06-k8s-triage

Pass

Audited by Gen Agent Trust Hub on Jun 21, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill implements a strictly read-only workflow for Kubernetes cluster investigation, including explicit prohibitions against commands that modify cluster state such as apply, delete, patch, or scale.
  • [SAFE]: Data protection measures are integrated into the instructions, specifically directing the agent to avoid printing the values of sensitive environment variables like secrets, tokens, and passwords during pod diagnostics.
  • [SAFE]: No remote code execution or external network operations were detected. All triage actions are performed locally using standard command-line tools.
  • [SAFE]: The skill uses a discovery-driven workflow where target resource names are retrieved directly from the cluster prior to inspection, reducing the risk of processing malicious inputs.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 21, 2026, 01:58 AM
Security Audit — agent-trust-hub — 06-k8s-triage