Jupyter MCP Integration

Example Use Cases

• 1

  Programmatically insert, edit, and execute Python or magic cells inside a Jupyter Notebook (v6.x) to run analyses or demos.

• 2

  Retrieve cell outputs (text and images) and notebook metadata so they can summarize results or include outputs in generated reports or messages.

• 3

  Build slide-style presentations inside a notebook by setting slideshow types and arranging concise markdown and code cells.

Description

This MCP server runs a WebSocket bridge inside Jupyter Notebook (6.x) and an MCP server that allows Claude Desktop to control notebooks: insert and edit cells, execute code, retrieve outputs (text and images), and save notebooks. It enables two-way communication so an AI agent can perform data analysis, visualizations, and even orchestrate multi-cell workflows. The project is experimental, local-only, and explicitly limited to Jupyter Notebook 6.x with safeguards and usage notes in the README.