The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes ffmpeg and ffprobe via Python's subprocess module to analyze video files. It passes arguments as lists and validates that the target video path is a valid file on disk, preventing command injection attacks.\n- [EXTERNAL_DOWNLOADS]: The text_look.py script downloads font files from Google Fonts to provide typeface lookalikes. This is a well-known, trusted service. Additionally, report.py can check technique names via the author's API (edit.trysometh.ing), which is a legitimate vendor resource.\n- [DYNAMIC_EXECUTION]: A dynamic import (__import__('os')) is used in analyze.py to check for a debug environment variable. This usage is static and does not involve untrusted user input.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted video files. It mitigates potential injection or misinterpretation by providing the AI agent with specific instructions to treat automated measurements (JSON output) as hints and to prioritize manual inspection of the generated frame sheets (images) for confirmation.

video-breakdown