gke-cost-optimization
Pass
Audited by Gen Agent Trust Hub on Jul 22, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- Kubernetes Command Execution: The skill provides instructions for using
kubectlto manage cluster resources, such as applyingResourceQuotaandVerticalPodAutoscalermanifests. This is standard behavior for a GKE management skill. - `kubectl apply -f
- <<EOF ... EOF` (used for manifest application)
kubectl get vpa {deployment_name}-vpa(used for resource inspection)- Resource Templating: The skill uses placeholders like
{namespace},{deployment_name}, and{image_name}to customize manifests and commands. This represents a potential interface for untrusted data, although it is a common pattern for infrastructure-as-code templates. - Ingestion points: Placeholders in
SKILL.md(e.g.,{namespace}). - Boundary markers: The skill uses heredocs (
EOF) to delimit manifest content, providing structural boundaries for the interpolated data. - Capability inventory: Access to
kubectl,gcloud, andbqcommands for cluster and billing management. - Sanitization: Sanitization is assumed to be handled by the executing agent when populating these template variables.
- Cloud Provider Integration: The skill references official Google Cloud services and tools, including BigQuery (
bq), GCloud Billing, and Autopilot features, which are appropriate for GKE cost optimization.
Audit Metadata