multi-clip-cinematic-generation
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill directs the agent to execute
ffprobeto analyze and verify the integrity of downloaded video assets. - [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided manifests and external media assets, creating a surface for instructions hidden in file metadata or structured manifest fields.
- Ingestion points: Reference images, predecessor video tails, and sequence manifests.
- Boundary markers: Visual receipt-checks and frame stability inspections are mandated.
- Capability inventory: Execution of
ffprobeand API interactions with video generation providers. - Sanitization: The skill relies on visual and technical inspection rather than automated sanitization of incoming text-based metadata.
- [EXTERNAL_DOWNLOADS]: The skill retrieves media content from external AI generation services, specifically referencing the MiniMax H3 model and the Fal.ai platform.
Audit Metadata