- Issue #199: proxy_ prefix on tool names is now automatically disabled
when routing to non-Claude backends (Gemini, Compatible, etc.).
Prevents tool name mismatches in OpenCode, Cursor, and other clients.
- Issue #198: Added customRpm/customTpm support per provider connection.
Users can configure custom rate limits via connection settings,
overriding the default auto-learned limits from response headers.
- Add type annotations to all providerModels helper functions
- Introduce LegacyProvider interface in providerRegistry
- Refactor translator system to use self-registering module pattern
with bootstrapTranslatorRegistry() and per-file imports
- Simplify translator/index.ts by delegating to modular translators
- Remove hardcoded Gemini OAuth client secret for security
Introduce `normalizeQuotaResponse` and `normalizeCombosResponse` helpers
to handle varying API response shapes (array vs wrapped object)
consistently across MCP server and advanced tools. Add optional `meta`
field to checkQuotaOutput schema and update sourceEndpoints to reflect
current API routes.
- Add unit tests for 8 MCP advanced tool handlers (Phase 3)
- Migrate test files from JavaScript to TypeScript (.ts/.tsx)
- Restructure file paths from app/ to src/app/ across all tests
- Refactor route assertions into reusable assertRouteMethods helper
- Add tests for new API routes (compliance, audit-log, evals/[suiteId])
- Update barrel export tests to use consolidated assertion pattern
Register 8 new advanced MCP tools (simulate_route, set_budget_guard,
set_resilience_profile, test_combo, get_provider_metrics,
best_combo_for_task, explain_route, get_session_snapshot) with their
handler implementations. Add A2A smart routing skill that routes
prompts through the OmniRoute pipeline with routing explanation,
cost envelope, and resilience trace metadata.
Add stdio-based MCP server support to OmniRoute CLI, enabling AI agents
in VS Code, Cursor, Claude Desktop, and Copilot to interact with
OmniRoute tools (health, combos, quota, routing). Update help text,
gitignore vscode-extension subproject, and include MCP/A2A strategy report.
Passthrough mode now detects Responses SSE payloads (parsed.type starts
with 'response.') and skips Chat Completions-specific sanitization:
- sanitizeStreamingChunk() only runs on Chat Completions payloads
- fixInvalidId() and hasValuableContent() checks skipped for Responses
- Usage extraction still runs for both formats
- Content length tracking adapted for Responses delta format
This prevents potential stream corruption when Responses SSE data
triggers idFixed or other Chat Completions-specific rewrite conditions.
- Added warning logs when generateProjectId() is used as fallback
in antigravity.ts and openai-to-gemini.ts (3 locations), helping
admins diagnose 404 errors from fake GCP project IDs (#176)
- Added blackbox.ai as new OpenAI-compatible provider with 6 models
(GPT-4o, Gemini 2.5 Flash, Claude Sonnet 4, DeepSeek V3,
Blackbox AI, Blackbox AI Pro) and provider logo (#175)
- Added tool_calls[].id and type:'function' to argument delta chunks
so OpenAI-compatible clients can associate argument fragments with
the correct tool call
- Changed finish_reason from hardcoded 'stop' to 'tool_calls' when
tool calls were emitted (in flush, response.completed, and final chunk)
- Fixes state desync in agentic clients (OpenCode, Claude Code, etc.)
where the assistant thinks tools already ran
- Added cleanJSONSchemaForAntigravity() to openaiToGeminiBase() tool conversion
- Both OpenAI-format and Claude-format tool parameters are now sanitized
- Also sanitized response_format json_schema using the same function
- Removes unsupported JSON Schema keywords (additionalProperties, $schema, etc.)
- All Gemini paths (standard, CLI, Antigravity) now consistently sanitize schemas
- syncAndVerify now returns cloudUrl in API response for frontend to use
- EndpointPageClient uses dynamic cloudBaseUrl state instead of relying on env var
- Falls back gracefully when NEXT_PUBLIC_CLOUD_URL is not set (Docker deployments)
- Fixed setInterval in accountFallback.ts global scope for Cloudflare Workers compat
- 24 unit tests for parseModelFromRegistry, getAllModelsFromRegistry, buildAuthHeaders
- Integration tests for video/music registries
- Media Playground dashboard page (Image/Video/Music tabs with model selector)
- Sidebar navigation entry for Media page
- i18n translations (EN + PT-BR)
- Fix Record<string, any> → Record<string, unknown> in registryUtils.ts
- Update /resolve-issues workflow to wait for user validation before commit/release
- Add path traversal validation for ElevenLabs voice_id and HuggingFace
model_id URL concatenation (prevents SSRF via ../ sequences)
- Add saveCallLog usage tracking to video and music handlers for
consistent analytics with imageGeneration.ts
- Extract shared upstreamErrorResponse() and audioStreamResponse()
helpers to reduce error handling duplication in audioSpeech.ts
- Extract shared upstreamErrorResponse() and isValidPathSegment()
helpers in audioTranscription.ts
- Add explicit format: "openai" to qwen TTS and STT provider entries
- Remove unused modelId parameter from handleCoquiSpeech and
handleTortoiseSpeech
- Filter cloud video/music providers by active status in models route
(local providers with authType: "none" always listed)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add 6 TTS providers (Nvidia NIM, ElevenLabs, HuggingFace, Coqui, Tortoise,
Qwen3), 3 STT providers (Nvidia NIM, HuggingFace, Qwen3), 2 local image
providers (SD WebUI, ComfyUI), and two new modalities — Text-to-Video
(/v1/videos/generations) and Text-to-Music (/v1/music/generations).
