generate-reference-inspired-brand-worlds
Pass
Audited by Gen Agent Trust Hub on Jul 19, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill prioritizes intellectual property safety by explicitly instructing the agent to avoid reproducing protected signature elements, such as logos, slogans, and distinctive motifs from reference images.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of user-provided images which are used as visual references for the generation pipeline.
- Ingestion points: Visual references provided by the user, labeled as
Image 1andImage 2inSKILL.md. - Boundary markers: The instructions mandate clear labeling of input images to differentiate them from the core prompt logic.
- Capability inventory: The skill utilizes an internal
imagegentool and manages files within the project workspace, including performing git operations. - Sanitization: Includes a mandatory validation checklist to verify the spelling of brand names and ensure that generated content adheres to the requested originality constraints.
- [COMMAND_EXECUTION]: The skill contains instructions for the agent to use git commands (
Stage and commit) for managing assets within the repository. These operations are limited to project maintenance and do not involve executing arbitrary or remote code.
Audit Metadata