mcp -openmetadata

This MCP server provides a standardized interface to interact with an OpenMetadata instance via the Model Context Protocol. It exposes MCP-compatible endpoints that wrap OpenMetadata's REST API, enabling MCP clients to discover and manipulate data entities such as tables, databases, dashboards, charts, pipelines, topics, metrics, containers, reports, and ML models. Clients can list, get by ID, get by name, create, update, and delete these entities through the MCP layer, while the underlying calls are translated to corresponding OpenMetadata REST API calls.

To use it, configure the environment to point at your OpenMetadata instance, start the MCP server, and connect MCP clients to the host and port you expose. Common MCP client capabilities you can leverage include: listing core data entities (tables, databases, database schemas), managing dashboards and charts, handling pipelines and topics, and working with governance assets like classifications, glossaries, and tags. The server also surfaces system resources such as users, teams, bots, services, and events, aligned with the OpenMetadata model, enabling integrated data governance workflows through MCP.

Typical workflow involves provisioning required credentials for OpenMetadata, starting the MCP server, and using an MCP client to perform create/read/update/delete operations against the standardized MCP endpoints, which are then proxied to the OpenMetadata REST API.

Prerequisites:

• Node.js (v14+ recommended) and npm installed on your system
• Access to the repository containing this MCP server

Install and run locally:

1. Clone the repository: git clone https://github.com/yangkyeongmo/mcp-server-openmetadata.git cd mcp-server-openmetadata

2. Install dependencies: npm install

3. Configure environment variables (examples):

   • OPENMETADATA_API_BASE_URL=https://openmetadata.example/api
   • OPENMETADATA_API_TOKEN=your_openmetadata_token
   • MCP_HOST=0.0.0.0
   • MCP_PORT=8080

4. Start the MCP server: npm run start

5. Verify the server is running by hitting the MCP endpoint (e.g., http://localhost:8080/health) or by using an MCP client to connect to the server.

Notes:

• If your deployment uses a different entry file, adjust the mcp_config accordingly (node path/to/entry.js).
• Ensure network access from MCP clients to the host/port where the server is listening.

Tips and common issues:

• Environment variables are required to route requests to OpenMetadata; ensure OPENMETADATA_API_BASE_URL is reachable and that any authentication tokens are valid.
• If you run behind a proxy or corporate firewall, configure your network settings to allow MCP client traffic to the server and to the OpenMetadata API.
• For production deployments, consider using a process manager (e.g., pm2) and setting up proper logging and restart policies.
• The MCP server exposes a comprehensive set of endpoints for core data entities, dashboards, charts, pipelines, topics, metrics, containers, reports, ML models, and governance resources (classifications, glossaries, tags). Use MCP clients to perform CRUD operations and to ensure data lineage and metadata are consistently synchronized with OpenMetadata.