Jupyter MCP Server

Active

GitHub Python BSD-3-Clause

Description

MCP server for Jupyter that enables AI agents to interact with Jupyter kernels, execute code, and manage notebooks.

Real-time notebook control with instant visibility into code changes and execution results
Smart execution feedback — auto-adjusts on cell failures with multimodal output support (images, plots, text)
Complete cell operation toolkit: read, insert, delete, move cells with code and Markdown support
Multi-notebook management — seamlessly switch between and manage multiple notebooks
Deep JupyterLab integration with auto-open notebooks and run-all-cells commands
OpenTelemetry observability with built-in hook system for tracing tool calls and kernel executions

💡 Let AI agents automatically write and execute data analysis code in Jupyter notebooks

💡 Connect Claude Desktop or Cursor via MCP protocol to control Jupyter kernels

💡 Automate ML experiment workflows: data preprocessing, model training, and result visualization

💡 Execute complex notebook operation sequences using natural language instructions

💡 Safely run agent code in JupyterHub or Datalayer hosted notebook environments

pip install jupyter-mcp-server then configure MCP_TOKEN and add the Jupyter MCP Server to your MCP client (e.g. Claude Desktop) to connect