The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses subprocess.run in scripts/stitch_videos.py and scripts/create_title_slides.py to execute ffmpeg and ffprobe. Commands are constructed using argument lists, which prevents shell injection.
[EXTERNAL_DOWNLOADS]: The scripts/gdrive_video_download.py script downloads video files from Google Drive using the pydrive2 library. As Google Drive is a well-known service, these downloads are considered safe within the intended context of the skill.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface (Category 8). 1. Ingestion points: Untrusted video filenames are retrieved from Google Drive in scripts/gdrive_video_download.py and processed in scripts/generate_youtube_description.py. 2. Boundary markers: The generated YouTube descriptions in scripts/generate_youtube_description.py do not utilize delimiters or specific warnings to ignore instructions embedded in segment titles. 3. Capability inventory: The skill has significant capabilities including executing system commands via subprocess.run (FFmpeg) and writing files to the local file system. 4. Sanitization: Filenames are cleaned using basic regex to remove episode prefixes and illegal characters, but this does not prevent instructions embedded in titles from being included in the generated markdown output.

video-production