ipynb-mcp logo

ipynb-mcp

Free

MCP server for AI agents to safely read, edit and run local Jupyter notebooks: zero service, CAS-guarded edits, resumable execution.

Visit Website
ai-agentclaudecursordata-scienceipynbaiai assistantautomation

Overview

ipynb-mcp-server is an MCP server that lets any AI agent read, edit, and run local Jupyter notebooks safely, with zero setup. It is designed for students, data people, teachers, and anyone pointing an agent at real files. The server provides safe editing by using compare-and-swap anchors, atomic writes, and rolling backups, so edits cannot silently corrupt the notebook. It also supports running only target cells with resume or replay modes, avoiding re-execution of long-running cells. The server requires no JupyterLab, no port, and no token; it is a stdio server that talks directly to a Jupyter kernel. Additional features include stale-cell analysis, image outputs as native MCP image blocks, background execution with polling, and per-notebook kernel lifecycle management.

Pricing

Free

Detailed plans have not been confirmed in our catalog. Check the official website for current limits and billing terms.

Visit Website

Prices and limits may change. Confirm the currency, billing period, seat minimum and usage caps on the official website.

Use Cases

  • Students use ipynb-mcp to let an AI agent read a course notebook, write the cells, and run them, with anchored edits that fail loudly if they race with manual edits.
  • Teachers use ipynb-mcp to let an agent read what each student's cells produce and edit submissions with anchored, atomic writes and rolling backups.
  • Anyone pointing an agent at real files uses ipynb-mcp to ensure the agent can show it did not corrupt the notebook.
  • Users can run long jobs without re-running expensive cells by using mode='resume' or mode='replay'.
  • Users can avoid setting up JupyterLab, managing tokens, or keeping a server running; the server is a stdio process with one line of config.
  • Users can get stale-cell analysis to identify which outputs are invalid because their inputs changed.

Features

Lets agents read local notebooks

Lets agents edit Jupyter notebooks

Runs local notebook code

Guards edits with CAS checks

Supports resumable notebook execution

Operates without a separate service

Reviews

No reviews yet. Be the first to share your experience!