Python Interpreter MCP

🚀 Python Interpreter MCP: An MCP Server for Running Scripts

This is a lightweight and experimental MCP server designed to execute arbitrary Python scripts in a structured and reproducible environment. It leverages uv to run isolated code snippets via subprocesses.

✨ Features

Tools

• run_script
  • Run a given script using uv run script.py.
  • Input:
    • code (str): The script to be run.
  • Return: The stdout output of the script.

📦 Installation

No installation steps provided in the original document, so this section is skipped.

💻 Usage Examples

Basic Usage

async with MCPServerStdio(
            params={
                "command": "uvx",
                "args": ["python-interpreter-mcp"],
            }
    ) as server
...

Configuration with Claude Desktop

Add the following content to your claude_desktop_config.json:

"mcpServers": {
  "interpreter": {
    "command": "uvx",
    "args": ["python-interpreter-mcp"]
  }
}

🔧 Technical Details

How It Works

• After receiving the script string, the MCP tool run_script processes the script.
• A hidden folder is created in the current working directory, and the script is saved as a .py file within it.
• Then, the script is executed using uv run to ensure dependency isolation.
• The stdout output of the script is captured and returned as the response.

📚 Documentation

Usage Precautions

This project is still in the development stage.

⚠️ Important Note

• It can execute arbitrary Python code, which means it can run anything, including malicious or destructive commands.
• It should only be used in a trusted and sandboxed environment.
• When integrating this tool with an LLM, always validate the input, set up guardrails, or impose restrictions.

📄 License

No license information provided in the original document, so this section is skipped.