---
name: postmortem-to-rule
description: "Applies “Every scar becomes a rule” as a repeatable product-quality gate. Use during operate work when the system must get smarter with every failure — or you'll pay for the same bug twice."
---

# Every scar becomes a rule

Apply this skill during **Operate** work. It supports Any stack.

The system must get smarter with every failure — or you'll pay for the same bug twice.

## Instructions

After every incident, regression, or repeated user complaint, write a 5-line blameless postmortem within 48 hours: what happened, impact, root cause, the one change that prevents recurrence, and where that change now lives (rule, test, or deleted code). Append rules to the project's single living rulebook with the incident cited — never scatter them across docs. Periodically challenge rules that no longer map to a real scar. If errors aren't reaching a human, treat that as its own sev-1.

## Detect the problem

- The same bug fixed twice
- Lessons living in chat threads
- Postmortems that name a person, not a cause

## Hold the gate

Do not declare this phase complete until every item passes:

- [ ] Every incident/user complaint gets a 5-line blameless postmortem within 48h
- [ ] Each postmortem produces exactly one new rule, test, or deleted footgun
- [ ] Rules live in ONE always-read document, each citing its incident
- [ ] A rule with no scar behind it is reviewed for deletion quarterly
- [ ] Error tracking pages a human; silence is treated as an outage of observability

## Verify the result

Map last month's incidents: each one points to exactly one rule, test, or deleted footgun.
