PRIMS – Python Runtime Interpreter MCP Server

PRIMS – Python Runtime Interpreter MCP Server
PRIMS is a lightweight, open‑source Model Context Protocol (MCP) server that gives AI assistants the ability to execute arbitrary Python code in a sandboxed, disposable environment. By exposing a single, well‑defined tool (), PRIMS removes the complexity of managing runtimes, dependencies, and security for developers who want to embed code execution into conversational agents or automated workflows.

The core problem PRIMS solves is the friction of safely running user‑supplied code within an LLM‑driven application. Traditional approaches require maintaining long‑lived virtual environments, handling dependency resolution, and ensuring isolation from the host system. PRIMS eliminates these concerns by spinning up a fresh virtual environment for every invocation, installing requested pip packages on demand, mounting optional read‑only files, and then tearing the workspace down. This guarantees reproducibility, prevents state leakage between calls, and protects the host from malicious code.

Key capabilities include:

• Secure sandboxing: Each execution runs in a freshly created environment that is destroyed afterward, ensuring no residual data or side effects.
• Dynamic dependency management: Callers can specify pip packages to install for that run, allowing the assistant to adapt to new libraries without manual updates.
• File handling tools: downloads a remote file once per session, while , , and provide convenient ways to inspect, preview, or persist results.
• Zero configuration: The server can be launched via MCP/stdio for local testing or packaged in Docker for production deployments, requiring no special setup beyond the standard MCP interface.

Real‑world use cases span data science pipelines where an assistant fetches a dataset, runs analysis code, and returns visualizations; automated testing frameworks that need to execute user scripts in isolation; or educational platforms where learners can submit code snippets and see immediate, sandboxed output. By integrating PRIMS into an AI workflow, developers can expose code execution as a first‑class tool in the MCP ecosystem, enabling richer interactions and seamless orchestration of external computations.