Documentation

FastMCP Python

FastMCP Python is a lightweight, high-performance Python framework that implements the Model Context Protocol (MCP). With KMCP, you can quickly create an MCP project that uses the FastMCP framework with a sample MCP server and echo tool that you can use as a boilerplate to develop your own tools.

Prerequisites

  • Install uv to build and run FastMCP servers.
  • Install the MCP inspector tool to access and test your MCP server. 
    npx @modelcontextprotocol/inspector

Create the MCP project

  1. Create a scaffold for your FastMCP server project.

    The following command creates an MCP server in the my-mcp-server directory and a sample echo tool that you can later use to test your server. Follow the interactive CLI prompts to optionally add a description and author for your project.

    kmcp init python my-mcp-server

  2. Review the project structure that was created for you.

    find my-mcp-server

    Example project structure:

    ├── src/
    |   ├── core/           # MCP server code and utilities
    │   ├── tools/          # Tool implementations
    │   └── main.py         # MCP server entry point
    ├── tests/              # Built-in test suite
    ├── Dockerfile          # Dockerfile for spinning up your MCP server in a Kubernetes environment
    ├── kmcp.yaml           # MCP container definition for spinning up your MCP server
    ├── pyproject.toml      # Python dependencies
    ├── .env.example        # Sample environment variables file for your MCP server
    └── README.md           # MCP Project documentation

Add tools

  1. Create a boilerplate for an MCP tool in the src/tools directory of your MCP project.

    The boilerplate includes code for an echo tool that is similar to the my-mcp-server/src/tools/echo.py tool that is automatically created when you initialize the project. You can use the boilerplate as a base to create your own tools.

    kmcp add-tool mytool --project-dir my-mcp-server

  2. Review the code for the mytool tool.

    cat my-mcp-server/src/tools/mytool.py

  3. Make changes to the mytool.py tool to implement your own tool logic.

Run locally

Use the kmcp run command to run your MCP server on your local machine so that you can test your MCP server and tools.

  1. Run your MCP server on your local machine. The command builds the Docker image for your MCP server and automatically opens the MCP inspector tool so that you can connect to your MCP server and test it.

    Tip: The following command automatically opens the MCP inspector tool so that you can test your MCP server. If you do not want to open the MCP inspector tool, add the --no-inspector option to your command.

    kmcp run --project-dir my-mcp-server

  2. In the MCP inspector tool, click Connect to connect to your MCP server. Note that it might take a few seconds for the MCP inspector tool to successfully connect to your MCP server. If the connection fails, make sure the following fields are set in the MCP inspector tool.

    • Select STDIO from the Transport Type drop down.
    • In the Command section, enter uv.
    • In the Arguments section, enter run python src/main.py.
    • In the Configuration section, enter the Proxy Session Token that was returned in the kmcp run command.
    Locally run your MCP server with the MCP inspector tool
    Locally run your MCP server with the MCP inspector tool

  3. Try out the mytool MCP tool.

    1. Go to the Tools tab and click List Tools.
    2. Select the mytool tool.
    3. Enter any string in the message field, such as Hello world and click Run Tool.
    4. Verify that you see your message echoed in the Tool result card.
    Successful run of the mytool tool
    Successful run of the mytool tool

Next