prop-asset
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references design guidelines, visual case studies, and a style library from a public GitHub repository (
freestylefly/awesome-gpt-image-2). These resources are used strictly as reference material for visual inspiration and utilize pinned commit hashes to ensure consistent content. - [INDIRECT_PROMPT_INJECTION]: The skill operates by ingesting user-provided scripts and external reference text to maintain asset continuity across scenes.
- Ingestion points: Processes user-supplied script text (described in SKILL.md) and retrieves documentation from external GitHub URLs (cinematic-image-direction.md).
- Boundary markers: The skill does not implement explicit boundary markers or XML tags to isolate external script content from its core instructions.
- Capability inventory: The skill is limited to text-based prompt generation and creating JSON metadata. It does not possess capabilities for shell execution, file system modification, or network data exfiltration.
- Sanitization: No specific sanitization or filtering of external input is defined before it is interpolated into image generation prompts.
- Mitigation: The skill includes a human-in-the-loop requirement, specifying that assets must be reviewed and approved by the user before being finalized.
Audit Metadata