video-face-image-replacement

Pass

Audited by Gen Agent Trust Hub on Oct 19, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses ffmpeg and ffprobe for legitimate media processing tasks, such as extracting frames, checking video duration, and muxing audio. These are standard operations for a video-related tool.
  • [EXTERNAL_DOWNLOADS]: The documentation references several reputable external resources, including official research papers on ArXiv, the InsightFace tutorial, Meta's SAM 2 project, and well-known GitHub repositories. These are used for documentation and research purposes, and no automatic execution of untrusted remote code was found.
  • [DATA_EXPOSURE_EXFILTRATION]: The skill processes local video files and generates image assets in a local directory (artifacts/). It does not demonstrate any patterns of sensitive data exfiltration or unauthorized network communication.
  • [SAFE]: The skill follows security best practices by recommending that source videos never be overwritten and that sensitive media remain uncommitted unless explicitly requested.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 19, 2026, 06:24 PM