PDF Knowledgebase MCP Server

# Quick Start Guide Get **pdfkb-mcp** running in minutes using Docker/Podman with DeepInfra as your AI provider. ## Prerequisites - **Container Runtime**: Docker or Podman installed - **DeepInfra API Key**: Get your free key at [deepinfra.com](https://deepinfra.com) - **Documents**: A folder with PDF or Markdown files to index ## Step 1: Get Your DeepInfra API Key 1. Visit [deepinfra.com](https://deepinfra.com) and sign up 2. Navigate to your dashboard and create an API key 3. Copy your API key - you'll need it for configuration 💡 **Why DeepInfra?** We recommend DeepInfra because it's extremely cost-effective, supports the latest Qwen3 models, and provides excellent performance for both embeddings and reranking. ## Step 2: Set Up Docker Compose 1. **Download the configuration**: ```bash curl -o docker-compose.yml https://raw.githubusercontent.com/juanqui/pdfkb-mcp/main/docker-compose.sample.yml ``` 2. **Create your documents directory**: ```bash mkdir -p ./documents ./cache ./logs ``` 3. **Edit the configuration**: ```bash # Edit docker-compose.yml with your preferred editor nano docker-compose.yml ``` 4. **Update these key settings**: ```yaml volumes: # Change this path to your document collection: - "/path/to/your/documents:/app/documents:rw" environment: # Add your DeepInfra API key: PDFKB_OPENAI_API_KEY: "your-deepinfra-api-key-here" ``` ## Step 3: Start the Server ```bash # Using Podman (recommended) podman-compose up -d # Or using Docker docker compose up -d ``` The server will start and be accessible at: - **Web Interface**: http://localhost:8000 - **MCP Endpoint**: http://localhost:8000/mcp/ (for HTTP transport) ## Step 4: Add Your Documents ### Option A: Web Interface 1. Open http://localhost:8000 in your browser 2. Use the upload interface to drag and drop PDF files 3. Watch the processing progress in real-time ### Option B: File System 1. Copy your PDF/Markdown files to your documents directory 2. The server automatically detects and processes new files 3. Check the logs: `podman-compose logs -f` (or `docker compose logs -f`) ## Step 5: Set Up Your MCP Client ### For Claude Desktop Add this to your Claude Desktop configuration file: **Location**: - **macOS**: `~/Library/Application Support/Claude/claude_desktop_config.json` - **Windows**: `%APPDATA%\\Claude\\claude_desktop_config.json` - **Linux**: `~/.config/Claude/claude_desktop_config.json` **Configuration**: ```json { "mcpServers": { "pdfkb": { "transport": "http", "url": "http://localhost:8000/mcp/" } } } ``` ### For VS Code with Continue Add to your `.continue/config.json`: ```json { "mcpServers": { "pdfkb": { "transport": "http", "url": "http://localhost:8000/mcp/" } } } ``` ## Step 6: Test Your Setup 1. **Restart your MCP client** (Claude Desktop, VS Code, etc.) 2. **Look for PDF KB tools**: You should see `add_document`, `search_documents`, `list_documents`, `remove_document`, `rescan_documents` 3. **Try searching**: Ask your AI assistant to search your documents Example queries: - "Search for information about neural networks in my documents" - "Find any mentions of API keys or authentication" - "What documents do I have about machine learning?" ## Verify Everything is Working ### Check Container Status ```bash # Using Podman podman-compose ps podman-compose logs -f # Using Docker docker compose ps docker compose logs -f ``` ### Test the Health Endpoint ```bash curl http://localhost:8000/health ``` Should return: `{"status": "healthy"}` ### Test the Web Interface Open http://localhost:8000 - you should see the document management interface. ## What's Configured in This Setup The default configuration includes: - **🚀 Fast Processing**: PyMuPDF4LLM parser for speed - **🔍 Advanced Search**: Hybrid search (semantic + keyword matching) - **📊 Optional Reranking**: Disabled by default; enable with `PDFKB_ENABLE_RERANKER=true`. Defaults to local `Qwen/Qwen3-Reranker-0.6B`. DeepInfra 8B available via `PDFKB_RERANKER_PROVIDER=deepinfra` and `PDFKB_DEEPINFRA_RERANKER_MODEL=Qwen/Qwen3-Reranker-8B`. - **🧠 Optional Summarization**: Disabled by default; enable with `PDFKB_ENABLE_SUMMARIZER=true` for automatic titles and summaries. - **🌐 Web + MCP**: Both web interface and MCP endpoints available - **💾 Persistent Storage**: Documents and cache preserved across restarts ## Next Steps - **[📄 Add More Documents](web-interface.md#uploading-documents)**: Learn about batch uploading and file monitoring - **[⚙️ Customize Configuration](configuration.md)**: Tune performance and features - **[🔧 Troubleshooting](troubleshooting.md)**: Common issues and solutions - **[🎯 Advanced Features](advanced.md)**: Document summarization and enterprise features ## Quick Commands Reference ```bash # Start services podman-compose up -d # Podman docker compose up -d # Docker # View logs podman-compose logs -f # Podman docker compose logs -f # Docker # Stop services podman-compose down # Podman docker compose down # Docker # Update and restart podman-compose pull && podman-compose up -d # Podman docker compose pull && docker compose up -d # Docker # Check status curl http://localhost:8000/health ``` 🎉 **Congratulations!** Your PDF knowledgebase is now ready to use with your favorite MCP-enabled AI assistant.