PyRunner

关于

PyRunner is an MCP (Model Context Protocol) Server designed to supercharge AI assistants by offering a persistent Python kernel. Unlike traditional execution environments where variables reset with each run, PyRunner maintains the kernel's state, allowing variables, functions, and loaded modules to persist across multiple executions. This eliminates the need to repeatedly load large datasets or import heavy libraries, significantly accelerating iterative development and analysis. It integrates features like ultra-fast package checks, semantic script searching, structured long-term memory, and robust shell command execution with automatic output redirection, addressing common bottlenecks faced by AI tools like Gemini CLI and VS Code Copilot.

主要功能

• Persistent Python Kernel for stateful script execution and variable retention
• Non-blocking Shell command execution with automatic stdout/stderr redirection for stability
• Rapid Python package dependency checking and installation in microseconds
• Semantic search capabilities to find scripts using descriptions and tags
• Structured long-term memory to store and recall preferences, projects, and commands
• 2 GitHub stars

使用案例

• Iterative data analysis and exploration with large datasets, avoiding repeated loading
• Interactive Python script development in a command-line environment with persistent context
• Machine learning model loading and repeated inference without re-initializing the model