The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes ffmpeg and ffprobe via the Python subprocess module in several scripts (clip_cut.py, add_end_card.py, burn_subtitles.py, and make_vertical.py). These executions use the secure list-of-arguments format rather than raw shell strings, which effectively prevents command injection vulnerabilities from user-supplied file paths or segments.
[DATA_EXPOSURE]: The skill processes media files (.mp4), subtitles (.srt), and text transcripts (.txt) stored within the local project's working/ and output/ directories. It does not attempt to access sensitive system files, environment variables, or private user data (e.g., .ssh or .aws directories).
[SAFE]: No evidence of prompt injection, obfuscation, or remote code execution was found. The workflow is consistent with its stated purpose of video editing and asset generation.

short-video-workflow