Reproduce “Anything2Skill: Compiling External Knowledge into Reusable Skills for 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

Qianjun Pan, Yutao Yang, Junsong Li, Jie Zhou, Kai Chen, Xin Li, Qin Chen, Liang He

Anything2Skill compiles supported procedures from external records into evidence-linked reusable skills. It retrieves those skills alongside task-specific documents, complementing rather than replacing retrieval-augmented generation (RAG). In tests on qsv and GitHub-CLI, adding skills to RAG raises GitHub-CLI task success from 76.50% to 94.10%, a 17.60-point increase. This finding is limited to the two command-line benchmarks.

Summary by alphaXiv.

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