kube-state-metrics

If you're running kube-prometheus-stack, you already have:

• kube_pod_status_phase{namespace,pod} — pod lifecycle (filter by pod=~"<agent>-pod")
• container_cpu_usage_seconds_total{pod=~"...pod$"} — per-agent CPU
• container_memory_working_set_bytes{pod=~"...pod$"} — per-agent memory
• kube_persistentvolumeclaim_resource_requests_storage_bytes{namespace} — workspace storage

We don't duplicate these in komputer-ai's metrics. Use them for resource-utilization views in your dashboards.