| name | cluster-audit |
| description | Full cluster state audit. Use when asked for cluster status, health check, or infrastructure audit. Produces comprehensive report of all nodes, pods, services, PVCs, GPUs, and issues. |
| allowed-tools | Bash, Read |
Cluster Audit
Run a comprehensive audit of the Kaizen K8s cluster. Execute ALL of these in sequence:
- Nodes:
kubectl get nodes -o wide
- All Pods:
kubectl get pods -A -o wide
- Problem Pods:
kubectl get pods -A --field-selector=status.phase!=Running,status.phase!=Succeeded
- Services:
kubectl get svc -A
- PVCs:
kubectl get pvc -A
- GPU Allocation:
kubectl describe nodes | grep -A5 "nvidia.com/gpu"
- Recent Events:
kubectl get events -A --sort-by=.lastTimestamp --no-headers | tail -20
- Inference Health:
curl -s --max-time 3 http://10.10.10.10:30000/v1/models and curl -s --max-time 3 http://10.10.10.10:30001/v1/models
- Cognitive Health:
curl -s --max-time 3 http://10.10.10.10:30800/health and curl -s --max-time 3 http://10.10.10.10:30810/health
Produce a summary table:
Flag any CrashLoopBackOff, OOMKilled, or Pending pods. Note GPU utilization vs allocation.
