SkillHone：一种通过持久决策历史实现智能体技能持续演化的框架

摘要

智能体技能通过任务专用流程、脚本和参考信息扩展了语言模型智能体的能力范围，但其面向的任务与环境持续变化。现有方法在有限运行中改进技能，仅保留最终产物，舍弃了后续智能体理解先前修订、评估和被否决替代方案所需的决策历史。我们提出SkillHone——一种基于持久化决策历史实现智能体技能持续演化的框架。SkillHone将技能修订与提供实践反馈的评估端证据配对，记录诊断、修订、证据和结果的结构化历史。采用角色分离的子智能体在实践探针上运行候选技能并进行脱敏报告，基于先前决策提出修订建议，无需重新发现过往推理即可实现跨会话优化。在深度研究基准测试中，SkillHone无需预集成搜索堆栈即可运行，其在GAIA和WebWalkerQA-EN上分别比商业支持的深度研究智能体高出15.8分和3.2分，同时也超越了先前的技能演化方法。我们进一步将SkillHone部署于内部工具介导的分析场景，在七种设置下平均准确率提升18.8分。

English

Agent skills extend language-model agents with task-specific procedures, scripts, and references, but the tasks and environments they target continually change. Existing methods improve skills in bounded runs and retain only the final artifact, discarding the decision history that later agents need to interpret prior revisions, evaluations, and rejected alternatives. We introduce SkillHone, a harness for continual agent skill evolution grounded in persistent decision history. SkillHone pairs skill revisions with evaluation-side evidence that supplies practice feedback, recording structured histories of diagnoses, revisions, evidence, and outcomes. Role-separated subagents run candidate skills on practice probes with redacted reporting and propose revisions informed by prior decisions, enabling cross-session refinement without rediscovering past rationale. On deep-research benchmarks, SkillHone runs without a pre-integrated search stack and outperforms the commercially backed deep-research agent by 15.8 points on GAIA and 3.2 points on WebWalkerQA-EN, while also exceeding prior skill-evolution methods. We further deploy SkillHone on internal tool-mediated analysis scenarios, where it improves accuracy by an average of 18.8 points across seven settings.