What it does
Runcell is a JupyterLab extension that adds an AI agent to notebooks for data analysis, data science and research. You describe the result you want, and it inspects the notebook and data, writes Python, runs the cells and fixes errors as it goes, rather than only suggesting the next line. It reads the outputs your cells produce, including tables, statistics, charts and images, so its next step is based on what actually ran. It can also answer questions about a single cell in context, and the site says it keeps memory of a dataset, decisions and project state across sessions. It works on existing .ipynb files, with a file tree, global search and git added inside Jupyter. Install it with pip install runcell or through the JupyterLab extension manager, then sign in. The site says access to models such as GPT, Claude and Gemini is included according to the plan, so no personal API key is needed. It is made by Kanaries Data Inc.
Capabilities
- Runs inside JupyterLab on your existing notebooks
- Plans multi-step workflows, writes Python and executes cells
- Diagnoses and fixes errors while working
- Reads tables, charts and other cell outputs
- Cross-session memory of data, decisions and project state
- Model access included with the plan, with no API key required
When to use it
- You analyse data in Jupyter and want an agent to run the steps, not just suggest code
- You need help debugging a notebook cell using its surrounding outputs
- You want to try several analytical methods side by side with real outputs
- You work on long notebook projects and want context kept between sessions
