Diagnose GKE Cloud TPU training throughput drops and step-time regressions (15%+ TPU duty-cycle drop) using ML Diagnostics Workload Monitoring (`gcloud alpha mldiagnostics monitored-events` / `hypercomputecluster.googleapis.com/v1alpha`) and 1-minute Cloud Monitoring system metrics (`kubernetes.io/node/accelerator/*`).
Install
npx skills add google/skills --skill gke-ai-troubleshooting-tpu-performance-degradationnpx skills add google/skills --skill gke-ai-troubleshooting-tpu-performance-degradation -a claude-code -gnpx skills add google/skills --skill gke-ai-troubleshooting-tpu-performance-degradation -a codex -ggemini skills install https://github.com/google/skills.git --path skills/cloud/gke-ai-troubleshooting-tpu-performance-degradationgh skill install google/skills gke-ai-troubleshooting-tpu-performance-degradationgit clone --depth 1 https://github.com/google/skills /tmp/google__skills
mkdir -p ~/.claude/skills ~/.agents/skills
cp -R /tmp/google__skills/skills/cloud/gke-ai-troubleshooting-tpu-performance-degradation ~/.claude/skills/gke-ai-troubleshooting-tpu-performance-degradation # Claude Code
cp -R /tmp/google__skills/skills/cloud/gke-ai-troubleshooting-tpu-performance-degradation ~/.agents/skills/gke-ai-troubleshooting-tpu-performance-degradation # Codex, Cursor, Gemini CLI, Copilot, OpenCode