Learning capture system that extracts HIGH/MED/LOW confidence patterns from conversations to prevent repeating mistakes. Use after user corrections ("no", "wrong"), praise ("perfect", "exactly"), or when discovering edge cases. Complements .squad/agents/{agent}/history.md and .squad/decisions.md.
Install
npx skills add github/gh-aw --skill reflectnpx skills add github/gh-aw --skill reflect -a claude-code -gnpx skills add github/gh-aw --skill reflect -a codex -ggemini skills install https://github.com/github/gh-aw.git --path .squad/templates/skills/reflectgh skill install github/gh-aw reflectgit clone --depth 1 https://github.com/github/gh-aw /tmp/github__gh-aw
mkdir -p ~/.claude/skills ~/.agents/skills
cp -R /tmp/github__gh-aw/.squad/templates/skills/reflect ~/.claude/skills/reflect # Claude Code
cp -R /tmp/github__gh-aw/.squad/templates/skills/reflect ~/.agents/skills/reflect # Codex, Cursor, Gemini CLI, Copilot, OpenCode