azure-kubernetes-automatic-readiness

Pass

Audited by Gen Agent Trust Hub on Jun 27, 2026

Risk Level: SAFE
Full Analysis
  • Indirect Prompt Injection Surface: The skill evaluates external Kubernetes manifests (processed in SKILL.md under Offline Mode), which creates a potential entry point for indirect instructions. This is mitigated by a mandatory human-in-the-loop requirement where users must explicitly review and approve all suggested YAML changes before they are applied.
  • Official Tool Integration: The skill integrates with the mcp_azure_mcp_aks tool for cluster discovery and assessment. All network interactions and documentation references (e.g., aka.ms/azure-mcp-setup) point to official Microsoft resources and operate within expected vendor boundaries.
  • Sensitive Data Protection: A core guardrail (Guardrail 2) explicitly instructs the agent to never transmit or display sensitive information such as secrets, connection strings, or service account tokens, ensuring data privacy during the assessment.
  • Administrative Boundary Controls: The skill is strictly limited to a read-only assessment role. It includes explicit instructions to avoid modifying cluster state or running commands like kubectl apply without direct user confirmation of generated diffs.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 27, 2026, 05:18 AM
Security Audit — agent-trust-hub — azure-kubernetes-automatic-readiness