CoEvoSkills: Self-Evolving Agent Skills via Co-Evolutionary Verification

TL;DR

CoEvoSkills is a framework that enables large language model agents to autonomously generate and refine complex multi-file skills through co-evolutionary verification, improving performance on multi-step tasks.

Contribution

It introduces a novel self-evolving skills framework coupling a skill generator with a surrogate verifier for autonomous skill construction.

Findings

Achieves highest pass rate on SkillsBench among five baselines.

Demonstrates strong generalization to six additional LLMs.

Enables autonomous construction of complex multi-file skills.

Abstract

Anthropic proposes the concept of skills for LLM agents to tackle multi-step professional tasks that simple tool invocations cannot address. A tool is a single, self-contained function, whereas a skill is a structured bundle of interdependent multi-file artifacts. Currently, skill generation is not only label-intensive due to manual authoring, but also may suffer from human--machine cognitive misalignment, which can lead to degraded agent performance, as evidenced by evaluations on SkillsBench. Therefore, we aim to enable agents to autonomously generate skills. However, existing self-evolving methods designed for tools cannot be directly applied to skills due to their increased complexity. To address these issues, we propose CoEvoSkills, a self-evolving skills framework that enables agents to autonomously construct complex, multi-file skill packages. Specifically, CoEvoSkills couples a Skill Generator that iteratively refines skills with a Surrogate Verifier that co-evolves to provide informative and actionable feedback without access to ground-truth test content. On SkillsBench, CoEvoSkills achieves the highest pass rate among five baselines on both Claude Code and Codex, and also exhibits strong generalization capabilities to six additional LLMs.