ffmpeg-opencv-integration

Pass

Audited by Gen Agent Trust Hub on Jun 13, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill contains no evidence of malicious patterns such as prompt injection, data exfiltration, or obfuscation.
  • [COMMAND_EXECUTION]: The skill provides Python code snippets using the subprocess module to execute ffmpeg and ffprobe binaries. These operations are the primary intended purpose of the skill and are documented clearly for video decoding, encoding, and metadata extraction.
  • [EXTERNAL_DOWNLOADS]: The documentation references several well-known and standard Python libraries for video processing, including ffmpegcv, vidgear, decord, av (PyAV), and ffmpeg-python. These are established tools in the developer community and are used here for their intended functionality.
  • [PROMPT_INJECTION]: As the skill is designed to process external video data, it theoretically possesses an indirect prompt injection surface if an agent were to process untrusted video metadata or embedded text. This is a characteristic common to all media processing skills and no specific exploitable patterns were identified.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 13, 2026, 03:39 AM
Security Audit — agent-trust-hub — ffmpeg-opencv-integration