transcript-ingest

Extract transcripts from video URLs and route atomic ideas into domain wiki pages.

Updated Feb 27, 2026
One-click install
npx skills add https://github.com/Dawaad/skills --skill transcript-ingest-dawaad
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: transcript-ingest
Source: https://github.com/Dawaad/skills/tree/main/skills/transcript-ingest
Command: npx skills add https://github.com/Dawaad/skills --skill transcript-ingest-dawaad

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires yt-dlp, openai-whisper, and includes scripts (resource) and references (resource) components.

What problem does it solve? Long-form videos and podcasts contain valuable ideas, but manually transcribing them and organizing the takeaways into a personal knowledge base is slow and error-prone. This Skill automates transcript extraction and turns one video URL into structured, cross-linked wiki pages. ## Core Features & Use Cases - Transcript Extraction: Pulls captions via yt-dlp from YouTube, Vimeo, podcasts, and other supported sources, with automatic Whisper fallback when captions are unavailable. - Atomic Topic Extraction: Reads the full transcript, filters out platitudes and doc-derivable content, and classifies surviving ideas by kind (decision, flow, insight, framework, sop, and more). - Multi-Wiki Routing: Routes each idea to the correct domain wiki (Dev, Personal, Meta, Marketing, and others), writes source pointers with timestamped evidence, and updates indexes and logs. - Use Case: Paste a two-hour conference talk URL and receive a confirmed proposal table of candidate pages, then approved pages written across the relevant wikis with timecoded transcript receipts. ## Quick Start Ingest this YouTube talk into my wiki and extract its key ideas as atomic pages: <video URL>.

Frequently Asked Questions about transcript-ingest

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I extract a transcript from a YouTube video?

Run the extract_transcript.py script with the video URL and a save directory. It first tries yt-dlp captions, then falls back to Whisper transcription, producing a timestamped transcript.md plus metadata and chapter files in a dated bundle.

What video sources does yt-dlp transcript extraction support?

Any URL yt-dlp can resolve works, including YouTube, Vimeo, podcast feeds, Twitter/X videos, lecture recordings, and direct MP3 or MP4 links. Instagram Reels and short-form content are explicitly out of scope for this pipeline.

What happens when a video has no captions available?

The script automatically falls back to Whisper transcription using the base model after downloading the audio track. This is slower, roughly real-time on CPU, but produces a usable timestamped transcript without manual intervention.

Why does yt-dlp metadata extraction fail on some URLs?

Failures usually mean the video is geo-blocked, private, or the installed yt-dlp version is outdated. The script surfaces the error rather than fabricating results, so upgrading yt-dlp or checking access is the fix.

When should I not use transcript-based wiki ingestion?

Skip it for Instagram Reels or short-form benchmark capture, which belongs to a different flow, and for arbitrary file or article ingestion handled by a generic ingest command. It is designed specifically for long-form video and podcast sources.