transcript-processing

To extract action items from meeting transcripts and YouTube videos, this Skill automatically detects input formats and uses an Operations mode to generate structured decisions and action items. It processes text, audio, and video sources with staged human-in-the-loop validation.

Yes, you need ffmpeg, whisper-cli, and yt-dlp installed locally to process audio transcripts. These dependencies handle automatic format detection, local speech-to-text conversion, and YouTube URL downloads before the transcript is transformed into structured output.

Yes, you can use yt-dlp to download YouTube URLs for transcript processing. The Skill accepts YouTube links directly, automatically downloads the source media using yt-dlp, and converts the audio to text via whisper-cli for downstream analysis.

The best way to turn spoken content into structured learning materials is using the Skill's Content mode. It processes transcripts from tutorials and podcasts to extract frameworks and insights, outputting structured summaries suitable for educational exploration.

Human-in-the-loop validation works during transcript processing by supporting staged workflows where you review and validate intermediate outputs. This ensures the accuracy of extracted decisions, action items, and frameworks before finalizing the structured meeting minutes or content summaries.

Limitations when processing long workshop or podcast transcripts depend on your local whisper-cli configuration and ffmpeg capabilities. The Skill processes audio and video locally, so hardware constraints may affect transcription speed and the overall structured output generation time.