mcp_pymdu

This MCP server exposes the PyMDU MCP interface via an OpenAPI-compatible HTTP server using mcpo as a lightweight proxy. The setup runs the mcpo tool through uvx to launch a FastMCP process that serves mcp_pymdu/server.py over HTTP on port 8000. You’ll interact with the PyMDU MCP endpoints through standard RESTful calls, allowing your LLM agents or tools to communicate with PyMDU without custom MCP glue code. To get started, start the server using the configuration shown in mcp_config, ensuring the PYTHONPATH points to the repository so Python can locate mcp_pymdu/server.py and its modules. You’ll need an API key as configured (top-secret in this example) to access the HTTP API, which helps secure access to the MCP proxy.

Once running, you can send MCP requests (translated to REST/OpenAPI calls by mcpo) to the HTTP endpoint exposed by the proxy. The setup leverages the standard PyMDU MCP server implementation (server.py) and exposes its capabilities through a conventional HTTP interface, enabling you to integrate Claude, Gemini, or other tools that expect REST/OpenAPI access to MCP-powered functionality.

Prerequisites:

• Python environment and a working internet connection
• Optional: micromamba (recommended for environment management)
• Poetry for Python dependencies
• Internet access to install mcpo and Python packages

Step-by-step installation:

1. Clone the repository: git clone <repo-url> cd mcp_pymdu

2. Create and activate a dedicated environment (using micromamba as shown in the README): micromamba env create -f environment.yml micromamba activate mcp_pymdu

3. Install Poetry for Python dependency management: pip install poetry

4. Install Python dependencies for the MCP server: poetry install

5. Install mcpo (the MCP HTTP proxy): pip install mcpo

6. Run the MCP server via the suggested uvx command (as shown in the README): uvx mcpo --port 8000 --api-key "top-secret" -- fastmcp run mcp_pymdu/server.py

Prerequisites recap:

• Ensure Python and pip are installed
• Install Poetry and project dependencies
• Install mcpo to expose MCP as an OpenAPI HTTP server
• Ensure the environment variable PYTHONPATH points to the repository path if needed

Tips and common considerations:

• The server expects an API key for HTTP access; keep this secure and rotate as needed.
• Set PYTHONPATH to the repository path to ensure imports work correctly in server.py.
• port 8000 is used in the example; choose a different port if needed and update the mcp_config accordingly.
• If you modify mcp_pymdu/server.py or its modules, run the proxy again to pick up changes.
• You can use the Gemini IAM workflow or Claude Desktop integration as shown in the README by pointing their MCP configurations to the same HTTP endpoint exposed by mcpo.
• If you encounter module import errors, verify that the virtual environment used by Python contains all dependencies and that PYTHONPATH includes the repo root.
• For debugging, you can run the underlying MCP command directly (fastmcp run mcp_pymdu/server.py) to ensure the MCP server starts before wrapping it with mcpo.