jupyter-live-kernel

You can keep Python variables persistent across executions by using a live Jupyter kernel that maintains runtime state. This stateful REPL approach lets you run code iteratively via CLI without restarting the session.

A stateful Python REPL is an interactive execution environment that preserves variables and runtime context across multiple code runs. It is needed for data-science exploration, ML prototyping, and API exploration where step-by-step construction requires maintaining state.

You run Python code iteratively against a live Jupyter kernel by using CLI and REST API tooling to coordinate execution. This sends code to the active kernel and returns results in real time while preserving variables.

Yes, you need uv and a running Jupyter server to use this stateful kernel approach. The hamelnb tooling coordinates code execution via CLI and REST API to ensure the live kernel remains active across runs.

Stateful Python execution is better for machine-learning experiments when you need to build complex computations step by step. Unlike standard scripts that restart state, a live Jupyter kernel preserves loaded models and variables across iterative runs.