Reproduce “LIBRA: Programmatic Induction of Executable Tool Libraries” 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
Zekai Li, Zhenke Liu, Yonghao Li, Junhan Liu, Xulong Xiao, Jiapeng Zhang
LIBRA improves tool retrieval by linking query-matched tools to prerequisites identified through executable contracts and sandbox tests. A dense retriever finds seed tools, then LIBRA propagates relevance through the tested graph. Across three benchmarks, it improves one-shot Recall by 7.9 percentage points and NDCG by 3.7 over QTS; these measure retrieval candidates, not task completion.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.