brandkit
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill features an intake process where it ingests user-supplied briefs and external visual references to inform the generation of brand assets. This constitutes an entry point for untrusted data that influences downstream AI tool usage.
- Ingestion points: User-provided brand briefs, category descriptions, and external reference files (SKILL.md, Intake section).
- Boundary markers: The skill uses natural language instructions to guide the agent's behavior but does not implement formal technical delimiters or boundary markers for the user input.
- Capability inventory: The skill utilizes an internal image-generation tool to create the final visual identity boards (SKILL.md, Generate section).
- Sanitization: While the instructions explicitly forbid copying unique assets or slogans from references, there is no technical sanitization or escaping of the user-provided text before it is interpolated into the generation prompt.
Audit Metadata