Happy slice of mcp pizzaMCP.pizza

jupyter-notebook-mcp

MCP.Pizza Chef: jjsantos01

Eleven actions are offered and all eleven work: add a cell and run it, edit a cell, run one cell or the whole file, read text or picture output back, save your work, and mark cells up as slides. The catch is narrow support — only Jupyter Notebook version 6, not JupyterLab, not Notebook 7, not VS Code or Colab. It writes and runs code on your own computer with nothing to sign up for, and has had no updates since April 2025.

Unmaintained · No commits in 18 months.
Coding
Data

Use This MCP server To

Build a chart from my data without writing the code Ask for an analysis and watch it run step by step Turn a notebook into a slide deck for a talk Fix a cell that failed and run it again Read back the picture a cell just produced Save my notebook once the work is done

README

JupyterMCP - Jupyter Notebook Model Context Protocol Integration

JupyterMCP connects Jupyter Notebook to Claude AI through the Model Context Protocol (MCP), allowing Claude to directly interact with and control Jupyter Notebooks. This integration enables AI-assisted code execution, data analysis, visualization, and more.

⚠️ Compatibility Warning

This tool is compatible ONLY with Jupyter Notebook version 6.x.

It does NOT work with:

  • Jupyter Lab
  • Jupyter Notebook v7.x
  • VS Code Notebooks
  • Google Colab
  • Any other notebook interfaces

Features

  • Two-way communication: Connect Claude AI to Jupyter Notebook through a WebSocket-based server
  • Cell manipulation: Insert, execute, and manage notebook cells
  • Notebook management: Save notebooks and retrieve notebook information
  • Cell execution: Run specific cells or execute all cells in a notebook
  • Output retrieval: Get output content from executed cells with text limitation options

Components

The system consists of three main components:

  1. WebSocket Server (jupyter_ws_server.py): Sets up a WebSocket server inside Jupyter that bridges communication between notebook and external clients
  2. Client JavaScript (client.js): Runs in the notebook to handle operations (inserting cells, executing code, etc.)
  3. MCP Server (jupyter_mcp_server.py): Implements the Model Context Protocol and connects to the WebSocket server

Installation

Prerequisites

jupyter-notebook-mcp FAQ

Which version of Jupyter does it work with?
Only Jupyter Notebook 6. It does not work with JupyterLab, Notebook 7, VS Code notebooks, or Google Colab, and version 6 is now the older release.
Do I need an account or a key?
No. Nothing is sent to any outside service and there is nothing to sign up for. You do need Python and the uv installer on your machine.
Can I use this to make a chart from my spreadsheet?
Yes. Load the file in a cell, then ask for the chart you want. The code is written, run, and the resulting picture handed straight back to you.
Is it safe to run?
Treat it carefully. It writes and runs Python on your own computer with no confirmation step, so it can change or delete files. The authors say the same and advise backing things up first.
Is the connection protected?
Barely. The link between your notebook and the assistant has no password. It only accepts connections from your own machine, but nothing checks who is connecting, including a web page open in your browser.
How hard is setup?
Involved. You download the project, set up a Python environment, install a custom kernel, paste a config block with a full folder path, then run a short snippet in a notebook cell each time you start.
What if it cannot find my notebook?
The notebook side quietly moves to the next free port number when the usual one is busy, but the assistant side keeps looking at the original, so the connection fails without saying why. Free up the usual port and start again.
Is the project still being worked on?
No changes have landed since April 2025, and the authors describe it as experimental.