Reproduce “Let Agents Design How They Search: Automatic Interface Optimization for Agentic Search” 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

Sangjune An, Jaeyoon Jung, Yumin Choi, Jinheon Baek, Junmo Cho, Sung Ju Hwang, Hae Beom Lee

SearchWeaver optimizes corpus-specific search interfaces by jointly revising prepared artifacts, executable tools, and usage instructions using task rewards and agent execution traces. A fixed search agent evaluates candidates, and the proposer selects the best on optimization queries. With GPT-5.4-nano, it raises BCP accuracy from 71.6% to 78.4% and average BRIGHT nDCG@10 from 45.7 to 55.0, though not on every subset.

Summary by alphaXiv.

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