AI 操作界面出错后，现在能自己改“技能”了
过去教 AI 操作手机或电脑界面，靠的是提前写死的“技能”——弹窗、加载慢、按钮挪了位置，任何一个意外都能让计划作废。这篇提出一个不重新训练就能自我修正的框架：AI 执行任务时，一旦失败，会有一个“批评者”在隔离环境里诊断原因，然后 AI 自己修改对应的技能文件，下次遇到同类问题就能绕开。在三个主流界面基准上，多个基础模型都稳定提升，最高涨了 16 个百分点，而且改过的技能还能复用到相关任务上。它不是你明天就能装上的工具，但“让 AI 从自己的错误里改行为”这件事，正在从实验室走向真实界面。
📄 原文摘要(英文)
GUI agents execute long-horizon tasks on dynamic graphical user interfaces, where pop-ups, delayed loads, and relocated widgets routinely invalidate plans fixed before execution. Recent agent-skill frameworks encapsulate reusable procedural knowledge to mitigate this, yet existing skill designs are largely developed without targeting GUI execution dynamics and treat skills as static artifacts produced before deployment rather than living procedural knowledge that improves through it. We argue that what GUI agents need is not better static skills, but skills that can be revised from execution feedback at deployment time, without additional training. We propose EvoSkill-GUI, a training-free framework in which each skill is a structured multi-file package containing retrieval metadata, executable plans, backup localization, failure-recovery rules, accessibility utilities, and failure cases. EvoSkill-GUI operates through a \emph{reflect-revise-reuse} loop: the executor performs instant in-rollout revisions, an isolated critic diagnoses failed trajectories under strict information isolation, and the executor edits specific skill files through a restricted tool interface. Across MobileWorld, AndroidWorld, and OSWorld, three mainstream GUI benchmarks spanning mobile and desktop platforms, EvoSkill-GUI consistently improves multiple base models without any training, with maximum gains of +16.2%, +6.0%, and +10.5% respectively, and evolved skill libraries continue to benefit related tasks rather than being rebuilt from scratch. Our code is available at https://github.com/ZJU-REAL/EvoSkill-GUI.