让智能体自我进化技能，比人工设计更高效。
Self-Evolving Skills via Surrogate-Guided Solve-and-Reproduce
- 通过‘解题-重演’协议分离求解与可复现交付，提升鲁棒性。
- 在弱模型上实现74.9%准确率，较人工基线提升14.8分。
- 适合想自动生成可靠任务技能的研究者或开发者。
智能体技能是部署时调用的指令与资源封装。当前自进化技能存在两大问题：一是从零演化出的技能性能低于人工设计，甚至不如不使用技能；二是演化阶段记录的仅是一次幸运轨迹，随机新代理在部署时往往无法复现。本文提出reSolve，一个基于三个组件的每任务、人工监督式框架。它将交互式求解与独立可重演的交付结果解耦，采用‘解题-重演’协议；通过不可访问隐藏测试或参考答案的代理验证器增强稀疏奖励信号；再在解法构建图上进行验证器引导的束搜索。在固定约束下，低成本模型自进化技能达到74.9%（均值3次），比人工基线60.1%高出14.8分，超越最强官方人工技能结果（67.3%，GPT-5.5/OpenHands）。还报告了失败案例与领域级表现，包括14个自然科学任务的结果，明确该方法适用边界。
原文摘要 · Abstract (English)
Agent skills are portable packages of instructions and resources an agent consults at deployment. Self-evolving them fails in two ways today. First, skills evolved from scratch underperform human-curated ones and, on a weak model, using no skill at all. Second, an evolution-time pass records one lucky trajectory that a fresh stochastic agent often fails to reproduce at deployment. We present reSolve, a per-task, oracle-in-the-loop framework built on three components. It decouples interactive solving from a self-contained deliverable that is independently re-executed in a fresh container, a protocol we call solve-and-reproduce. It enhances the sparse reward signal with a surrogate verifier that cannot access hidden tests or reference answers. It then runs verifier-guided beam search over a solution-construction graph. Within a fixed harness, a cheap model self-evolves skills that reach $74.9\%$ mean-of-3, $+14.8$ points over the $60.1\%$ human-curated baseline, exceeding the strongest official curated-skill result ($67.3\%$, GPT-5.5/OpenHands). We also report observed failure cases and domain-level results, including performance on the 14 Natural Science tasks, to clarify when the approach does and does not help.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。