ci-cd-acrazie
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill instructions demonstrate a strong security posture, explicitly prohibiting the handling of secret values and warning against common pitfalls like host Docker socket access.
- [INDIRECT_PROMPT_INJECTION]: The skill operates on untrusted repository metadata and configuration files. It mitigates this risk through strict 'read-only' discovery rules and requiring human verification of all proposed changes.
- [CREDENTIALS_UNSAFE]: Test fixtures within the skill contain illustrative hardcoded credentials (e.g., 'insecure-example' in a Jenkinsfile). These are provided specifically as diagnostic targets for the skill's modernization capabilities and do not represent a vulnerability in the skill's own implementation.
Audit Metadata