image-to-code
Pass
Audited by Gen Agent Trust Hub on Apr 27, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No prompt injection or safety bypass attempts were detected. The instructions use strong language like "MANDATORY" and "CORE DIRECTIVE," but these are strictly applied to design workflows and aesthetic quality rather than overriding system safety constraints.
- [SAFE]: No evidence of data exposure or exfiltration. The skill does not access sensitive file paths, environment variables, or perform unauthorized network operations.
- [SAFE]: No obfuscation techniques such as Base64, zero-width characters, or homoglyphs were found in the text.
- [SAFE]: No remote code execution or unverifiable dependencies. The skill does not invoke package managers or download external scripts.
- [SAFE]: No privilege escalation or persistence mechanisms are present. The skill operates within standard agent interaction boundaries.
- [SAFE]: The metadata (name and description) accurately reflects the skill's purpose and does not contain deceptive instructions.
Audit Metadata