getting-started

GKE Infrastructure Skills Available in the Google Skills Repository

September 5, 2026 google/skills ↗

The Google Skills repository maintains eight production-ready GKE infrastructure skills spanning workload troubleshooting, security hardening, autoscaling, cluster maintenance, monitoring, AI migration, enterprise search deployment, and multi-agent security architectures.

The google/skills repository hosts self-contained SKILL markdown files that provide operational guidance for managing Google Kubernetes Engine resources. These GKE infrastructure skills reside under skills/cloud/ and contain executable gcloud and kubectl commands, architectural patterns, and troubleshooting workflows designed for production-grade cluster management.

Core Workload Management Skills

GKE Workload Troubleshooting

Located at skills/cloud/gke-workload-troubleshooting/SKILL.md, this skill diagnoses common pod failures including CrashLoopBackOff, OOMKilled, ImagePullBackOff, and Pending states. It aggregates container logs, Kubernetes events, and Cloud Logging queries to identify root causes.


# List failing pods in the current namespace

kubectl get pods --field-selector=status.phase=Failed

# Retrieve recent logs from a specific container

kubectl logs <pod-name> -c <container-name> --tail=100

# Query Cloud Logging for GKE events

gcloud logging read 'resource.type="k8s_container"' --limit=50 --format="json"

GKE Workload Security

The skills/cloud/gke-workload-security/SKILL.md file implements defense-in-depth strategies including Workload Identity, GKE Sandbox (gVisor), Shielded Nodes, Network Policies, and Pod Security Standards.

Enable Workload Identity on existing clusters:

gcloud container clusters update $CLUSTER \
  --workload-pool="${PROJECT}.svc.id.goog"

Apply restricted Pod Security Standards:

kubectl apply -f - <<EOF
apiVersion: policy/v1beta1
kind: PodSecurityPolicy
metadata:
  name: restricted
spec:
  privileged: false
  seLinux:
    rule: RunAsAny
  runAsUser:
    rule: MustRunAsNonRoot
  supplementalGroups:
    rule: MustRunAs
  fsGroup:
    rule: MustRunAs
  volumes:
  - "configMap"
  - "secret"
  - "emptyDir"
EOF

GKE Workload Scaling

Found in skills/cloud/gke-workload-scaling/SKILL.md, this skill configures Horizontal Pod Autoscaler (HPA), Vertical Pod Autoscaler (VPA), and custom metrics for elastic workloads.

Example HPA configuration targeting 65% CPU utilization:

apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: web-hpa
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: web
  minReplicas: 2
  maxReplicas: 10
  metrics:
  - type: Resource
    resource:
      name: cpu
      target:
        type: Utilization
        averageUtilization: 65

kubectl apply -f horizontal-pod-autoscaler.yaml

Cluster Lifecycle and Observability

GKE Upgrades and Maintenance

The skills/cloud/gke-upgrades/SKILL.md file provides workflows for planning and executing control-plane and node-pool upgrades. It covers release channels, maintenance windows, and rollback procedures for GKE infrastructure skills related to cluster lifecycle.

Enable automatic upgrades via the Regular release channel:

gcloud container clusters update $CLUSTER --release-channel=regular

Perform manual control plane upgrades:

gcloud container clusters upgrade $CLUSTER \
  --master --cluster-version=1.30.3-gke.1234000

GKE Alert Configuration

Located at skills/cloud/gke-alert-configuration/references/gke_configuration_prerequisites.md, this skill establishes Cloud Monitoring alerts for CPU saturation, memory pressure, and node-pool health events.

Create infrastructure health alerts:

gcloud monitoring policies create \
  --policy-from-file=alerts/gke-cpu-saturation.json

Advanced Architecture Skills

Guided GKE AI Migration

The skills/cloud/google-cloud-solution-guided-gke-ai-migration/SKILL.md skill migrates AI workloads from Cloud Run or Gemini API to self-hosted inference on GKE. It covers cluster discovery, Workload Identity configuration, GPU/TPU provisioning, and Gateway API exposure.

Discover clusters and deploy inference services:


# List available clusters

gcloud container clusters list

# Check accelerator availability in target zone

gcloud compute accelerator-types list --filter="zones:$ZONE"

# Deploy vLLM inference service

kubectl apply -f assets/vllm-deployment.yaml.tmpl

RAG Enterprise Search on GKE and AlloyDB

Found in skills/cloud/google-cloud-solution-rag-enterprise-search-gke-sqldb/SKILL.md, this skill deploys Retrieval-Augmented Generation stacks using GKE Autopilot clusters with Ray workers, Cloud Storage FUSE, and AlloyDB vector search.

Provision the RAG infrastructure:


# Create GKE Autopilot cluster

gcloud container clusters create-auto rag-cluster \
  --region=$REGION --project=$PROJECT

# Deploy Ray workers for chunking operations

kubectl apply -f assets/ray-cluster.yaml.tmpl

# Provision AlloyDB instance with private IP

gcloud alloydb instances create rag-db --region=$REGION --private-ip

Multi-Agent Security with GKE Internal Load Balancing

The skills/cloud/google-cloud-solution-multi-agent-security/SKILL.md skill configures private GKE internal load balancers for MCP tool servers, enabling secure egress in multi-agent architectures.

Configure the internal load balancer:


# Reserve static internal IP

gcloud compute addresses create gke-ilb-ip \
  --region=$REGION --subnet=$SUBNET --addresses=10.0.1.50

# Deploy internal load balancer service

kubectl apply -f - <<EOF
apiVersion: v1
kind: Service
metadata:
  name: mcp-internal-lb
spec:
  type: LoadBalancer
  loadBalancerIP: 10.0.1.50
  selector:
    app: mcp-tool
  ports:
  - port: 443
    targetPort: 8443
EOF

Summary

The Google Skills repository provides eight specialized GKE infrastructure skills for comprehensive cluster management:

Frequently Asked Questions

What are GKE infrastructure skills in the Google Skills repository?

GKE infrastructure skills are self-contained markdown files stored in skills/cloud/ that provide executable operational guidance for Google Kubernetes Engine. Each skill contains specific gcloud and kubectl commands, architectural patterns, and troubleshooting procedures designed for agent consumption or direct operator use.

How do I troubleshoot CrashLoopBackOff using the GKE infrastructure skills?

According to skills/cloud/gke-workload-troubleshooting/SKILL.md, first identify failing pods with kubectl get pods --field-selector=status.phase=Failed, then inspect logs using kubectl logs <pod-name> --tail=100, and correlate events with Cloud Logging queries targeting resource.type="k8s_container".

Which skill handles AI workload migration to GKE?

The Guided GKE AI Migration skill at skills/cloud/google-cloud-solution-guided-gke-ai-migration/SKILL.md manages migration from Cloud Run or Gemini API to self-hosted inference, including GPU/TPU node pool configuration and Gateway API exposure for external traffic.

Where are the GKE skill files located in the repository?

All GKE infrastructure skills reside under the skills/cloud/ directory in the google/skills repository, with each skill stored in its own subdirectory containing a SKILL.md file and optional references/ folder for supplementary documentation such as version tables and CSI driver guides.

Have a question about this repo?

These articles cover the highlights, but your codebase questions are specific. Give your agent direct access to the source. Share this with your agent to get started:

Share the following with your agent to get started:
curl -s "https://instagit.com/install.md"
Works with
Claude Codex Cursor VS Code OpenClaw Any MCP Client

Maintain an open-source project? Get it listed too →