MCP Jina Supabase RAG

# Quick Start Guide Get up and running with mcp-jina-supabase-rag in under 10 minutes. ## Prerequisites - [ ] Python 3.12+ - [ ] Supabase account (free tier works) - [ ] OpenAI API key - [ ] Jina AI API key (optional but recommended) ## Step 1: Clone and Install ```bash cd ~/repos git clone <your-repo-url> mcp-jina-supabase-rag cd mcp-jina-supabase-rag # Install dependencies uv venv source .venv/bin/activate # or .venv\Scripts\activate on Windows uv pip install -e . crawl4ai-setup ``` ## Step 2: Set Up Supabase 1. Go to [https://supabase.com](https://supabase.com) and create a project 2. In SQL Editor, run the schema: ```bash # Copy the contents of supabase_schema.sql and run in Supabase SQL Editor ``` 3. Get your credentials: - Project URL: Settings → API → Project URL - Service Key: Settings → API → service_role key (keep secret!) ## Step 3: Configure Environment ```bash cp .env.example .env nano .env # or your favorite editor ``` Fill in: ```bash # Required OPENAI_API_KEY=sk-... SUPABASE_URL=https://your-project.supabase.co SUPABASE_SERVICE_KEY=eyJ... # Optional (recommended for faster extraction) JINA_API_KEY=jina_... ``` Get API keys: - OpenAI: [https://platform.openai.com/api-keys](https://platform.openai.com/api-keys) - Jina AI: [https://jina.ai](https://jina.ai) (free tier available) ## Step 4: Run the Server ```bash # Start the MCP server python src/main.py # You should see: # Starting MCP server on 0.0.0.0:8052 with sse transport # MCP server initialized successfully ``` ## Step 5: Connect to Claude Code ```bash # Add the MCP server claude mcp add --transport sse jina-supabase http://localhost:8052/sse # Verify it's connected claude mcp list ``` ## Step 6: Create Slash Command ```bash # Create the command directory if it doesn't exist mkdir -p ~/.claude/commands # Copy the command template cp SLASH_COMMAND.md ~/.claude/commands/jina.md # Edit if needed nano ~/.claude/commands/jina.md ``` ## Step 7: Test It Out! In Claude Code, try: ```bash # Index a small documentation site /jina https://docs.anthropic.com/claude/docs/intro-to-claude anthropic-test # Wait for it to complete, then search "Search the anthropic-test project for information about API keys" # List all projects "List all indexed projects" ``` ## Troubleshooting ### Server won't start **Check logs:** ```bash python src/main.py # Look for error messages ``` **Common issues:** - Missing environment variables → Check `.env` file - Port already in use → Change `PORT` in `.env` - Missing dependencies → Run `uv pip install -e .` ### Claude Code can't connect ```bash # Check if server is running curl http://localhost:8052/sse # Should return SSE connection info # Restart Claude Code claude mcp restart jina-supabase ``` ### Crawling is slow - **Use sitemap**: `/jina https://docs.example.com/* project sitemap` - **Get Jina API key**: Jina extraction is 3-5x faster than Crawl4AI - **Reduce max_urls**: Edit `URLDiscoverer` max_urls parameter ### Indexing fails **Check Supabase:** 1. Verify schema is installed (check "documents" table exists) 2. Check service key has correct permissions 3. Look at Supabase logs for errors **Check embeddings:** ```bash # Test OpenAI API key python -c "import openai; openai.api_key='YOUR_KEY'; print(openai.models.list())" ``` ## Next Steps 1. **Index your favorite docs**: Try indexing documentation you use frequently 2. **Create search command**: Add a `/search` command for quick queries 3. **Adjust chunking**: Modify `CHUNK_SIZE` in `.env` for your use case 4. **Set up multiple projects**: Index multiple doc sites for comprehensive RAG ## Configuration Options Edit `.env` to customize: ```bash # Discovery DEFAULT_DISCOVERY_METHOD=auto # auto, sitemap, crawl MAX_URLS=1000 # Extraction DEFAULT_EXTRACTION_METHOD=auto # auto, jina, crawl4ai MAX_PARALLEL_REQUESTS=10 # Chunking CHUNK_SIZE=1000 CHUNK_OVERLAP=200 # Embeddings EMBEDDING_MODEL=text-embedding-3-small EMBEDDING_DIMENSIONS=1536 ``` ## Example Workflows ### Index Multiple Documentation Sites ```bash /jina https://docs.supabase.com/docs/* supabase /jina https://docs.anthropic.com/claude/* anthropic /jina https://nextjs.org/docs/* nextjs /jina https://python.langchain.com/docs/* langchain ``` ### Search Across All Projects ``` "Search all projects for rate limiting best practices" ``` ### Project-Specific Search ``` "Search the supabase project for authentication examples" ``` ### Clean Up Old Projects ``` "Delete the test-project" ``` ## Support - GitHub Issues: [Create an issue](https://github.com/yourusername/mcp-jina-supabase-rag/issues) - Check logs: `python src/main.py` for detailed error messages - Supabase logs: Check your Supabase project logs for database errors ## What's Next? See the main [README.md](README.md) for: - Architecture details - Tool reference - Development guide - Advanced configuration