The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides documentation for using ffmpeg and ffprobe and includes a Python utility for media auditing.
scripts/media_probe.py uses subprocess.run to invoke the ffprobe executable.
The implementation uses an argument list and avoids the shell (shell=False), which is the recommended method to prevent shell injection attacks. The commands documented in SKILL.md are standard media processing utilities.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external media files and interpret their metadata.
Ingestion points: The media_probe.py script (called by the agent) ingests the output of ffprobe in JSON format.
Boundary markers: The script employs a structured internal report schema to isolate external data and uses explicit logic to compare results against expectations.
Capability inventory: The skill allows for local file reading and metadata extraction (via ffprobe) and media transcoding (via ffmpeg).
Sanitization: The Python helper implements strict data parsing using the json module, type validation, and finiteness checks for all numeric metadata, protecting the agent from malformed metadata fields.

ffmpeg-media-finishing