agency-cloud-security-architect
Pass
Audited by Gen Agent Trust Hub on Jul 6, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill's documentation includes GitHub Actions templates that reference external actions from well-known sources. These include official actions from GitHub (
actions/checkout), AWS (aws-actions/configure-aws-credentials), and established security vendors like Bridgecrew (bridgecrewio/checkov-action), Gitleaks (gitleaks/gitleaks-action), and Aqua Security (aquasecurity/trivy-action). These references are standard for cloud-native deployment pipelines and are used here to demonstrate security best practices. - [COMMAND_EXECUTION]: The skill provides example commands for Terraform and GitHub Actions (
terraform apply,gitleaks, etc.). These are static examples provided within the markdown documentation for educational and architectural purposes, not executable scripts that run automatically. - [PROMPT_INJECTION]: The skill uses instructional language such as 'Critical Rules You Must Follow' and defines a specific persona. These instructions are focused on maintaining security posture (e.g., 'Never allow long-lived credentials', 'Always encrypt data') and do not attempt to bypass the AI agent's underlying safety filters or override system-level instructions.
- [DATA_EXFILTRATION]: No patterns of data exfiltration or unauthorized network operations were found. The skill specifically advocates for the protection of sensitive data through encryption, centralized logging, and the use of short-lived credentials via OIDC federation.
Audit Metadata