What problem does it solve? Kubernetes misconfigurations like missing resource limits, broken probes, and unpinned image tags cause OOMKills, CrashLoopBackOffs, and downtime during node drains. This Skill audits manifests and Helm charts against a severity-ordered checklist so problems are caught before kubectl apply. ## Core Features & Use Cases - Severity-Ordered Review: Blockers (resources, probes, secrets, image tags, securityContext) first, then availability concerns (replicas, PodDisruptionBudgets, HPA, rolling updates), then config and networking (Ingress TLS, NetworkPolicy, RBAC). - Symptom-to-Cause Cheatsheet: Maps live cluster symptoms like CrashLoopBackOff, OOMKilled, Pending pods, and ImagePullBackOff to their most likely root causes and first diagnostic commands. - Runnable Verification: Provides kubectl dry-run, diff, QoS inspection, and rollout status commands, plus optional kube-score and kubeconform validation. - Use Case: Before merging a new Deployment, run the review to catch a missing memory limit and a liveness probe that checks the database, then apply the corrected YAML snippets provided in the findings table. ## Quick Start Review the Kubernetes manifests in ./manifests for production readiness and list blockers with corrected YAML.
