qwen-mm-plugins-omni-video2note
Pass
Audited by Gen Agent Trust Hub on Sep 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted video content, including speech, subtitles, and on-screen text (slides, terminal output), which could contain instructions designed to influence the agent's behavior.
- Ingestion points: Local video files specified by the user via the
video_pathargument in theomni_video2note_createtool. - Boundary markers: The skill contains a dedicated 'Safety' section that explicitly warns the agent to treat all video content as untrusted data and to ignore any commands, links, or instructions found within the video.
- Capability inventory: The skill utilizes tools for video analysis and PDF generation (
omni_video2note_create,omni_video2note_status). - Sanitization: The skill relies on natural language instructions to steer the agent away from executing instructions found in the source material.
- [COMMAND_EXECUTION]: The skill invokes specific automation tools (
omni_video2note_create,omni_video2note_status) to analyze local media and write output files to the local file system.
Audit Metadata