English

SkillEvolver: Skill Learning as a Meta-Skill

Artificial Intelligence 2026-05-12 v1

Abstract

Agent skills today are static artifact: authored once -- by human curation or one-shot generation from parametric knowledge -- and then consumed unchanged, with no mechanism to improve from real use. We propose \textbf{SkillEvolver}, a lightweight, plug-and-play solution for online skill learning, in which a single meta-skill iteratively authors, deploys, and refines domain-specific skills. The learning target of SkillEvolver is the skill's prose and code, not model weights, so that the resulting artifact drops into any agent without retraining; and the meta-skill itself is just another skill, loaded through the same interface by any protocol-compliant CLI-agent. Unlike trace-distillation, the meta-skill refines only after deploying the learnt skill, such that the learning signal comes from failures another agent encounters while using it -- not from exploratory traces alone. Refinement iterations are governed by a fresh-agent overfit audit that catches possible leakage as well as deployed-skill-specific failures, including the silent-bypass mode in which a skill appears valid in content but is never invoked at runtime. On 8383 SkillsBench tasks spanning 15+15^{+} domains, SkillEvolver reaches 56.8%56.8\% accuracy versus 43.6%43.6\% for curated human skills and 29.9%29.9\% for the no-skill baseline; on three GPU kernel optimization tasks from KernelBench, it also raises mean speedup from 1.161.16 to 1.511.51 on average.

Keywords

Cite

@article{arxiv.2605.10500,
  title  = {SkillEvolver: Skill Learning as a Meta-Skill},
  author = {Genrui Zhang and Erle Zhu and Jinfeng Zhou and Caiyan Jia and Hongning Wang},
  journal= {arXiv preprint arXiv:2605.10500},
  year   = {2026}
}