Retrospective & Continuous Learning

This skill exists to stop: incidents passing without producing a rule — or lessons written as "be more careful," which nobody can reuse.

📁 Source note: [sage] = upstream Sage repo (github.com/xoai/sage, public) — optional deeper reading; this skill runs fully on the rules inlined here. A step marked MUST READ points at a file in your own project (e.g. an event registry) — if it is missing, stop and ask instead of improvising.

🤖 0. HOW TO USE (agent workflow)

A. A/B test retro: compare the original bet ("we believed X would move KR Y by Z because…") with actuals; if multiple variables were mixed, report "result unreadable" instead of inventing a conclusion. B. Incident post-mortem: 5-whys to the root; lesson in strict WHEN/CHECK/BECAUSE form; log it in the team lesson book. C. Mint new rules: every rule must trace to a real failure (date/ticket/witness — kaizen) AND answer "what mechanism keeps it" — a rule that's only a reminder gets demoted to SHOULD. 📄 Full original (203 lines): [sage] skills/sage-reflect/SKILL.md.

Core stance: "Sprint fast, A/B relentlessly, fix what breaks. No solution is perfect on paper — but never step in the same pothole twice."