Reproduce “Crayotter: Traceable Multi-Agent Workflows for Long-Form Video Editing” 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

Lecheng Yan, Yichong Zhang, Xiantao Xu, Jianze Lin, Ben Pan, Xiaoyu Zheng, Jiawei Qian, Anqi Wu and 7 more

Crayotter makes long-form video editing an inspectable, repairable workflow over source footage. It retrieves and analyzes candidate material, builds a time-grounded editing blueprint, then executes it with tools while retaining reports, renders, and tool events for diagnosis and localized re-execution. Across 23 themes, human scores averaged 3.40/5 overall versus 3.23/5 for ChatCut; this system-level comparison does not isolate traceability’s effect.

Summary by alphaXiv.

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