ugc-ad-remake
Pass
Audited by Gen Agent Trust Hub on Nov 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external reference material like transcripts and images to generate prompts. This ingestion of untrusted data represents an indirect prompt injection surface. The risk is considered negligible because the agent's output is restricted to text prompts in code blocks, and it lacks dangerous capabilities like shell execution or network access.
- Ingestion points: User-supplied videos, screenshots, transcripts, and images referenced in SKILL.md.
- Boundary markers: Spoken lines are wrapped in double quotes in templates within references/clip-prompts.md.
- Capability inventory: The skill only generates text; no code execution, file writing, or network operations are present.
- Sanitization: The instructions do not specify any filtering or sanitization for input data.
Audit Metadata