate = gates.get(semaphoreKey) ?? ensureGate(semaphoreKey, 1);
+export function markBlocked(key: string, until: Date | string | number): void {
+  const untilMs =
+    until instanceof Date
+      ? until.getTime()
+      : typeof until === "number"
+        ? Date.now() + Math.max(0, until)
+        : new Date(until).getTime();
+  if (!Number.isFinite(untilMs) || untilMs <= Date.now()) return;
+  const gate = ensureGate(key, gates.get(key)?.maxConcurrency ?? 1);
   clearCleanupTimer(gate);
-  gate.blockedUntil = Date.now() + safeCooldownMs;
+  gate.blockedUntil = untilMs;
+}
 
-  const timer = setTimeout(() => {
-    const nextGate = gates.get(semaphoreKey);
-    if (!nextGate) return;
-    if (nextGate.blockedUntil && Date.now() >= nextGate.blockedUntil) {
-      nextGate.blockedUntil = null;
-      drainQueue(semaphoreKey);
-      if (nextGate.running === 0 && nextGate.queue.length === 0) {
-        scheduleCleanup(semaphoreKey);
-      }
-    }
-  }, safeCooldownMs + 50);
-
-  timer.unref?.();
+export function unblock(key: string): void {
+  const gate = gates.get(key);
+  if (!gate) return;
+  gate.blockedUntil = null;
+  drainQueues();
+  cleanupGateIfIdle(key);
 }
 
-/**
- * Return the current in-memory semaphore snapshot.
- */
 export function getStats(): Record<string, AccountSemaphoreStatsEntry> {
   const stats: Record<string, AccountSemaphoreStatsEntry> = {};
-
   for (const [key, gate] of gates) {
     stats[key] = {
       running: gate.running,
@@ -340,48 +330,50 @@ export function getStats(): Record<string, AccountSemaphoreStatsEntry> {
       blockedUntil: gate.blockedUntil ? new Date(gate.blockedUntil).toISOString() : null,
     };
   }
-
   return stats;
 }
 
-/**
- * Check if an account semaphore key is currently at or over its max concurrency limit.
- * Returns true if running >= maxConcurrency or blocked.
- */
 export function isAccountSemaphoreFull(
   provider: string,
   accountKey: string,
   maxConcurrency?: number | null
 ): boolean {
   if (isBypassed(maxConcurrency)) return false;
-  const key = buildAccountSemaphoreKey({ provider, accountKey });
-  const gate = gates.get(key);
+  const gate = gates.get(buildAccountSemaphoreKey({ provider, accountKey }));
   if (!gate) return false;
   const effectiveCap = maxConcurrency ?? gate.maxConcurrency;
-  if (isBypassed(effectiveCap)) return false;
-  return gate.running >= effectiveCap || isBlocked(gate);
+  return !isBypassed(effectiveCap) && (gate.running >= effectiveCap || isBlocked(gate));
 }
 
-/**
- * Reset a single key and reject queued waiters.
- */
-export function reset(semaphoreKey: string): void {
-  const gate = gates.get(semaphoreKey);
+export function reset(key: string): void {
+  const gate = gates.get(key);
   if (!gate) return;
-
   clearCleanupTimer(gate);
-  for (const entry of gate.queue) {
-    clearTimeout(entry.timer);
-    entry.reject(new Error("Semaphore reset"));
+  const error = createSemaphoreError("SEMAPHORE_RESET", `Semaphore reset for ${key}`);
+  const rejections: AcquireRequest[] = [];
+  for (const queued of [...gate.queue]) {
+    const request = queued.request;
+    if (request.settled) continue;
+    request.settled = true;
+    removeRequest(request);
+    rejections.push(request);
   }
-  gates.delete(semaphoreKey);
+  gates.delete(key);
+  for (const request of rejections) request.reject(error);
+  drainQueues();
 }
 
