YouTube Video Analysis

Overview

Transforms YouTube videos into searchable, analyzable Obsidian notes with personalized suggestions. Extracts transcripts, generates summaries connecting insights to your work priorities, and creates clickable timelines for easy reference.

When to Use

•Analyzing conference talks or product videos for insights
•Extracting key learnings from technical tutorials
•Processing educational content into knowledge base
•Building searchable reference library from videos
•Need personalized suggestions connecting video to work priorities
•Want to save time by reading transcripts instead of watching

Quick Reference

Task	Command	Output
Basic analysis	`/youtube URL`	Full analysis with transcript
With questions	`/youtube URL What does this say about X?`	Analysis + relevance section
Long video (>60min)	Automatic	Summary mode (timeline only)

Output location: {{ folders.youtube }}/

Process Summary

•Extract video ID from URL
•Download transcript using yt-dlp
•Load {{ profile.interest_profile }} for context
•Generate analysis (summary, findings, suggestions, timeline)
•Save to Obsidian vault

Duration-based strategy:

•< 15min: Full transcript
•15-60min: Full transcript (chunked reading)
•

60min: Summary + timeline only

Common Mistakes

Mistake	Fix
Missing yt-dlp	Install with `brew install yt-dlp` before running
Video has no transcript	Check if captions exist; some videos don't have transcripts
Reading large transcript with Read tool	Always use head/sed/tail via Bash to avoid token limits
Missing {{ profile.interest_profile }}	Create profile in vault root first for personalized suggestions
Not handling long videos	Automatically uses summary mode for >60min videos

Dependencies

•yt-dlp – brew install yt-dlp
•python3 – For VTT parsing (pre-installed macOS)
•Obsidian MCP – For vault access
•{{ profile.interest_profile }} – In vault root

Detailed Documentation

For complete workflow, code examples, and troubleshooting, see youtube-workflow.md.