imagegen-frontend-web
Pass
Audited by Gen Agent Trust Hub on Jul 5, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [PROMPT_INJECTION]: The skill uses strong imperative language, such as 'HARD OUTPUT RULE,' 'CRITICAL OUTPUT RULE,' and 'No exceptions,' and explicitly states that these instructions 'override any model default.' These markers are frequently associated with prompt injection attempts to bypass safety filters; however, in this context, they are clearly used as a standard prompt engineering technique to ensure the model adheres to specific formatting requirements (generating separate images for each section) instead of collapsing output. This behavior is benign and aligned with the skill's creative purpose.
- [NO_CODE]: The provided skill consists exclusively of Markdown-formatted instructions and YAML metadata. No code files, shell scripts, or dependency manifests (e.g., requirements.txt, package.json) are present. The skill operates solely by providing context and guidelines to the agent's pre-existing image generation capabilities.
Audit Metadata