-/**
- * Reset all keys and reject queued waiters.
- */
 export function resetAll(): void {
-  for (const key of gates.keys()) {
-    reset(key);
+  const error = createSemaphoreError("SEMAPHORE_RESET", "Semaphore reset");
+  const rejections: AcquireRequest[] = [];
+  for (const request of [...queuedRequests]) {
+    if (request.settled) continue;
+    request.settled = true;
+    removeRequest(request);
+    rejections.push(request);
   }
+  for (const gate of gates.values()) clearCleanupTimer(gate);
+  gates.clear();
+  queuedRequests.clear();
+  for (const request of rejections) request.reject(error);
 }

open-sse/services/autoCombo/strictZeroCostFilter.ts

@@ -50,6 +50,7 @@
  */
 import {
   FREE_MODEL_BUDGETS,
+  grantsFreeAccess,
   type FreeModelBudget,
 } from "@omniroute/open-sse/config/freeModelCatalog.ts";
 import { SYNTHETIC_NOAUTH_CONNECTION_ID } from "./resilienceCandidateFilter";
@@ -184,7 +185,7 @@ export function evaluateCandidateConnections(
     // entries today, so it will correctly exclude).
     if (isGenuineNoAuthCandidate) return [SYNTHETIC_NOAUTH_CONNECTION_ID];
   }
-  if (budgetEntry.freeType === "discontinued") return [];
+  if (!grantsFreeAccess(budgetEntry.freeType)) return [];
   if (isGenuineNoAuthCandidate) return []; // no-auth path but a non-keyless catalog entry: contradictory metadata, fail closed
 
   // Every remaining freeType (recurring-*, one-time-initial, a keyless entry

open-sse/services/combo/quotaScoring.ts

@@ -261,7 +261,7 @@ function getWindowsMapQuotaWindow(
   );
 }
 
-function resolveQuotaWindowByName(
+export function resolveQuotaWindowByName(
   quota: unknown,
   windowName: ResetWindowName
 ): QuotaWindowSnapshot | null {
@@ -321,8 +321,8 @@ export function scoreResetAwareQuota(
   if (quota.limitReached === true) return { score: -Infinity };
 
   const overallPercentUsed = clamp01(finiteNumberOrNull(quota.percentUsed) ?? 0.5);
-  const sessionWindow = getQuotaWindow(quota, "window5h");
-  const weeklyWindow = getQuotaWindow(quota, "window7d") || getQuotaWindow(quota, "windowWeekly");
+  const sessionWindow = resolveQuotaWindowByName(quota, "session");
+  const weeklyWindow = resolveQuotaWindowByName(quota, "weekly");
   const sessionRemaining = clamp01(1 - (sessionWindow?.percentUsed ?? overallPercentUsed));
   const weeklyRemaining = clamp01(1 - (weeklyWindow?.percentUsed ?? overallPercentUsed));
   const sessionScore = scoreQuotaWindow(

open-sse/services/compression/eval/executorModelClient.ts

@@ -16,9 +16,10 @@ export function createExecutorModelClient(
   credentials: ProviderCredentials,
   costPerKTokenOut?: number
 ): ModelClient {
-  const executor = getExecutor(provider);
   return {
     async complete(model: string, messages: ChatTurn[]): Promise<ModelCallResult> {
+      // #11220: getExecutor is async (lazy registry) — resolve per call.
+      const executor = await getExecutor(provider);
       const body = { model, messages, stream: false };
       const input: ExecuteInput = {
         model,

open-sse/services/genericQuotaFetcher.ts

@@ -150,16 +150,65 @@ export function convertUsageToQuotaInfo(usage: unknown): QuotaInfo | null {
 
   if (Object.keys(windows).length === 0) return null;
 
+  const normalized = normalizeQuotaWindows(windows);
+  Object.assign(windows, normalized);
+
   return {
     used: 0,
     total: 0,
     percentUsed: worstPercent,
     resetAt: worstResetAt,
     windows,
+    ...normalized,
     limitReached: worstPercent >= 1 - 1e-9,
   };
 }
 
