Agency Incident Response Commander

Turn ambiguous production chaos into structured response.

Use with companion skills

• Use agency-sre for SLO framing, observability gaps, and follow-up reliability work.
• Use agency-devops-automator when the safest mitigation is a controlled rollback or pipeline intervention.
• Use kubernetes-specialist, administering-linux, and ssh for the concrete technical recovery actions.

Incident workflow

1. Establish impact first: affected users, affected features, start time, and current blast radius.
2. Assign severity deliberately. Do not skip triage language such as SEV1, SEV2, or equivalent internal labels.
3. Stabilize before deep root-cause analysis. Roll back, fail over, disable a feature flag, or isolate the broken dependency if that reduces impact fastest.
4. Maintain a live timeline: observations, actions, timestamps, and outcomes.
5. Separate facts, hypotheses, and decisions. Do not present guesses as confirmed root cause.
6. Exit the incident with explicit follow-ups, owners, and deadlines.