Gke Scaling

This handles GKE autoscaling across the stack: HPA for pod replicas, VPA for rightsizing requests, and Node Auto Provisioning for cluster capacity. The golden path defaults to OPTIMIZE_UTILIZATION profile with VPA enabled and NAP turned on, which is aggressive about cost but solid for most workloads. You get a practical rightsizing workflow (run VPA in Off mode for a day, apply recommendations with 20% buffer) and clear guidance on avoiding HPA/VPA metric conflicts. The MCP tool callouts are helpful since some operations like kubectl scale have no MCP equivalent. If you're running production GKE and still manually setting replica counts or guessing at resource requests, this will save you money and headaches.

Install to Claude Code

npx -y skills add google/skills --skill gke-scaling --agent claude-code