大型语言模型 (LLM) 代理依靠可重用技能来解决复杂任务,，但现有的技能创建方法通常将技能视为孤立的, 静态工件, 限制了可重用性, 可靠性, 和长期改进。我们提出MUSE-Autoskill代理(内存利用技能进化),一个以技能为中心的代理框架，它在统一的生命周期:创建,内存,管理,评估,和细化下创建,重用,并细化技能。 MUSE 根据需要创建技能, 将它们存储在任务中, 通过技能目录检索它们, 并积累每项技能的经验以供以后重用和适应。在 SkillsBench 和 SkillLearnBench, 上报告的主要设置中，MUSE-Autoskill 的性能优于 Hermes, Codex, 和 Claude Code。在 SkillsBench, 上，其自行创建的技能在成功覆盖的子集 (85.24% 上超越了人类创作的技能，对比 81.17%), 表明，生命周期管理的技能可以将座席经验提炼为高效的可重用资产; MUSE 创建的技能也比 Codex 或 Claude 创建的技能更有效地转移到 Hermes,，在转移下达到 51.90% 的准确率。这些结果凸显了将技能视为长期,、经验感知, 和可测试资产的重要性。
Large language model (LLM) agents rely on reusable skills to solve complex tasks, but existing skill creation approaches often treat skills as isolated, static artifacts, limiting reusability, reliability, and long-term improvement. We propose MUSE-Autoskill Agent (Memory-Utilizing Skill Evolution), a skill-centric agent framework that creates, reuses, and refines skills under a unified lifecycle: creation, memory, management, evaluation, and refinement. MUSE creates skills on demand, stores them across tasks, retrieves them through a skill catalog, and accumulates per-skill experience for later reuse and adaptation. Across the main reported settings on SkillsBench and SkillLearnBench, MUSE-Autoskill outperforms Hermes, Codex, and Claude Code. On SkillsBench, its self-created skills surpass human-authored skills on the successfully covered subset (85.24% vs. 81.17%), showing that lifecycle-managed skills can distill agent experience into highly effective reusable assets; MUSE-created skills also transfer to Hermes more effectively than Codex- or Claude-created skills, reaching 51.90% accuracy under transfer. These results highlight the importance of treating skills as long-lived, experience-aware, and testable assets.
科目: 人工智能 (cs.AI); 计算和语言 (cs.CL); 机器学习 (cs.LG); 多代理系统 (cs.MA)
Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG); Multiagent Systems (cs.MA)