+/**
+ * Map provider-native window keys to canonical structural windows so that
+ * reset-aware / reset-window scoring works without knowing every provider's
+ * naming convention.
+ *
+ *   - Claude: "session (5h)" → window5h, "weekly (7d)" → window7d
+ *   - Antigravity: worst per-model quota → window5h; worst *_weekly quota → window7d
+ */
+function normalizeQuotaWindows(
+  windows: Record<string, { percentUsed: number; resetAt: string | null }>
+): Record<string, { percentUsed: number; resetAt: string | null }> {
+  const normalized: Record<string, { percentUsed: number; resetAt: string | null }> = {};
+
+  // Claude-style explicit time windows.
+  if (windows["session (5h)"] && !normalized.window5h) {
+    normalized.window5h = windows["session (5h)"];
+  }
+  if (windows["weekly (7d)"] && !normalized.window7d) {
+    normalized.window7d = windows["weekly (7d)"];
+  }
+
+  // Antigravity-style per-model 5h windows: pick the worst (most used) model quota.
+  const modelWindows = Object.entries(windows).filter(
+    ([key]) =>
+      key !== "credits" &&
+      !key.endsWith("_weekly") &&
+      !key.startsWith("window") &&
+      !key.includes("(5h)") &&
+      !key.includes("(7d)")
+  );
+  if (modelWindows.length > 0 && !normalized.window5h) {
+    const worst = modelWindows.reduce((a, b) => (a[1].percentUsed > b[1].percentUsed ? a : b));
+    normalized.window5h = worst[1];
+  }
+
+  // Antigravity-style weekly family buckets: pick the worst *_weekly quota.
+  const weeklyWindows = Object.entries(windows).filter(([key]) => key.endsWith("_weekly"));
+  if (weeklyWindows.length > 0 && !normalized.window7d) {
+    const worst = weeklyWindows.reduce((a, b) => (a[1].percentUsed > b[1].percentUsed ? a : b));
+    normalized.window7d = worst[1];
+  }
+
+  return normalized;
+}
+
 /**
  * Fetch quota for a connection by delegating to the appropriate
  * provider-specific usage fetcher and reshaping its output into the

open-sse/services/quotaPreflight.ts

@@ -48,6 +48,16 @@ export interface QuotaInfo {
    * (e.g. "session", "weekly", "monthly").
    */
   windows?: Record<string, QuotaWindowInfo>;
+  /**
+   * Structural, canonical window snapshots used by reset-aware / reset-window
+   * scoring. Providers that expose time-based windows (5h, weekly, monthly)
+   * populate these in addition to the provider-native `windows` map so the
+   * scorer does not need to know every provider's key naming convention.
+   */
+  window5h?: QuotaWindowInfo;
+  window7d?: QuotaWindowInfo;
+  windowWeekly?: QuotaWindowInfo;
+  windowMonthly?: QuotaWindowInfo;
   /** True when the upstream usage endpoint explicitly reports exhausted quota. */
   limitReached?: boolean;
 }

open-sse/services/volcengineConsoleAutoLogin.ts

@@ -96,13 +96,6 @@ const ARK_CONSOLE_URL =
 /** Cookie names required for a valid console session (mirrors tokenExtractionConfig) */
 const REQUIRED_COOKIES = ["digest", "AccountID", "csrfToken", "userInfo"] as const;
 
-/** Exact-domain match for session cookies — substring checks would also accept
- * look-alike hosts (e.g. `volcengine.com.evil.test`). Playwright may report the
- * domain with or without a leading dot. */
-function isVolcengineCookieDomain(domain: string): boolean {
-  return domain === "volcengine.com" || domain.endsWith(".volcengine.com");
-}
-
 const DEFAULT_SESSION_TIMEOUT = 300_000;
 const SUBMIT_COOKIE_TIMEOUT = 90_000;
 const CAPTURE_POLL_INTERVAL = 1_000;
@@ -625,7 +618,7 @@ export class VolcengineConsoleAutoLoginService {
       for (const cookie of cookies as Array<{ name: string; domain: string; value: string }>) {
         if (
           REQUIRED_COOKIES.includes(cookie.name as (typeof REQUIRED_COOKIES)[number]) &&
-          isVolcengineCookieDomain(cookie.domain)
+          cookie.domain.includes("volcengine.com")
         ) {
           credentials[cookie.name] = cookie.value;
         }
@@ -770,7 +763,7 @@ export class VolcengineConsoleAutoLoginService {
           domain: string;
         }>;
         const present = REQUIRED_COOKIES.filter((name) =>
-          cookies.some((c) => c.name === name && isVolcengineCookieDomain(c.domain))
+          cookies.some((c) => c.name === name && c.domain.includes("volcengine.com"))
         );
         parts.push(
           `cookies=[${present.join(",") || "none of digest/AccountID/csrfToken/userInfo"}]`

