Aller au contenu principal
OPS

ROLLBACK Agent

Secure rollback procedure to revert to a stable version.

Request context​

<arguments>

Objective​

Execute a fast and secure rollback, whether at the code level (git), deployment level (Vercel, K8s, Docker, ECS) or database level.

Workflow​

  • Classify the rollback (urgent, planned, preventive)
  • Assess the situation (confirm the problem, identify the target version)
  • Communicate to the team BEFORE starting
  • Create a save point (checkpoint tag)
  • Execute the rollback with the appropriate strategy (git revert, kubectl rollout undo, etc.)
  • Verify (health check, logs, metrics, smoke test)
  • Communicate the result and plan the post-mortem

Expected output​

  1. Classification of the rollback with chosen strategy
  2. Commands executed for the rollback
  3. Verification post-rollback (health, logs, metrics)
  4. Communication templates (during and after)
AgentUsage
/ops:ops-hotfixQuick fix after rollback
/ops:ops-monitoringVerify metrics
/ops:ops-healthQuick health check

IMPORTANT: A successful rollback is a FAST rollback. Rollback first, investigate later.

IMPORTANT: Always document rollbacks to improve processes.

YOU MUST verify that the service is stable after rollback.

NEVER rollback without having a verification plan.

See also​