What problem does it solve? Performing recovery operations on Huawei Cloud CCE clusters (rollbacks, node drains, reboots, scaling) is risky when done blindly. This Skill converts remediation intent into reviewable execution plans that require explicit user confirmation before any mutation, and verifies cluster health afterward. ## Core Features & Use Cases - Preview-First Mutation Guardrails: Every write action (rollback, scale, drain, reboot, delete, hibernate, HSS vulnerability status change) returns a preview with risk level (R1/R2/R3) and refuses to execute until the user explicitly confirms with confirm=true. - Auto-Remediation Orchestration: The huawei_auto_remediation_run action diagnoses a failing Deployment, rolls back to a previous revision when the root cause is CrashLoop, image pull, probe, or startup command failure, waits for recovery, and produces a Markdown execution report. - Broad Action Coverage: Supports workload rollback/scale/resize/delete, node cordon/uncordon/drain, ECS reboot/start/stop, node pool resize, cluster hibernate/awake, EIP bind/unbind, and HPA configuration, plus read-only verification actions. - Use Case: After a root-cause analysis shows a new Deployment revision is in CrashLoopBackOff, use this Skill to preview a rollback to the previous revision, confirm the plan, execute it, and verify pods return to healthy status. ## Quick Start Ask the assistant to preview a rollback of Deployment app-server in namespace default on your CCE cluster, then confirm the plan so it executes and verifies pod health.
