cloud-architect

Pass

Audited by Gen Agent Trust Hub on Jul 31, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [SAFE]: No malicious patterns, obfuscation, or unauthorized data access mechanisms were detected in the instructions or metadata. The skill follows best practices for the domain.- [INDIRECT_PROMPT_INJECTION]: The skill possesses an inherent attack surface for indirect prompt injection based on its intended functionality.
  • Ingestion points: The skill processes external user-supplied architectural requirements and references a local resource file, resources/implementation-playbook.md.
  • Boundary markers: There are no instructions or delimiters provided to ensure the agent ignores instructions that may be embedded within these external inputs.
  • Capability inventory: The skill is capable of generating Infrastructure as Code (Terraform, CDK) and recommends executing security and infrastructure testing tools (Checkov, Terrascan, Terratest).
  • Sanitization: The skill does not include logic for sanitizing or validating requirements before they are incorporated into architectural designs or code generation.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 31, 2026, 06:17 AM
Security Audit — agent-trust-hub — cloud-architect