Reproduce “Reactive Python Notebooks as Tools for Coding Agents” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Trevor Manz, Myles Scolnick, Akshay Agrawal

marimo pair lets an agent explore data in a live notebook by running Python actions in the kernel, where it can inspect values and receive execution feedback. The runtime governs which changes become durable notebook cells, checking invariants before applying them and rerunning affected cells. The paper describes an implemented interface, but reports no empirical evaluation of task performance.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.