jupyter-ai-agents

Jupyter AI Agents exposes an MCP-enabled interface that lets an AI agent interact with Jupyter notebooks, execute code, manage files, and perform common workspace tasks directly from a chat interface. The server is designed to be used as a Jupyter server extension, integrating MCP tooling so the agent can call notebook tools, run code cells, manipulate notebooks, and manage files within the Jupyter environment. Users can access the chat UI inside JupyterLab and issue natural language prompts that are translated into MCP tool invocations by the agent.

Prerequisites:

• Python 3.8+ and pip
• JupyterLab (optional for local testing) if you want to run with the notebook UI

Installation steps:

1. Install the package from PyPI:

   pip install jupyter_ai_agents
   

2. Resolve potential environment issues noted in the project docs (e.g., specific CRDT-related dependencies if you rely on CLI features):

   pip uninstall -y pycrdt datalayer_pycrdt
   pip install datalayer_pycrdt==0.12.17
   

3. (Optional) Run from source for development:

   git clone https://github.com/datalayer/jupyter-ai-agents
   cd jupyter-ai-agents
   pip install -e .
   

4. Launch JupyterLab to access the MCP-enabled chat interface:

   jupyter lab
   

Note: The MCP integration is provided as part of the Jupyter MCP Server workflow. The agent connects to Jupyter tools through the MCP interface exposed by the server extension.

Tips and notes:

• This MCP server relies on JupyterLab and the Jupyter MCP Server integration; ensure your Jupyter environment is set up as described in the project docs.
• API keys for model providers are configured outside the MCP server (e.g., ANTHROPIC_API_KEY, OPENAI_API_KEY, Azure/OpenAI keys) and should be exported in your environment when using the chat agents.
• If you encounter issues with pycrdt or related dependencies, follow the documented steps to reinstall specific versions (as shown in the README) and restart the shell/Jupyter session.
• The MCP toolset enables actions such as listing notebooks, executing code cells, and managing files through natural language prompts issued by the AI agent.