HuggingFace Daily Papers·· 1 天前
Skill^{infty}：面向 Agent 技能优化的递归自我改进方法
Skill^{infty}: Recursive Self-Improvement for Agent Skill Optimization
AI 导读
Skill^{infty} 让技能优化本身递归自我改进：元 Agent 依照 meta-skill 优化任务技能，并用同一流程改写 meta-skill 本身。meta-skill 与任务技能解耦，在多个技能优化任务上打磨后可跨领域复用。在 83 个 SkillsBench 任务上，其通过率为 60.7%，高于人工整理技能的 56.4% 与 SkillEvolver 的 54.5%。
正文 · 原文
Abstract:Recursive self-improvement (RSI), in which a system also improves the way it improves, is a long-standing aspiration on the path to AGI. Language agents increasingly store procedural knowledge in reusable skills. Optimizing skills offers a direct way to achieve RSI, because the optimization procedure itself can be encoded as a skill. Existing skill optimizers, however, follow a hand-designed procedure. Even methods that evolve a meta-skill confine it to a fixed pipeline and relearn it for each task. We present Skill$^{\infty}$, which makes skill optimization recursively self-improving: a meta agent follows a meta-skill to optimize task skills and optimizes the meta-skill itself using the same procedure. The optimization is agent-led. The meta agent diagnoses execution traces, proposes candidates, requests independent reviews, and allocates the search budget. The meta-skill defines these steps, allowing self-improvement to rewrite them. This meta-skill is also decoupled from task skills, refined on multiple skill optimization tasks and then reused to optimize task skills across diverse domains. As in deployment, skills at both levels are standard packages loaded through native skill discovery. On 83 SkillsBench tasks, Skill$^{\infty}$ reaches a pass rate of 60.7%, compared with 56.4% for human-curated skills and 54.5% for SkillEvolver.
| Comments: | 17 pages, 5 figures |
| Subjects: | Artificial Intelligence (cs.AI); Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.35434 [cs.AI] |
| (or arXiv:2610.35434v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.35434 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Wangcheng Tao [view email]
[v1]
Mon, 26 Oct 2026 13:37:56 UTC (40 KB)
来源：HuggingFace Daily Papers · arxiv.org