self-improve
仓库创建 2026年7月6日最近提交 21 天前SkillHot 收录 21 天前
▸ 精选理由
把临时卡顿转为长期技能改进助推团队成长
▸ 风险提示
会修改技能文件,须用户多选确认并审计变更
这个 Skill 做什么
检测会话中技能/流程的不足并建议把改进写入技能文件。
会在一次以技能驱动的流程结束后或者用户要求复盘时,把会话里卡住的地方自动找出来并列成清单,然后用多选的方式问你是否要把改进写进技能文件。平常没问题就不打扰,只有确有证据表明某个技能流程不顺才会出现建议。重点是把实际摩擦变成可执行的技能修订,并且只有在你同意后才会改动。
▸ 展开 SKILL.md 英文原文
Retrospective mechanism that detects weaknesses in the skills/workflows just used, lists them, and asks via multiple choice whether to codify improvements into the skill files. Runs at the end of a skill-driven workflow when concrete friction was observed, or on demand when the user asks for learnings, a retro, or how a skill could be better. Only surfaces when there is a real, evidence-backed weakness — stays silent otherwise.
0
Stars
0
Forks
10
仓库内 Skill
+0
7 日增星
安装 / 使用
给你的 Agent 一句话(通用)
帮我安装这个 skill:https://raw.githubusercontent.com/Seismoforg/claude-config/main/skills/self-improve/SKILL.md或 curl 直取 SKILL.md
curl -fsSL "https://raw.githubusercontent.com/Seismoforg/claude-config/main/skills/self-improve/SKILL.md"SKILL.MD 节选查看完整文件 ↗
# SELF-IMPROVE Turn friction observed this session into general skill-file edits. Only with the user's multiple-choice approval. Only on a real, observed weakness. # WHEN TO RUN - **On demand** — user asks for "learnings", "retro", "how could this skill be better", "self-improve". - **End of a skill-driven workflow** — when you saw concrete friction tracing to a skill (see WHAT COUNTS). Other skills point here at their end. - **Brevity pass** — also scan the just-used skill(s) for bloat/redundancy; propose trims (same approval gate). Tight skills yield nothing → silence. Hard gate: **no evidence-backed weakness AND no real bloat → do nothing, say nothing.** Silence is default. A prompt ne
via SKILL·HOT · 数据来自 GitHub 公开信息 · 原文版权归作者所有