Python Runner MCP Server

What is Python Runner MCP Server?

A Python code execution server built on the FastMCP framework, designed for data science and machine learning workflows. It executes Python code in isolated namespaces and comes with pre-installed libraries such as NumPy, pandas, scikit-learn, Matplotlib, and more.

How to use Python Runner MCP Server?

Run the server directly with uvx python-runner (no installation required) or install manually via uv/pip and run python main.py. For use with Claude Desktop, add the server configuration to your claude_desktop_config.json file.

Key features of Python Runner MCP Server

• Safe Python code execution in isolated namespaces
• Pre-installed data science and machine learning libraries
• Real-time capture of stdout, stderr, and return values
• Fully compatible with the Model Context Protocol
• Simple API: execute_python(code) returns a dict with output, error, and success

Use cases of Python Runner MCP Server

• Run data analysis with pandas, NumPy, or SciPy
• Train and evaluate scikit-learn models
• Generate plots with Matplotlib, Seaborn, or Plotly
• Test and debug Python snippets within Claude Desktop
• Perform scientific computing or network analysis with SymPy or NetworkX

FAQ from Python Runner MCP Server

What are the runtime requirements?

Python 3.12 or higher and the uv package manager (recommended) or pip.

How can I run the server without installing anything?

Use uvx python-runner directly. This requires no prior installation.

Is it safe to execute arbitrary code?

Code runs in isolated namespaces and output is captured separately. However, the tool does execute arbitrary Python code; only use it in a trusted environment.

How do I call the server from Python directly?

After manual installation, import execute_python from main and pass a code string. The function returns a dict with output, error, and success keys.