seedance-cinematic-film
Pass
Audited by Gen Agent Trust Hub on Sep 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill purely contains instructional content and prompt templates for video generation. No unauthorized network activity, file access, or code execution patterns were found.
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-defined scene descriptions to generate structured prompts for the higgsfield_generate tool. While it lacks explicit delimiters for user input, the rigid 'six-block' prompt architecture serves as a natural constraint, mapping user intent to specific cinematic parameters and minimizing the risk of unintended instruction following. Evidence: Ingestion point (user requests in SKILL.md), capabilities (higgsfield_generate tool in SKILL.md), boundaries (absent), sanitization (implicit structural transformation in SKILL.md).
Audit Metadata