youtube-to-blog

name	youtube-to-blog
description	Use when asked to create a blog post from a YouTube video URL. Triggers on phrases like "create a blog post from this video", "blog post from YouTube", or when given a YouTube URL with intent to document learnings.

YouTube to Blog Post

Overview

Extracts a YouTube video transcript, analyzes the content for key technical insights, then delegates to the blog-post-manager agent to create a properly formatted blog post.

Workflow

Step 1: Extract Video ID

Parse the video ID from the URL:

• https://www.youtube.com/watch?v=VIDEO_ID → VIDEO_ID
• https://youtu.be/VIDEO_ID → VIDEO_ID

Step 2: Get Video Title

python3 -c "
import urllib.request, json
url = 'https://www.youtube.com/oembed?url=https://www.youtube.com/watch?v=VIDEO_ID&format=json'
data = json.loads(urllib.request.urlopen(url).read())
print('Title:', data.get('title', 'Unknown'))
print('Author:', data.get('author_name', 'Unknown'))
"

Step 3: Get Transcript

Method 1 (primary) — youtube-transcript-api:

# Use a venv to avoid Homebrew Python conflicts
python3 -m venv /tmp/yt-env
/tmp/yt-env/bin/pip install youtube-transcript-api -q
/tmp/yt-env/bin/python3 -c "
from youtube_transcript_api import YouTubeTranscriptApi
api = YouTubeTranscriptApi()
segments = api.fetch('VIDEO_ID')
transcript = ' '.join(s.text for s in segments)
print(transcript)
" 2>&1

Method 2 (fallback) — yt-dlp:

# Install if missing: brew install yt-dlp
yt-dlp --write-auto-sub --skip-download --sub-format vtt \
  -o "/tmp/%(id)s" "https://www.youtube.com/watch?v=VIDEO_ID"
cat /tmp/VIDEO_ID.en.vtt | grep -v '^[0-9]' | grep -v '^\.' | grep -v '^WEBVTT' | grep -v '^$'

If no transcript is available (private video or no captions), inform the user and stop.

Step 4: Analyze Content

With the transcript in hand, identify:

• Main topic: What is the video fundamentally about?
• Key concepts: 3–5 major technical ideas or techniques
• Tools/technologies: Libraries, frameworks, or platforms discussed
• Code patterns: Any specific code approaches demonstrated
• Practical takeaways: What can a reader immediately apply?
• Suggested blog structure: Logical section order (intro → concepts → examples → conclusion)

Keep the analysis concise — it becomes the brief for the blog-post-manager agent.

Step 5: Delegate to blog-post-manager Agent

Invoke the blog-post-manager agent with this structured brief:

Create a technical blog post based on this YouTube video.

VIDEO: [TITLE] by [AUTHOR]
URL: https://www.youtube.com/watch?v=VIDEO_ID
DATE: [TODAY'S DATE]

KEY TOPICS:
- [topic 1]
- [topic 2]
- [topic 3]

SUGGESTED SLUG: YYYY-MM-DD-kebab-case-title

SUGGESTED CATEGORIES: ['tag1', 'tag2', 'tag3']

CONTENT BRIEF:
[2–3 paragraph summary of what the video covers, written as a blog author would introduce it]

TRANSCRIPT CONTEXT:
[Paste the full transcript or the most relevant 2000–3000 word excerpt]

STRUCTURE:
- Intro: Hook the reader with the problem/opportunity the video addresses
- [Section per key concept]
- Code examples if the video demonstrates code
- Practical takeaways
- Attribution: Credit the original video and creator at the end

ATTRIBUTION FOOTER (include verbatim):
> This post is based on learnings from "[TITLE]" by [AUTHOR]. Watch the original video [here](https://www.youtube.com/watch?v=VIDEO_ID).

Frontmatter Reference

The blog-post-manager will create the file, but confirm these fields are included:

Field	Format	Example
title	String	"Understanding MCP Servers"
slug	YYYY-MM-DD-kebab-title	2026-06-07-mcp-servers
description	140–160 chars	One-sentence summary
categories	Array of strings	['mcp', 'ai-agents']
published	ISO 8601	2026-06-07T00:00:00.000Z
author	String	"Dale Nguyen"
coverImage	Full URL	https://dalenguyen.me/assets/images/blog/slug.png
profileImage	Relative path	assets/images/dale-nguyen-avatar.webp

Common Issues

Problem	Fix
TranscriptsDisabled error	Video has no captions — try yt-dlp with --write-auto-sub
youtube-transcript-api not installed	Run pip install youtube-transcript-api
Transcript is too long (>5000 words)	Summarize by keeping intro, each section header + first paragraph, and conclusion
Private video / members-only	Cannot extract transcript; ask user to paste it manually
oEmbed returns 404	Video may be private or deleted; verify URL first