generate-reference-inspired-brand-worlds
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill's primary purpose is image-based brand analysis and prompt generation. It follows best practices for managing user-supplied visual references and emphasizes the creation of original work through a 'similarity dial' system and explicit exclusion of protected signature elements.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-provided images as primary references, which is an inherent part of its workflow. This presents a potential surface for indirect prompt injection where instructions could be embedded in image content or metadata.
- Ingestion points: User-supplied visual references (
Image 1andImage 2) are loaded and inspected inSKILL.md. - Boundary markers: The skill uses a structured 'prompt spine' in
SKILL.mdto encapsulate the generated brand details, though it does not explicitly define 'ignore' boundaries for text extracted from images. - Capability inventory: The agent can generate images via the
imagegentool, persist files to the project workspace, and perform git operations (staging/committing) as noted in the 'Save and hand off' section ofSKILL.md. - Sanitization: The skill provides an 'Extract visual DNA' and 'Validate every output' step to manually verify that the generated results conform to safety and originality rules, but lacks automated sanitization of data derived from the input images.
Audit Metadata