Key design decisions:
- Format-based unified providers: local providers grouped by API format
(comfyui, sdwebui, coqui, tortoise, openai-compatible) with configurable
base URLs and expandable model lists
- Cloud providers kept separate (unique auth and API shapes)
- Local providers use authType: "none" — credential checks bypassed at both
route and handler level
- Shared ComfyUI client (comfyuiClient.ts) reused across image/video/music
- Shared registry utilities (registryUtils.ts) for model parsing and listing
- Qwen3 TTS/ASR use format: "openai" — no custom handler needed
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add collapsible '16 Real Pain Points' section to all 30 READMEs
- Fix 5 files bypassing dataPaths.ts with hardcoded os.homedir() (closes#156)
- Add per-provider User-Agent env var overrides in base executor (closes#155)
- Sync .env and .env.example with 9 provider UA defaults
- Update CHANGELOG.md for v1.7.0
The URL parser silently strips default ports (80 for HTTP, 443 for HTTPS)
when constructing URL objects. This caused proxy connections to use port
8080 instead of the user-specified port 80, resulting in connection
timeouts. Fix by extracting the port from the raw URL string before
parsing and building the normalized URL manually to avoid the serializer
stripping it.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- URL-encode proxy credentials to handle special characters in passwords
- Decode URL-encoded credentials during legacy proxy migration
- Fix HTTPS proxy default port (443 instead of 8080) in frontend and migration
- Add dispatcher cache invalidation when proxy config changes
- Cast proxy port to number for SQLite INTEGER column in proxy logger
- Fix redundant .replace("//", "") in migration protocol parsing
- Copy postinstall script in Dockerfile before npm install
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Added ApiKeyMetadata interface to replace 'any' types in apiKeyPolicy.ts
- Added error logging in catch blocks for getApiKeyMetadata() and checkBudget()
- Fixed claude-sonnet-4-6-thinking → claude-sonnet-4-6 mismatch in usage.ts importantModels
Follow-up fixes for merged PRs #131 and #128
Next.js inlines process.env at build time, causing clientSecret defaults to
be baked as empty strings. Desktop/CLI OAuth secrets are not confidential
per Google's documentation.
Fixes: client_secret is missing error on VPS deployment
- Add modelsUrl field to RegistryEntry interface for custom models endpoints
- Configure kilocode provider with dedicated models URL (https://api.kilo.ai/api/openrouter/models)
- Expand kilocode model list from 8 to 26 models including free options:
- openrouter/free (Free Models Router)
- stepfun/step-3.5-flash:free
- arcee-ai/trinity-large-preview:free
- Additional Qwen, DeepSeek, Llama, Mistral, Grok, and Kimi models
- Update validateOpenAILikeProvider to accept custom modelsUrl parameter
- Fix models URL derivation for base URLs ending with /chat/completions
- Add kilocode config to PROVIDER_MODELS_CONFIG
- Fix API_KEY_SECRET undefined crash with deterministic fallback
- Add code_review_rate_limit parsing for Team/Plus/Pro plan support
- Move feature-81 doc to completed/
- Bump version to v1.1.1
- fix(chatCore): stream defaults to false per OpenAI spec (body.stream === true)
closes#89
- fix(models): custom OpenAI-compatible providers now appear in /v1/models
Added raw providerId check against activeAliases
closes#90
- fix(OAuthModal): Google OAuth providers force localhost redirect URI
Targeted amber warning + redirect_uri_mismatch error guidance for remote deployments
closes#91
- docs: Add 'OAuth em Servidor Remoto' tutorial in README with 7-step Google Cloud guide
- docs: .env.example prominent warning block for Google OAuth remote setup
- chore: bump version 1.0.6 → 1.0.7
- docs: CHANGELOG entry for v1.0.7
OmniRoute is an intelligent API gateway that unifies 20+ AI providers behind a single
OpenAI-compatible endpoint. Features include intelligent routing with 6 strategies,
multi-format translation (OpenAI/Claude/Gemini/Responses API), circuit breakers,
semantic caching, combo fallback chains, real-time health monitoring, and a full
dashboard with provider management, analytics, and CLI tool integration.
Key highlights:
- 20+ providers (Claude Code, Codex, Gemini CLI, GitHub Copilot, iFlow, Qwen, Kiro, etc.)
- 6 routing strategies (Fill First, Round Robin, P2C, Random, Least Used, Cost Optimized)
- Export/Import database backup with full archive support
- Translator Playground with 4 modes (Playground, Chat Tester, Test Bench, Live Monitor)
- 100% TypeScript across src/ and open-sse/
- Docker support with multi-stage builds
- Comprehensive documentation and 9 dashboard screenshots
