superplan-postmortem

Retrospective Postmortem

Overview

Turn completed or failed work into reusable learning.

This skill exists to reduce repeated mistakes, not to assign blame.

Trigger

Use when:

• a task finished and the team wants to learn from it
• a task failed, thrashed, or took an unexpected path
• a recurring trap or process weakness became visible

Stay Out

Do not use when:

• execution is still actively blocked on basic facts
• the task is mid-flight and the right next move is still direct execution
• there is nothing durable to learn beyond ordinary completion notes

Core Rules

• separate facts, causes, and follow-up changes
• capture what went wrong and what almost went wrong
• identify one or two concrete rules, checks, or habits that would prevent recurrence
• keep the writeup concise enough to be reusable
• record recurring traps in .superplan/gotchas.md
• record process or policy changes in .superplan/decisions.md

Forbidden Behavior

• blame-oriented writing
• vague lessons like "be more careful"
• rewriting history to make the path look cleaner than it was
• producing a long transcript instead of actionable learning

Handoff

Likely follow-ons:

• superplan-guard when a test or proof path should be added
• superplan-docs when the docs or help misled the work
• superplan-shape when the underlying task breakdown needs to change