# SkillEvolver: Skill Learning as a Meta-Skill

Source: [SkillEvolver: Skill Learning as a Meta-Skill](<https://www.sota2.com/research/paper/6a02b47edbc836b87d824c68>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-05-13T05:00:20.520Z

Authors: Genrui Zhang, Erle Zhu, Jinfeng Zhou, Caiyan Jia, Hongning Wang

Published: 2026-05-11T12:58:25.000Z

## 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 $83$ SkillsBench tasks spanning $15^{+}$ domains, SkillEvolver reaches $56.8\\%$ accuracy versus $43.6\\%$ for curated human skills and $29.9\\%$ for the no-skill baseline; on three GPU kernel optimization tasks from KernelBench, it also raises mean speedup from $1.16$ to $1.51$ on average.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Skill execution on SkillsBench](<https://www.sota2.com/research/sota/skill-execution-on-skillsbench.md>): Skill execution; SkillsBench; Overall Success Rate (avg@5): 56.9.

- [Continuous-reward optimization on KernelBench (val)](<https://www.sota2.com/research/sota/continuous-reward-optimization-on-kernelbench-val.md>): Continuous-reward optimization; KernelBench (val); MLP Speedup: 1.089.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2605.10500>)

[PDF](<https://arxiv.org/pdf/2605.10500.pdf>)