incident-postmortem-drafter

When to invoke

• You have an outage/incident timeline and need a consistent postmortem draft.
• You want to standardize incident write-ups across teams.

Inputs needed

• Incident metadata: title, date/time, duration, severity.
• Impact summary: customer impact, internal impact, key metrics.
• Timeline of events (timestamp + event).
• Contributing factors (optional) and what went well/poorly.
• Action items (optional; can be generated placeholders).

Workflow

1. Validate required fields and normalize timestamps.
2. Derive key milestones (start, detection, mitigation, resolution).
3. Summarize impact and customer communication.
4. Generate a blameless narrative: what happened and why (avoid personal attribution).
5. Produce corrective and preventive actions (CPAs) with owners and due dates if provided; otherwise create placeholders.

Output format

• Summary
• Customer impact
• Timeline
• Root cause / contributing factors (blameless)
• Detection and response analysis
• What went well / what didn’t
• Action items (table)
• Follow-ups and references

Guardrails

• Do not assign blame to individuals; focus on systems and process.
• If inputs are missing, call them out explicitly rather than inventing facts.
• Keep timestamps and units consistent.

Reference code

• postmortem_drafter.py

  reads a JSON incident file and writes a Markdown postmortem.