shot-specifier
Pass
Audited by Gen Agent Trust Hub on Jun 19, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill demonstrates safe behavior, focusing on creative workflow automation without attempting unauthorized data access, network exfiltration, or system modification.
- [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection as it processes untrusted data from a scene inventory and a keyword library to construct prompts for image and video generation models.
- Ingestion points: Ingests project-specific metadata and keywords from
{project}_scene_inventory.mdand{project}_prompt_keywords.md(SKILL.md Phase 1). - Boundary markers: Uses structured prompt tags such as
[STYLE],[FILMSTOCK], and[SCENE]to delimit injected content (SKILL.md Phase 7). - Capability inventory: Utilizes the
nanobananaMCP for image generation and performs file system writes to create storyboard frames and manifest files (SKILL.md Phases 5 and 8). - Sanitization: Content from the keyword library is copied verbatim, which is the intended behavior for maintaining production consistency, though it lacks input validation.
Audit Metadata