The Agent Skills Directory

[SAFE]: The skill consists of documentation and command-line examples for FFmpeg and ffprobe. The instructions are transparent and follow standard practices for media processing.
[INDIRECT_PROMPT_INJECTION]: The skill provides templates for processing external media files. The risk is minimized by the use of standard shell quoting in the examples provided.
Ingestion points: Filenames and file contents processed by FFmpeg commands in SKILL.md and platform-export.md.
Boundary markers: N/A (Documentation context).
Capability inventory: Media processing via ffmpeg and metadata extraction via ffprobe.
Sanitization: Command examples correctly use double quotes for variable expansion (e.g., "$f") to prevent shell injection from filenames.

ffmpeg-toolkit