菜单
Run SkillAdaptor to evolve agent skills from task failures. Use when improving SKILL.md files from trajectories or after benchmark runs.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/zjunlp/SkillAdaptor --skill skill-adaptor命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
基于 SOC 职业分类
正在显示 SKILL.md
| name | skill-adaptor |
| description | Run SkillAdaptor to evolve agent skills from task failures. Use when improving SKILL.md files from trajectories or after benchmark runs. |
Evolve reusable SKILL.md files from failure trajectories. Python CLI in this repo.
From the SkillAdaptor repo (set skillAdaptorRoot to the directory containing run_plugin.py):
cd <skillAdaptorRoot>
python run_plugin.py init --workspace <workspace> --harness codex
# add task briefs under <workspace>/input_task/*.md
python run_plugin.py --workspace <workspace> --harness codex --max-iterations 2
Outputs: <workspace>/skills/<id>/SKILL.md — auto-synced to ~/.codex/skills/ and <workspace>/.agents/skills/.
Use --harness codex (not openclaw or claude-code) so adopted skills appear in Codex skill discovery paths.
See plugin/codex/README.md in the SkillAdaptor repository.