The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/douyin_extract.py uses subprocess.run to invoke curl for fetching video metadata via Server-Side Rendering (SSR) pages. This is a common pattern for scraping metadata from dynamic websites like Douyin without requiring browser automation.
[COMMAND_EXECUTION]: The script scripts/vtt.py executes system utilities avconvert (on macOS) or ffmpeg (on Linux) to extract audio from video files before transcription. These are standard media processing tools.
[EXTERNAL_DOWNLOADS]: The skill uses pip to install faster-whisper, a well-known machine learning library for speech-to-text. It also downloads the 'tiny' Whisper model (~75MB) from official sources upon first use, which is documented and expected behavior for this type of application.
[DATA_EXPOSURE]: The script scripts/douyin_extract.py reads a user-provided environment variable DOUYIN_COOKIE to facilitate authenticated API requests if the SSR method fails. This follows best practices for secret management by not hardcoding credentials.

video-to-text