Overview
JupyterMCP enables Claude AI to directly interact with and control a Jupyter Notebook (version 6.x) through the Model Context Protocol (MCP). The solution establishes two-way, WebSocket-based communication between Claude and a running Jupyter instance, allowing AI-driven operations such as inserting and executing cells, saving notebooks, and retrieving notebook information. The system comprises three main components: a WebSocket Server inside Jupyter (jupyter_ws_server.py) that bridges notebook communication, a Client JavaScript (client.js) that runs in the notebook to perform operations, and an MCP Server (jupyter_mcp_server.py) that implements MCP and connects to the WebSocket server. Installation steps describe prerequisites (Python 3.12+, uv), a kernel install, and Claude desktop integration configuration to launch the MCP server via uv. Usage highlights include starting the Jupyter nbclassic server, selecting the jupyter-mcp kernel, initializing the WebSocket server from Python, and launching Claude with MCP. Available Claude tools cover code insertion, execution, notebook and cell management, and retrieving outputs. This integration is experimental and should be used with caution while evaluating its capabilities in AI-assisted notebook workflows.
Features
Two-way communication
Establish two-way WebSocket-based communication between Claude and the Jupyter notebook, enabling bidirectional control and data exchange.
Cell manipulation
Insert, execute, and manage notebook cells from Claude.
Notebook management
Save notebooks and retrieve notebook information.
Cell execution
Run specific cells or all cells in a notebook.
Output retrieval
Get output content from executed cells, with options to limit text.
WebSocket Server inside Jupyter
jupyter_ws_server.py sets up a WebSocket server inside Jupyter that bridges notebook communication with external clients.
Client JavaScript integration
client.js runs in the notebook to handle operations (inserting cells, executing code, etc.).
MCP Server implementation
jupyter_mcp_server.py implements the Model Context Protocol and connects to the WebSocket server.
Who Is This For?
- Claude AI users:Control and interact with Jupyter Notebooks from Claude to perform AI-assisted coding, data analysis, and visualization.
- Data scientists / researchers:Leverage Claude to orchestrate notebook operations and execute analytics workflows within Jupyter 6.x.
- Jupyter developers / MCP integrators:Install and run the MCP components to bridge Claude and Jupyter notebooks via WebSocket.
