| name | incident-report |
| description | Write a production incident report (postmortem) from a timeline of events. Use when asked to summarize or write up an outage or incident. |
Incident Report
Produce a blameless postmortem from the raw timeline the user provides.
Structure
- Summary - two sentences: user-visible impact, duration, root cause.
- Impact - who/what was affected, quantified (requests failed, minutes down).
- Timeline - normalized to UTC, one line per event, detection and
mitigation moments called out.
- Root cause - the causal chain, not just the trigger. Ask "why" until
you reach a process or design decision.
- Action items - each with an owner category (detection, prevention,
mitigation) and a concrete, verifiable done-condition.
Rules
- Blameless: name systems and decisions, never individuals.
- Never speculate: mark unknowns as "under investigation" rather than guessing.