Datalayer's open-source Jupyter MCP Server connects AI assistants to a running JupyterLab or Jupyter server so they can read notebooks, insert and edit cells, execute code and see outputs, including images. It runs locally via uvx (or Docker) and authenticates to Jupyter with its server URL and token.
Connect
- 1Add the server to Codex
codex mcp add jupyter --env JUPYTER_URL=YOUR_JUPYTER_URL --env JUPYTER_TOKEN=YOUR_JUPYTER_TOKEN -- uvx jupyter-mcp-server@latest
Configs are generated from the publisher's documented endpoint. Check their docs for plan requirements and scopes. New to this? Read the step-by-step tutorial.
Things to ask
- “Load sales.csv in my notebook and plot monthly revenue”
- “Run all cells and fix any that error”
- “Add a cell that trains a logistic regression on this dataframe”
- “Summarize what this notebook does”