// Automating SKILL.md Generation // Increasingly, mining sessions is one of the best ways to impro...
TL;DR · AI 摘要
文章介绍了一种通过挖掘会话数据自动生成SKILL.md的方法，但实际效果有限。
核心要点
- OpenAI 的 Codex 能够从交互中打包技能，但效果有限。
- 论文中提出的三阶段流程在技能聚类上表现较好，但实际应用效果不佳。
- GRPO 方法仅将技能步骤准确率从 18.5% 提高到 20.5%，效果有限。
结构提纲
按章节快速跳转。
思维导图
用一张图看清主题之间的关系。
查看大纲文本（无障碍 / 无 JS 友好）
- SKILL.md 自动生成方法
- OpenAI Codex 方法
- 从交互中打包技能
- 论文提出的三阶段流程
- 分割 GUI 轨迹
- 聚类技能
- 训练技能感知策略
- 方法的局限性
- GRPO 方法效果有限
- 三个主要问题：弱边界检测器、无序的段表示、离线奖励模型
金句 / Highlights
值得收藏与分享的关键句。
论文中提出的三阶段流程在技能聚类上表现较好，但实际应用效果不佳。
GRPO 方法仅将技能步骤准确率从 18.5% 提高到 20.5%，效果有限。
作者指出三个主要问题：弱边界检测器、无序的段表示和离线奖励模型。
elvis on X: "// Automating SKILL.md Generation // Increasingly, mining sessions is one of the best ways to improve your agents. OpenAI released something similar yesterday that lets Codex package skills from interactions. (bookmark it) This paper explains a related approach. They run a https://t.co/HyVvbtF1MN" / X
elvis
@omarsar0
// Automating SKILL.md Generation // Increasingly, mining sessions is one of the best ways to improve your agents. OpenAI released something similar yesterday that lets Codex package skills from interactions. (bookmark it) This paper explains a related approach. They run a three-stage pipeline that segments GUI trajectories, clusters them into candidate skills, and trains a skill-aware policy. The clusters are genuinely readable, with five of eight hitting 0.95 or higher purity against ground-truth workflow labels. But readability does not transfer. GRPO lifts skill-step accuracy only from 18.5% to 20.5%, leaves BrowseComp+ flat, and loses to trivial frequency priors. The authors name the three culprits: a weak boundary detector, an orderless segment representation, and an offline reward model. Paper:
Learn to build effective AI agents in our academy:
academy.dair.ai
3:04 PM · Jun 19, 2026
7.2K
Views
1
3
13
5
15
2
8
128
0
180
Read 13 replies