Jupyter AI

Jupyter AI is a JupyterLab extension that connects AI agents and chat-based assistants to your notebooks, letting you prompt models, generate code, and have agents edit cells inside reproducible workflows.

It targets JupyterLab and integrates with frontier coding agents such as Claude, Codex, and Gemini through the Agent Client Protocol (ACP), and each agent exposes its own models, modes, and effort settings in the chat. Jupyter AI ships without a bundled agent, so you install the agent CLIs you want along with any ACP adapter they require, and Jupyter AI detects them automatically. LiteLLM-backed model access stays available through the optional Jupyternaut persona and the magic commands add-on.

In JupyterLab, you interact with Jupyter AI through a chat side panel, where you use the persona picker in the chat input toolbar to choose which AI persona replies. The selected persona can write, debug, and run notebook cells on your behalf. Agents reach notebooks and files through a built-in Jupyter MCP server, and a permission system makes them ask for approval before writing files, running commands, or calling MCP tools. Custom MCP servers can be registered in .jupyter/mcp_settings.json.

The %%ai cell magic and %ai line magic are now an optional add-on, installed with python -m pip install jupyter-ai-magic-commands and loaded with %load_ext jupyter_ai_magic_commands.

Official website: jupyter-ai.readthedocs.io

Related Resources

Tutorial

JupyterLab for an Enhanced Notebook Experience

In this tutorial, you'll learn how to use the JupyterLab authoring environment and what it brings to the popular computational notebook Jupyter Notebook. You'll learn about its different tools and discover how they can work together to enhance your notebook experience.

intermediate data-science editors tools

For additional information on related topics, take a look at the following resources:

• Using Jupyter Notebooks (Course)
• Jupyter Notebook: An Introduction (Tutorial)