SearchSkill: Teaching LLMs to Use Search Tools with Evolving Skill Banks

TL;DR

SearchSkill introduces an evolving skill bank framework that enhances language models' ability to plan and execute search queries more effectively in open-domain question answering.

Contribution

It presents a novel two-stage training approach that explicitly models skill selection and grounding, improving retrieval efficiency and answer accuracy.

Findings

Improves exact match scores on knowledge-intensive QA benchmarks.

Reduces repeated and broad queries, focusing on atomic hops.

Achieves better retrieval behavior with fewer search steps.

Abstract

Teaching language models to use search tools is not only a question of whether they search, but also of whether they issue good queries. This is especially important in open-domain question answering, where broad or copied queries often waste retrieval budget and derail later reasoning. We propose \Ours, a framework that makes query planning explicit through reusable search skills. At each step, the model first selects a skill, then generates a search or answer action conditioned on the selected skill card. The skill inventory itself is not fixed: SearchSkill maintains an evolving SkillBank, expands or refines it from recurrent failure patterns, and reconstructs affected trajectories before supervised training. The resulting two-stage SFT recipe aligns training with the inference-time protocol of skill selection followed by skill-grounded execution. Across open-source and closed-source models, SearchSkill improves exact match on knowledge-intensive QA benchmarks and yields better retrieval behavior, including fewer copied first queries, more atomic hop-focused queries, and more correct answers within a small search budget. These results suggest that explicit skill-conditioned query planning is a lightweight alternative to treating search as an undifferentiated action.