Cluster Health

Run read-only Kubernetes health checks and report cluster status with evidence. This skill works without private overlays by requiring an explicit kube context or confirmed current context. Local users may add ignored protected overlays for aliases and environment-specific checks.

When to use

• User asks to check cluster health, status, diagnostics, node status, or post-maintenance state
• Verifying cluster-wide symptoms after upgrades, reboots, Helm changes, GitOps syncs, or incidents
• Gathering read-only evidence across nodes, workloads, events, ingress, storage, logs, and policy
• Producing a short traffic-light report from Kubernetes and related observability signals

When NOT to use

• Writing or reviewing Kubernetes manifests - use kubernetes
• Writing Helm charts, Kustomize overlays, or IaC - use kubernetes or terraform
• Changing resources, restarting pods, deleting objects, or applying fixes - ask for explicit escalation
• Debugging one application deeply after the broad sweep identifies it - use the relevant domain skill