package.json

@@ -261,7 +261,7 @@
     "alibaba:sync-allowlist": "node --import tsx/esm scripts/ops/sync-alibaba-allowlist.mjs"
   },
   "dependencies": {
-    "@aws-sdk/client-bedrock-runtime": "^3.1112.0",
+    "@aws-sdk/client-bedrock-runtime": "^3.1116.0",
     "@dnd-kit/core": "^6.3.1",
     "@dnd-kit/sortable": "^10.0.0",
     "@dnd-kit/utilities": "^3.2.2",
@@ -280,29 +280,29 @@
     "commander": "^15.0.0",
     "cron-parser": "^5.10.0",
     "csv-stringify": "^6.8.3",
-    "dompurify": "^3.4.13",
+    "dompurify": "^3.4.14",
     "express": "^5.2.1",
     "fetch-socks": "^1.3.3",
     "fflate": "^0.8.3",
-    "fumadocs-core": "^16.14.4",
-    "fumadocs-ui": "^16.14.4",
+    "fumadocs-core": "^16.15.1",
+    "fumadocs-ui": "^16.15.1",
     "http-proxy-middleware": "^4.0.0",
     "https-proxy-agent": "^9.0.0",
     "ink": "^7.0.3",
     "ink-spinner": "^5.0.0",
     "ink-text-input": "^6.0.0",
     "ioredis": "^5.10.1",
-    "jose": "^6.2.9",
+    "jose": "^6.2.10",
     "js-yaml": "^5.3.0",
     "jsonc-parser": "^3.3.1",
     "lowdb": "^7.0.1",
-    "lucide-react": "^1.21.0",
-    "marked": "^18.0.9",
+    "lucide-react": "^1.33.0",
+    "marked": "^18.0.10",
     "marked-terminal": "^7.3.0",
     "material-symbols": "^0.46.0",
-    "mermaid": "^11.15.0",
+    "mermaid": "^11.17.0",
     "monaco-editor": "^0.56.0",
-    "next": "16.3.1",
+    "next": "16.3.2",
     "next-intl": "^4.13.7",
     "next-themes": "^0.4.6",
     "node-machine-id": "^1.1.12",
@@ -332,7 +332,7 @@
     "turndown-plugin-gfm": "1.0.2",
     "undici": "^8.10.0",
     "update-notifier": "^7.3.1",
-    "uuid": "^14.0.0",
+    "uuid": "^14.0.2",
     "ws": "^8.21.3",
     "xxhash-wasm": "^1.1.0",
     "yazl": "^3.3.1",
@@ -348,11 +348,12 @@
     "onnxruntime-node": "1.24.3",
     "sqlite-vec": "^0.1.9",
     "tls-client-node": "^0.2.0",
-    "wreq-js": "^3.0.0"
+    "wreq-js": "^3.1.0"
   },
   "devDependencies": {
     "@axe-core/playwright": "^4.13.0",
     "@cyclonedx/cyclonedx-npm": "6.0.1",
+    "@eslint/compat": "^2.1.0",
     "@playwright/test": "^1.62.1",
     "@size-limit/file": "^13.0.3",
     "@stryker-mutator/core": "^10.0.0",
@@ -363,24 +364,25 @@
     "@testing-library/react": "^16.3.2",
     "@testing-library/user-event": "^14.6.6",
     "@types/better-sqlite3": "^9.6.0",
-    "@types/bun": "latest",
+    "@types/bun": "1.4.0",
     "@types/node": "^26.2.0",
     "@types/react": "^19.2.18",
     "@types/react-dom": "^19.2.4",
     "@types/safe-regex": "^1.1.6",
     "@types/ws": "^8.18.0",
-    "@vitejs/plugin-react": "^6.0.5",
-    "bun": "1.3.14",
+    "@vitejs/plugin-react": "^6.1.0",
+    "bun": "1.4.0",
     "c8": "^12.0.0",
     "concurrently": "^10.0.5",
     "cross-env": "^10.1.0",
     "ctrf": "^0.3.0",
     "dpdm": "^4.3.0",
-    "eslint": "^9.39.4",
-    "eslint-config-next": "16.3.1",
+    "eslint": "^10.9.0",
+    "eslint-config-next": "16.3.2",
     "eslint-plugin-sonarjs": "^4.1.0",
+    "espree": "^11.2.0",
     "fast-check": "^4.8.0",
-    "fumadocs-mdx": "^15.2.3",
+    "fumadocs-mdx": "^15.3.1",
     "glob": "^13.0.6",
     "httpyac": "^6.16.7",
     "husky": "^9.1.7",
@@ -392,7 +394,7 @@
     "lint-staged": "^17.3.0",
     "lockfile-lint": "^5.0.1",
     "node-loader": "^2.1.0",
-    "opencode-ai": "1.18.18",
+    "opencode-ai": "1.18.21",
     "playwright-ctrf-json-reporter": "^0.0.29",
     "prettier": "^3.9.6",
     "promptfoo": "^0.122.0",
@@ -401,7 +403,7 @@
     "type-coverage": "^2.30.1",
     "typescript": "^6.0.3",
     "typescript-eslint": "^8.67.0",
-    "vitest": "^4.1.7",
+    "vitest": "^4.1.11",
     "wait-on": "^9.1.0",
     "wtfnode": "^0.10.1"
   },
@@ -435,12 +437,13 @@
   },
   "overrides": {
     "onnxruntime-node": "1.24.3",
+    "eslint-plugin-react-hooks": "7.0.1",
     "fast-xml-parser": "^5.10.1",
     "sharp": "^0.35.3",
     "postcss": "^8.5.18",
     "ip-address": "^10.3.1",
     "qs": "^6.15.2",
-    "uuid": "^14.0.0",
+    "uuid": "^14.0.2",
     "form-data": "^4.0.6",
     "vite": "^8.0.16",
     "protobufjs": "^7.6.5",
@@ -468,10 +471,6 @@
     },
     "socket.io-parser": "^4.2.7",
     "tar": "^7.5.21",
-    "brace-expansion": "^5.0.9",
-    "minimatch": {
-      "brace-expansion": "^1.1.18"
-    },
     "libxmljs2": {
       "minimatch": {
         "brace-expansion": "^2.1.4"
@@ -496,7 +495,7 @@
     },
     "nanoid": "^3.3.17",
     "monaco-editor": {
-      "dompurify": "^3.4.13"
+      "dompurify": "^3.4.14"
     }
   }
 }

