DataViz Storyteller

DataViz Storyteller

Documentation for the conversational data analysis app: setup, Docker & local dev, Jupyter & MCP, sample prompts, concepts, and customization.

Demo video

Watch a quick walkthrough of the app:

Start here

• Get started — install & run the app
• Introduction — core concepts and what the app can do
• Jupyter & MCP — embed Jupyter and connect the Jupyter MCP server

Docker setup (Jupyter + MCP)

Use Docker to spin up JupyterLab and the Jupyter MCP server that the app integrates with.

1. Start JupyterLab via Docker Compose

From the project root (dataviz-storyteller):

docker compose -f docker-compose.jupyter-mcp.yml up -d jupyterlab

This runs JupyterLab with a token and exposes it on port 8888 (see the compose file for details).

2. Run the Jupyter MCP server

Still from the project root:

pip install jupyter-mcp-server
npm run jupyter-mcp

This starts the MCP server at http://localhost:3002/mcp and points it at the work/notebook.ipynb notebook.

3. Connect from the app

In the DataViz Storyteller UI:

1. Open the MCP config modal in the chat toolbar.
2. Add a server with URL http://localhost:3002/mcp.
3. (Optional) Set NEXT_PUBLIC_JUPYTER_LAB_URL in .env.local so the real Jupyter notebook appears in the right panel.

See Jupyter & MCP for a deeper explanation of how the browser MCP connection works, troubleshooting tips, and advanced options.