ugc-fashion
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill templates interpolate user-supplied text into prompts for a video generation API.\n
- Ingestion points: SKILL.md contains numerous placeholders such as [OUTFIT DESCRIPTION], [GARMENT], [SIZE], and [STYLING STEPS] that ingest untrusted user data.\n
- Boundary markers: The templates use brackets for placeholders, but they lack explicit instructions or delimiters to ensure the agent ignores potential instructions embedded within the user-supplied content.\n
- Capability inventory: The skill instructions in SKILL.md direct the agent to use the generated strings as the 'prompt' field in a POST /videos/generate tool call.\n
- Sanitization: No input validation, filtering, or sanitization logic is defined in the skill to handle potentially malicious user input.
Audit Metadata