Skip to main content
MCP server CX Codex

Python Interpreter — MCP server: what it does, how to install

Python Interpreter — an MCP server that runs Python snippets with stateful session memory. It retains imports, variables, and calculations across tool calls.

no license updated 2026-10-29

Open on GitHub

What Python Interpreter does

This server supplies an AI agent with the execute_code tool to run Python commands interactively. Instead of terminating after each snippet, the execution runtime mimics a Jupyter notebook. Every variable, imported library, and function stays in memory for subsequent steps. The agent can verify its logic before returning output to you.

Running it locally via uv executes commands directly on your machine. To isolate executions from the host system, the author recommends the Docker MCP Toolkit with a volume mapped to /app/notebooks.

What it helps with

  • run Python snippets directly through model tool calls
  • retain state, imports, and variables across separate prompts
  • let the model test code output before writing the final solution
  • isolate script execution inside a Docker container

Example prompts for the agent

  • Calculate 50 factorial using execute_code and store it in variable n
  • Use the variable n from the previous step and convert it to string
  • Define a data cleanup function and verify it with sample inputs

How to install Python Interpreter

Install

The MCP registry has no launch command for it: see the repository README

Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more

Questions about Python Interpreter

How do you install Python Interpreter locally using uv?

Run uv pip install . inside the repository root, then register the server using uv run mcp install src/server.py.

What does Python Interpreter do?

It gives LLMs a persistent Python environment to test snippets, preserve variables, and verify outputs incrementally.

How can you test the tools before connecting an agent?

Execute uv run mcp dev src/server.py from the project root to open the developer web interface.

Similar MCP servers

All MCP servers for Agent setup: memory, context, orchestration →

Project sponsors

Companies that keep this analytics open