wc-seed-youtube

You can discover and transcribe YouTube videos by providing seed queries that filter for relevance and credibility, then saving the resulting transcripts as structured markdown files under a dedicated seed/youtube directory for knowledge-base creation.

YouTube video transcriptions are automatically deduplicated by enforcing URL checks during the discovery process, ensuring your knowledge base contains only unique video transcripts without redundant entries across multiple seed queries.

Filtering YouTube video content by relevance, credibility, and appropriate length is handled by aligning discovery queries with specific sub-questions and pain points, producing expert-curated video summaries tailored to your target audience segment.

Yes, parallel Task calls with max_turns set to 50 are used to process multiple seed queries concurrently, enabling rapid YouTube video discovery and transcription while writing structured markdown files with frontmatter metadata.

Transcribed YouTube videos are saved as markdown files containing frontmatter metadata including url, title, channel, source_type, date_scraped, pain_point, and sub_question to support structured content curation and knowledge-base organization.

No external dependencies are required to run YouTube video discovery and transcription, as the skill operates independently using parallel Task calls to find, filter, and transcribe content directly into structured markdown seed files.