The previous merge used sourceFormat check which broke Cursor when it
sends openai-responses format requests. Now uses user-agent detection:
- Droid CLI (user-agent contains 'droid' or 'codex-cli') → passthrough
- Other clients (Cursor, etc.) → translate to Chat Completions format
This fixes the API translation for both clients.
Co-authored-by: Hellodebasishsahu <itsyourboydevil@gmail.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
- Support multiple OpenAI-compatible providers with custom prefix/baseUrl
- Add provider nodes CRUD (create/read/update/delete)
- URL building: baseUrl + /chat/completions or /responses
- Model import from /models endpoint
- API key validation via /models
- Usage type safety across all translators
- OAuth token auto-refresh for expired tokens
- Force streaming for Codex/OpenAI models to fix non-streaming bug
- Strip unsupported params (user, metadata, stream_options, prompt_cache_retention)
- Force response translation from openai-responses to openai format
- Migrate middleware.js to proxy.js for Next.js 16
- Use webpack explicitly in dev/build scripts
- Updated Codex User-Agent
Adds local token usage tracking for all AI providers. Usage data is
captured during stream processing and stored in a local database.
Includes a new Usage tab in the Providers dashboard to visualize
historical token consumption.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