scripts/build/afterpack-copy-node-modules.mjs

@@ -0,0 +1,34 @@
+import { cpSync, existsSync, readdirSync, rmSync } from "node:fs";
+import { join } from "node:path";
+
+// electron-builder >=26 injects an "!**/node_modules/**" ignore into every
+// extraResources/extraFiles pattern list (app-builder-lib/out/fileMatcher.js),
+// and that ignore cannot be overridden by any later positive filter pattern —
+// verified empirically with a minimal fixture on 26.15.3. The standalone
+// server resolves better-sqlite3 (and other runtime deps) from the *primary*
+// node_modules at resources/app/node_modules (see
+// prepare-electron-standalone.mjs: "Verify better-sqlite3 Node-API prebuilds in
+// the primary node_modules"), so without this hook the packaged desktop app silently loses its native
+// SQLite driver and falls back to sql.js — the exact regression guarded by
+// issue #7592's cold-restart smoke check.
+export default async function afterPack(context) {
+  const stagingNodeModules = join(
+    context.packager.projectDir,
+    "..",
+    ".build",
+    "electron-standalone",
+    "node_modules"
+  );
+  const destNodeModules = join(context.appOutDir, "resources", "app", "node_modules");
+
+  if (!existsSync(stagingNodeModules)) {
+    console.warn(`[afterpack] no staged node_modules at ${stagingNodeModules} — skipping restore`);
+    return;
+  }
+
+  rmSync(destNodeModules, { recursive: true, force: true });
+  cpSync(stagingNodeModules, destNodeModules, { recursive: true });
+  console.log(
+    `[afterpack] restored ${readdirSync(destNodeModules).length} runtime module(s) into resources/app/node_modules`
+  );
+}

scripts/build/optionalPackStaging.mjs

@@ -89,16 +89,20 @@ function moveTree(src, dest) {
   return { removedFiles: files, removedBytes: bytes };
 }
 
