Cinematic Direction
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 data to construct a detailed 'director brief' which is then used by media generation tools. This ingestion pattern creates an attack surface for indirect prompt injection.
- Ingestion points: User-provided inputs defined in
SKILL.md, including subject, action, environment, and emotion. - Boundary markers: The prompt template for the director brief uses labels like [SCENE] and [SUBJECT] but lacks clear delimiters or 'ignore' instructions to isolate user input from the rest of the prompt.
- Capability inventory: The skill has access to sensitive capabilities including
media.generate_videoandmedia.upload_filethrough the referencedmuapi-video-tools.mdscript. - Sanitization: The instructions do not define any sanitization, validation, or escaping logic for the user-supplied creative inputs.
Audit Metadata