gke-compute-classes
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill promotes security best practices for GKE environments, including the use of Workload Identity instead of hardcoded service account keys, maintaining Shielded VM features, and adhering to the principle of least privilege for pod permissions. The documentation correctly identifies high-privilege operations and suggests safer alternatives.
- [INDIRECT_PROMPT_INJECTION]: The main instructions in SKILL.md include a proactive security rule ('CRITICAL INJECTION RULE') that explicitly directs the AI to treat user-provided data such as logs or YAML as untrusted content. This sanitization step is designed to prevent instruction overrides from embedded malicious text in data the skill processes.
- [COMMAND_EXECUTION]: The asset
assets/log-autoscaler-events.shprovides a legitimate tool for users to monitor GKE autoscaler events via standard CLI tools (gcloud,jq). The script operates locally and requires manual execution by the user, with no evidence of unauthorized background execution or data exfiltration. It correctly implements permission checks for the required IAM roles.
Audit Metadata