| title | SkillSeek: Revisiting Agent Skill Retrieval at Marketplace Scale |
|---|---|
| authors | Guanqun Yang, Wenlong Zhang, Tian Shi, Ping Wang |
| date | 2026-09-30 |
| venue | arXiv (preprint, not peer reviewed) |
| id | arXiv:2609.38822 |
| affiliation | Stevens Institute of Technology |
| upvotes | 6 (Hugging Face Daily Papers, 2026-11-01) |
| links | abs pdf hf code |
| abstract | Anthropic's Agent Skills package reusable procedural know-how for an LLM agent into SKILL.md directories, and open-source aggregations have grown past 230,000 skills, making selection rather than authoring the bottleneck. The standing answer in the literature outsources selection to the agent itself: an LLM-mediated retrieval loop that rewrites queries and refines candidates inside the agent's decision loop, paying LLM tokens on every task. We present SkillSeek, an open-source two-stage skill retriever built from the standard IR recipe (a BGE-base bi-encoder feeding a small cross-encoder, exposed over MCP). Across a 4 times 11 grid of pool, backbone, and method on the 89-task SkillsBench benchmark, SkillSeek reaches observed parity with the LLM-mediated loop of Liu et al. at essentially no extra cost: plain bm25 alone records a pass rate at or above their refined loop on three of four settings, and a small cross-encoder covers the remaining difference on the fourth. A first-stage recall ceiling explains the pattern, and total per-trial spend drops from USD 51.30 to USD 27.54 (within fifty cents of the no-skill baseline). Under the SkillsBench tasks and OpenHands harness we tested, this positions the standard IR recipe as a strong default for agent-skill retrieval, with LLM-mediated alternatives a natural fit for cases where deterministic methods fall short. |
cite
@misc{yang2026skillseek,
title={SkillSeek: Revisiting Agent Skill Retrieval at Marketplace Scale},
author={Guanqun Yang and Wenlong Zhang and Tian Shi and Ping Wang},
year={2026},
eprint={2609.38822},
archivePrefix={arXiv},
url={https://arxiv.org/abs/2609.38822}
}