# Gke Productionize

> Use when assessing and preparing GKE clusters and workloads for production readiness.

- Skill: `loopyluci/gke-productionize` (Agent Skill)
- Install (CLI): `npx skillmds@latest add loopyluci/gke-productionize`
- Raw SKILL.md: https://api.skillmd.com/api/skills/loopyluci/gke-productionize/raw
- Safety review: pending (external: skill-scanner PASS, skillspector PASS)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: LoopyLuci (https://skillmd.com/u/loopyluci)
- Updated: 2026-09-08
- Page: https://skillmd.com/skills/loopyluci/gke-productionize

---


# GKE Productionize Skill

Orchestrates comprehensive production readiness reviews for GKE clusters and workloads across scalability, security, reliability, observability, backup/DR, and cost optimization.

## Discovery Phase

```bash
# Cluster discovery
gcloud container clusters describe {cluster_name} --location {location} --project {project}

# Workload discovery
kubectl get deployment {app_name} -n {namespace} -o yaml
kubectl get hpa -n {namespace}
kubectl get pdb -n {namespace}
kubectl get networkpolicy -n {namespace}
```

## Production Readiness Areas

| Area | Action | Key Skill |
|------|--------|-----------|
| Scalability | Configure HPA, VPA, resource limits | `gke-workload-scaling` |
| Observability | Cloud Logging, Monitoring, Managed Prometheus | `gke-observability` |
| Reliability | Regional clusters, PDBs, health probes | `gke-reliability` |
| Security | Workload Identity, Network Policies, Shielded Nodes | `gke-platform-security` |
| Backup/DR | Backup for GKE, restore procedures | `gke-backup-dr` |
| Cost | Rightsizing, quotas, Spot VMs | `gke-cost-optimization` |

## Scoring

After assessment, provide RAG (Red/Amber/Green) status for each area with an overall readiness score.

## Common Pitfalls

- **Skipping discovery**: Always run discovery commands before making recommendations
- **Missing security basics**: Always check Pod Security Standards, dedicated service accounts, and NetworkPolicies
- **No backup plan**: Ensure Backup for GKE is configured for stateful workloads
- **Single-zone clusters**: Production clusters should be regional (multi-zone)

## Verification Checklist

- [ ] Cluster details gathered (Autopilot vs Standard, release channel)
- [ ] Workload resource requests/limits configured
- [ ] HPA and PDB configured for critical workloads
- [ ] Health probes (liveness, readiness, startup) configured
- [ ] Network policies enforce least-privilege
- [ ] Workload Identity configured for GCP API access
- [ ] Backup for GKE configured