alvii147/piston-mcp

Overview

This MCP server connects LLMs to the Piston code execution engine. It allows language models to run code snippets in various languages through the Model Context Protocol. The server is written in Python and provides a sandboxed execution environment.

Best for

Best for
Developers building LLM tools that need safe code execution

Use cases

• Running code snippets from LLM conversations
• Testing code generation outputs in real time
• Providing LLMs with sandboxed code execution

How to use

Tools exposed

• uv

Tested with

Claude Desktop

Example client config

{\n  "mcpServers": {\n    "piston": {\n      "command": "uv",\n      "args": [\n        "tool",\n        "run",\n        "--from",\n        "git+https://github.com/alvii147/piston-mcp.git@main",\n        "piston_mcp"\n      ]\n    }\n  }\n}

Notes

This MCP server connects LLMs to the Piston code execution engine. It allows language models to run code snippets in various languages through the Model Context Protocol. The server is written in Python and provides a sandboxed execution environment.

8 stars on GitHub. Last updated 2026-02-22. Licensed MIT.

Use cases

• Running code snippets from LLM conversations
• Testing code generation outputs in real time
• Providing LLMs with sandboxed code execution

Pros

• Leverages Piston’s sandboxing for safe execution
• Simple integration with MCP-compatible clients
• Lightweight Python implementation

Cons

• Low community adoption with only 8 stars
• Limited to languages supported by Piston
• Requires running a separate server process

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.