h3-long-video
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The workflow includes a standard shell command using
ffmpegto extract the final frame of a video file. This is a legitimate functional requirement for maintaining frame-to-frame consistency and does not involve elevated privileges, network access, or sensitive paths. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest user-provided storyboards and descriptions to generate prompts for an external video model. While it handles external input without explicit boundary markers, the risk is minimal as the output is restricted to generative prompts and localized media processing for the user's project.
Audit Metadata