panfangmin

You can diagnose Pod, node, network, storage, workload, and autoscaling failures; query logs, events, and metrics; correlate AOM alarms; manage clusters, node pools, addons, and workloads; run k6 pressure tests; forecast capacity; optimize costs; scan availability risks; plan upgrades and migrations; and execute preview-confirmed remediation actions.

They target SREs, platform engineers, and cloud operations teams running Kubernetes workloads on Huawei Cloud CCE. Typical users handle incident response, daily cluster inspections, on-call reporting, capacity planning, cost reduction, and cluster upgrade or migration projects across single or multi-cluster environments.

Skills rely on the hcloud CLI (KooCLI) for Huawei Cloud API calls, kubectl with the kubectl-cce plugin for read-only cluster evidence, and k6 for load generation. A dedicated installer skill sets up kubectl and kubectl-cce when missing. Some analyzers use a Python SDK dispatcher for cloud service queries.

Most skills are strictly read-only, gathering evidence for diagnosis and reports. The auto-remediation runner performs mutations like drain, cordon, scale, restart, and hibernate only through a preview-first, confirm-required flow with post-execution verification; confirmation is never auto-applied, keeping changes under explicit user control.

Yes. The cost advisor identifies idle resources, oversized requests, and low-utilization nodes with HPA and autoscaler recommendations. The capacity forecaster analyzes trends, predicts resource bottlenecks, simulates elasticity policies, and generates capacity curve reports, all as read-only analysis requiring confirmation before any change.