jupyter-live-kernel

A stateful Python REPL maintains variables and imports across executions, enabling iterative data exploration. By using a live Jupyter kernel, variables persist across cells so you can build complex workflows incrementally without losing context.

The best way to run a stateful Python REPL for ML experimentation is connecting to a live Jupyter kernel. This provides a notebook-like workflow accessed via script, allowing you to incrementally inspect dataframes and results without losing context.

Yes, you need a running Jupyter server to use a live kernel for interactive Python sessions. The setup requires uv, JupyterLab, and a running server to start and manage the live kernel session with a persistent Python environment.

Yes, you can use a live Jupyter kernel for API exploration and debugging instead of a standard notebook. It turns ad-hoc experiments into repeatable sessions, allowing you to incrementally inspect results through a script-based skill.

A stateful Jupyter kernel differs from standard Python scripts by persisting state and variables across executions. This persistent environment enables incremental evaluation for data science, preventing the context loss that occurs when rerunning standard scripts.