short-drama-production
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user themes and drama ideas to generate content (scripts, art descriptions) that is subsequently passed to external generation tools.
- Ingestion points: User themes, project details (title, genre, duration), and iterative feedback provided in SKILL.md.
- Boundary markers: Absent in code, but conceptually implemented via mandatory approval gates between each phase of production.
- Capability inventory: Writing project files to the local file system; invoking sub-skills for image generation, video generation, and GPU instance management.
- Sanitization: Content is validated through manual user review at each 'Confirmation Gate' before being used as input for the next stage.
- [SAFE]: The 'Engineering Redlines' section mandates that the agent must never write sensitive information like SSH passwords or API tokens into project files, and must perform pre-execution capability checks (e.g., verifying the presence of required environment variables) before proceeding with resource-intensive tasks.
Audit Metadata