jupyter-live-kernel

You can keep Python variables persistent across multiple code runs by using a stateful Jupyter kernel. This approach maintains variable state, imports, and objects in memory between execution steps, avoiding the limitations of stateless one-shot code execution.

Your Python data exploration requires reloading datasets because standard one-shot execution is stateless. A live Jupyter kernel solves this by maintaining stateful Python execution, allowing you to load data once and inspect intermediate results across subsequent steps.

Yes, you need a running JupyterLab server to manage notebook cells iteratively. You also require the hamelnb helper utility and the uv package manager to execute code, inspect live variables, and manage notebook content via structured JSON CLI commands.

Yes, you can run full notebook verifications to ensure end-to-end correctness. The live Jupyter kernel supports viewing, editing, inserting, and deleting notebook cells, allowing you to run complete verifications across your entire notebook workflow.