Searchcraft MCP Server

# Open WebUI Setup Guide for Searchcraft MCP Server This guide shows you how to integrate the Searchcraft MCP Server with Open WebUI using the Pipelines framework. ## Overview Open WebUI supports MCP servers through its **Pipelines** framework, which acts as a bridge between Open WebUI and MCP servers. This setup allows you to use Searchcraft's 30+ tools directly within Open WebUI's chat interface. ## Prerequisites - Docker and Docker Compose installed - Searchcraft MCP Server built and ready (`yarn build`) - Basic understanding of Open WebUI and Pipelines ## Quick Setup (Recommended) ### Step 1: Start Searchcraft MCP Server ```bash # In your searchcraft-mcp-server directory yarn start # Starts HTTP server on port 3100 ``` ### Step 2: Create Docker Compose Configuration Create a `docker-compose.yml` file: ```yaml services: openwebui: image: ghcr.io/open-webui/open-webui:main ports: - "3000:8080" volumes: - open-webui:/app/backend/data environment: - OPENAI_API_BASE_URL=http://pipelines:9099 - OPENAI_API_KEY=0p3n-w3bu! depends_on: - pipelines pipelines: image: ghcr.io/open-webui/pipelines:main ports: - "9099:9099" volumes: - pipelines:/app/pipelines - ./open_webui_pipeline.py:/app/pipelines/searchcraft_mcp_pipeline.py environment: - PIPELINES_API_KEY=0p3n-w3bu! extra_hosts: - "host.docker.internal:host-gateway" volumes: open-webui: pipelines: ``` ### Step 3: Copy the Pipeline File Copy the provided `open_webui_pipeline.py` to your current directory (it should be in the same folder as your `docker-compose.yml`). ### Step 4: Start the Services ```bash docker compose up -d ``` ### Step 5: Configure Open WebUI 1. Open http://localhost:3000 in your browser 2. Create an admin account 3. Go to **Admin Settings** → **Pipelines** 4. You should see the "Searchcraft MCP Pipeline" listed 5. Click on it and configure the valves: - `MCP_SERVER_URL`: `http://host.docker.internal:3100/mcp` - `ENDPOINT_URL`: Your Searchcraft cluster URL - `ADMIN_KEY`: Your Searchcraft admin key - `ENABLE_MCP_TOOLS`: `true` ### Step 6: Test the Integration 1. Start a new chat in Open WebUI 2. Try asking: "List all my Searchcraft indexes" 3. The pipeline should enhance your message with Searchcraft context 4. The AI will have access to information about available Searchcraft tools ## Manual Setup (Alternative) If you prefer to set up components individually: ### 1. Start Pipelines Server ```bash docker run -d -p 9099:9099 \ --add-host=host.docker.internal:host-gateway \ -v pipelines:/app/pipelines \ --name pipelines \ ghcr.io/open-webui/pipelines:main ``` ### 2. Install the Pipeline Copy the pipeline code from `open_webui_pipeline.py` and: 1. Go to http://localhost:9099 (Pipelines admin interface) 2. Add the pipeline code 3. Configure the valves ### 3. Start Open WebUI ```bash docker run -d -p 3000:8080 \ -e OPENAI_API_BASE_URL=http://host.docker.internal:9099 \ -e OPENAI_API_KEY=0p3n-w3bu! \ -v open-webui:/app/backend/data \ --name open-webui \ ghcr.io/open-webui/open-webui:main ``` ## Configuration Options ### Pipeline Valves - **MCP_SERVER_URL**: URL of your Searchcraft MCP server (default: `http://localhost:3100/mcp`) - **ENDPOINT_URL**: Your Searchcraft cluster endpoint URL - **ADMIN_KEY**: Your Searchcraft admin key for full access - **ENABLE_MCP_TOOLS**: Enable/disable MCP integration (default: `true`) ### Environment Variables For the Searchcraft MCP server, you can also set: ```bash # In your .env file ENDPOINT_URL=https://your-cluster.searchcraft.io ADMIN_KEY=your_admin_key_here ``` ## How It Works 1. **User Input**: User asks about Searchcraft operations in Open WebUI 2. **Pipeline Processing**: The pipeline detects Searchcraft-related keywords 3. **Context Enhancement**: Adds information about available MCP tools 4. **AI Response**: The AI model responds with knowledge of Searchcraft capabilities 5. **Tool Execution**: When needed, the pipeline can facilitate tool calls to the MCP server ## Troubleshooting ### Pipeline Not Loading - Check that the pipeline file is properly mounted - Verify the pipeline syntax is correct - Check Pipelines logs: `docker logs pipelines` ### MCP Connection Issues - Ensure Searchcraft MCP server is running on port 3100 - Check network connectivity between containers - Verify the MCP_SERVER_URL in pipeline valves ### Open WebUI Not Connecting to Pipelines - Confirm OPENAI_API_BASE_URL points to Pipelines - Check that both services are running - Verify the API key matches ## Advanced Usage ### Custom Tool Integration You can extend the pipeline to handle specific Searchcraft operations: ```python def _handle_search_request(self, query: str): """Handle direct search requests through MCP""" # Implementation for direct MCP tool calls pass ``` ### Multiple MCP Servers The pipeline can be extended to work with multiple MCP servers by modifying the `_initialize_mcp` method. ## Support For issues specific to: - **Searchcraft MCP Server**: Check the main README.md - **Open WebUI**: Visit https://github.com/open-webui/open-webui - **Pipelines**: Visit https://github.com/open-webui/pipelines - **General questions about Searchcraft**: Join our [Discord server](https://discord.gg/RQnGD63qWw). - **Issues / Feature Requests**: Visit https://github.com/searchcraft-inc/searchcraft-issues