Jupyter MCP Server
About this tool
Datalayer's MCP server connects agents to local JupyterLab, JupyterHub or hosted notebooks. Its 22 tools connect servers, read notebooks, insert or edit cells and run code with outputs. Optional extensions run code in cloud sandboxes such as Datalayer, Colab or Kaggle. Useful for notebook-based analysis, modeling and reproducible experiment records.
- Feature 1: 22 tools read, insert, edit, move, delete and execute notebooks and cells.
- Feature 2: execute_code runs in the kernel without writing to the notebook.
- Feature 3: Local JupyterLab, JupyterHub and Datalayer-hosted notebooks.
- Feature 4: Optional cloud sandboxes include Datalayer, Kaggle, Colab, Modal and E2B.
- Feature 5: Claude Code plug-in, Docker image and hosted endpoint.
Resource information
- Maintainer
- Datalayer; company project.
- Type
- MCP
- Category
- Experiments & data analysis
- Runs on
- MCP
- Who it's for
- Data-analysis, ML and computational-science researchers using agents to clean data, model and plot in notebooks with code and outputs retained for review.
- License at time of listing
- BSD-3-Clause
- Source
- GitHub
