gke-compute-classes
Pass
Audited by Gen Agent Trust Hub on Jun 25, 2026
Risk Level: SAFE
Full Analysis
- Diagnostic Script for GKE Events: The skill includes a shell script (
assets/log-autoscaler-events.sh) designed to monitor GKE cluster autoscaler visibility logs. The script uses standard tools (gcloud,jq) to read logs from the user's Google Cloud project and provides real-time feedback in the terminal. It does not perform any external network operations or sensitive data exfiltration. - Safety Instructions for AI Agent: The instructions in
SKILL.mdinclude specific rules that direct the AI to handle user-provided content (like logs or YAML) as untrusted data. This is a security best practice designed to mitigate potential indirect prompt injection attacks by ensuring the AI prioritizes its core safety and configuration rules over directives embedded in user-supplied data. - Resource and Schema Validation: The skill enforces strict schema rules for GKE ComputeClass resources and requires the AI to provide warnings about resource alignment with Committed Use Discounts (CUDs) and Reservations, which helps prevent configuration errors and unexpected costs.
Audit Metadata