What problem does it solve? When a Kubernetes Deployment rollout fails or production behavior changes after a recent deploy, on-call engineers need a structured, read-only way to gather rollout evidence before deciding on a rollback. This Skill collects Deployment status, rollout history, and ReplicaSet state so the rollback decision is based on facts rather than guesswork. ## Core Features & Use Cases - Rollout Evidence Collection: Runs four deterministic kubectl commands to capture Deployment JSON, rollout history, rollout status, and ReplicaSet lists. - Decision Branches: Provides structured reasoning branches for ProgressDeadlineExceeded, image regressions, replica availability gaps, and ReplicaSet sprawl. - Read-Only Safety: Never executes rollbacks, scaling, or deletions; it only gathers evidence and escalates with a complete handoff package. - Use Case: A Deployment shows Progressing == False after a new image push. Use this Skill to compare the current and previous ReplicaSet images, confirm the regression timing, and escalate with a recommended rollback revision. ## Quick Start Ask the agent to investigate why the deployment named in $DEPLOYMENT_NAME failed to roll out in the given namespace and whether a rollback is warranted.
