Avatar / UGC
Pass
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied scripts, audio recordings, and image/video references which may contain embedded instructions designed to manipulate the agent's behavior during the video generation process.
- Ingestion points: User-provided script text, audio sources, and media references collected during the 'Consent gate' phase.
- Boundary markers: Step 1 in the 'Execute' section requires freezing the script and marking lines as supplied or approved.
- Capability inventory: The skill utilizes powerful tools including media.lipsync, media.generate_video, and media.generate_speech.
- Sanitization: The instructions focus on identity drift and authorization verification but do not detail specific techniques for sanitizing or escaping the content of input scripts.
Audit Metadata