The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses subprocess.run to call FFmpeg and FFprobe for video processing, quality assurance, and metadata extraction. These operations are essential for the primary purpose of the skill and use safe argument handling to prevent injection.
[EXTERNAL_DOWNLOADS]: Automatically fetches required design fonts from well-known repositories such as Google Fonts (via jsDelivr) and GitHub to ensure consistent visual output across different environments.
[DATA_EXPOSURE]: Employs an internal guard script to prevent pipeline metadata (such as model names or speed multipliers) from being rendered onto video frames and implements strict path validation to keep file operations within the project scope.
[INDIRECT_PROMPT_INJECTION]: Processes user-provided text for scripts and captions. This attack surface is protected by content linting for platform taboos and structured artifact management, resulting in a low risk profile.
[DYNAMIC_EXECUTION]: Uses headless Chrome for high-quality text rendering in cover images. The rendering process is handled through a deterministic template system with input sanitization, avoiding arbitrary code execution.

video-editing