MCP Server Example (Python)

A simple Model Context Protocol (MCP) server implementation in Python.

What is MCP?

The Model Context Protocol (MCP) is an open protocol that enables AI assistants like Claude to securely interact with external tools and data sources. MCP servers expose tools that clients can discover and call.

Project Structure

mcp_playground/
├── server.py           # Main MCP server implementation
├── pyproject.toml      # Project metadata and dependencies
├── .python-version     # Python version specification (3.13)
├── .gitignore         # Git ignore rules
└── README.md          # This file

Setup

Prerequisites

• Python 3.13
• uv - Fast Python package installer and manager

Installing uv

If you don't have uv installed yet:

# On macOS/Linux
curl -LsSf https://astral.sh/uv/install.sh | sh

# On Windows
powershell -c "irm https://astral.sh/uv/install.ps1 | iex"

Installation

1. Clone this repository (if you haven't already)

2. Install dependencies using uv:

uv sync

This will:

• Automatically install Python 3.13 if needed
• Create a virtual environment
• Install all dependencies from pyproject.toml

Running the Server

To run the MCP server:

uv run server.py

Or activate the virtual environment first:

source .venv/bin/activate  # On Windows: .venv\Scripts\activate
python server.py

The server uses stdio (standard input/output) for communication, which is the standard way MCP servers communicate with clients.

Available Tools

This example server provides three simple tools:

1. add_numbers

Adds two numbers together.

Parameters:

• a (number): First number
• b (number): Second number

Example:

{
  "a": 5,
  "b": 3
}

2. greet

Generates a personalized greeting.

Parameters:

• name (string): Name of the person to greet

Example:

{
  "name": "Alice"
}

3. reverse_string

Reverses a string.

Parameters:

• text (string): Text to reverse

Example:

{
  "text": "Hello World"
}

Using with Claude Code

To use this MCP server with Claude Code, you need to configure it in your MCP settings:

1. Open your Claude Code MCP configuration file (typically at ~/.config/claude/claude_desktop_config.json)

2. Add this server to the mcpServers section:

{
  "mcpServers": {
    "example-server": {
      "command": "uv",
      "args": [
        "--directory",
        "/path/to/mcp_playground",
        "run",
        "server.py"
      ]
    }
  }
}

3. Restart Claude Code

4. The tools will now be available for Claude to use!

Customizing the Server

Adding New Tools

To add a new tool to the server:

1. Add the tool definition to the list_tools() function
2. Add the tool implementation to the call_tool() function

Example:

# In list_tools():
Tool(
    name="my_new_tool",
    description="Description of what the tool does",
    inputSchema={
        "type": "object",
        "properties": {
            "param1": {
                "type": "string",
                "description": "Parameter description"
            }
        },
        "required": ["param1"]
    }
)

# In call_tool():
elif name == "my_new_tool":
    param1 = arguments["param1"]
    result = do_something(param1)
    return [TextContent(
        type="text",
        text=f"Result: {result}"
    )]

Advanced Features

The MCP protocol supports additional features beyond basic tools:

• Resources: Expose data that can be read (files, API data, etc.)
• Prompts: Provide templated prompts
• Sampling: Allow the server to request LLM completions

Check the MCP documentation for more information.

Development

Running in Development Mode

For development, you might want to add logging:

import logging

logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)

Testing

You can test your MCP server using the MCP Inspector tool or by integrating it directly with Claude Code.

Troubleshooting

"Module not found" errors

Make sure you've installed the dependencies:

uv sync

Server not appearing in Claude Code

• Check that the path in your MCP configuration is correct
• Make sure the virtual environment is activated (if using one)
• Try restarting Claude Code

Tool calls failing

• Check the server logs for error messages
• Verify that the input schema matches what you're passing
• Make sure all required parameters are provided

Why uv?

uv is a modern Python package manager that's:

• 10-100x faster than pip
• Automatically manages Python versions
• Has built-in virtual environment support
• Compatible with pip and pyproject.toml standards

Resources

• MCP Documentation
• MCP Python SDK
• MCP Specification
• uv Documentation

License

This is an example project for educational purposes.