This handles GKE cost optimization: rightsizing pods with VPA recommendations, configuring Spot VMs for 60-90% savings on fault-tolerant workloads, and selecting the right machine families through ComputeClasses. It walks you through the golden path defaults already baked into Autopilot clusters, then shows you where to optimize further. The Spot guidance is solid, including explicit warnings about preemption and which workloads are suitable. Also covers CUDs, dev cluster management, and has good tables comparing machine families by cost. If you're running GKE and want to cut your bill without compromising reliability, this gives you the playbook.
npx -y skills add google/skills --skill gke-cost --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
juliusbrussee/caveman
mattpocock/skills
obra/superpowers
forrestchang/andrej-karpathy-skills
vercel-labs/skills