normalize
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- Prompt Injection (SAFE): The prompt uses instructional markers like 'CRITICAL' and 'IMPORTANT' solely to emphasize design principles and UX best practices, with no evidence of jailbreak or bypass attempts. |
- Data Exposure (SAFE): Access is limited to grep-based searches for design system documentation within the project repository. No sensitive files or credentials are targeted. |
- Indirect Prompt Injection (SAFE): Vulnerability surface identified: (1) Ingestion point: 'feature' argument; (2) Boundary markers: Absent; (3) Capabilities: grep, file modification; (4) Sanitization: Absent. Risk is negligible within the intended design normalization context. |
- Execution Safety (SAFE): The skill specifies standard frontend refactoring tasks. It does not request elevated privileges, package installations, or remote code execution.
Audit Metadata