-function tarPack(packOutDir, tarballPath) {
+export function tarPack(packOutDir, tarballPath) {
   // bsdtar ships with macOS, Linux images, and Windows runners (System32\tar.exe).
+  // GNU tar (common in Git-for-Windows environments) treats `C:\...` in `-f` as a
+  // remote rsh target ("Cannot connect to C:"), so always pass a bare filename
+  // and point cwd at the tarball directory instead.
   const result = spawnSync(
     process.platform === "win32" ? "tar.exe" : "tar",
-    ["-czf", tarballPath, "-C", packOutDir, "node_modules"],
-    { stdio: "pipe" }
+    ["-czf", path.basename(tarballPath), "-C", packOutDir, "node_modules"],
+    { stdio: "pipe", cwd: path.dirname(tarballPath) }
   );
   if (result.status !== 0) {
     throw new Error(
-      `optional-pack tar failed for ${path.basename(tarballPath)} (exit ${result.status})`
+      `optional-pack tar failed for ${path.basename(tarballPath)} (exit ${result.status})` +
+        `: ${(result.stderr || result.stdout || "").toString().slice(-600)}`
     );
   }
 }

scripts/dev/smoke-electron-packaged.mjs

@@ -360,7 +360,7 @@ function isInsideDir(parentDir, candidateDir) {
   return candidate === parent || candidate.startsWith(parent + sep);
 }
 
-async function ensureSmokeEnvDirs(smokeEnv, dataDir) {
+export async function ensureSmokeEnvDirs(smokeEnv, dataDir) {
   const dirNames = [
     "DATA_DIR",
     "HOME",
@@ -388,6 +388,16 @@ async function ensureSmokeEnvDirs(smokeEnv, dataDir) {
       dirs.push(join(smokeEnv.APPDATA, subdir));
     }
   }
+  // Electron resolves the Roaming profile from %USERPROFILE%\AppData\Roaming
+  // (USERPROFILE takes precedence over the APPDATA env var) and the path
+  // service throws — rather than creates — when that directory is missing,
+  // which makes requestSingleInstanceLock() return false and the app exit(0)
+  // before app.whenReady(). Pre-create the derived tree as well.
+  if (platform() === "win32" && smokeEnv.USERPROFILE) {
+    for (const subdir of ["omniroute-desktop", "OmniRoute", "omniroute"]) {
+      dirs.push(join(smokeEnv.USERPROFILE, "AppData", "Roaming", subdir));
+    }
+  }
 
   await Promise.all(dirs.map((dir) => mkdir(dir, { recursive: true })));
 }
@@ -506,7 +516,14 @@ async function waitForReady({ logs, smokeUrl, timeoutMs, settleMs, exitState })
  * by the single-launch path and the cold-restart (two-launch) path so both
  * exercise identical spawn/readiness/shutdown behavior.
  */
-async function launchAndCollectLogs({ appExecutable, smokeUrl, dataDir, timeoutMs, settleMs, streamLogs }) {
+async function launchAndCollectLogs({
+  appExecutable,
+  smokeUrl,
+  dataDir,
+  timeoutMs,
+  settleMs,
+  streamLogs,
+}) {
   const smokeEnv = buildSmokeEnv({ dataDir });
   await assertPortIsFree(smokeUrl);
   await ensureSmokeEnvDirs(smokeEnv, dataDir);
@@ -568,7 +585,14 @@ async function main() {
     !process.env.ELECTRON_SMOKE_DATA_DIR && process.env.ELECTRON_SMOKE_KEEP_DATA !== "1";
 
   try {
-    await launchAndCollectLogs({ appExecutable, smokeUrl, dataDir, timeoutMs, settleMs, streamLogs });
+    await launchAndCollectLogs({
+      appExecutable,
+      smokeUrl,
+      dataDir,
+      timeoutMs,
+      settleMs,
+      streamLogs,
+    });
 
     if (!coldRestart) return;