yt-research

Search YouTube for videos on any topic and return structured metadata (titles, views, authors, duration, URLs). Use when the user asks to research, find, or search for YouTube videos on a topic.

SKILL.md

GitHub

YouTube Research Skill

Search YouTube for videos on a topic and return structured metadata.

CRITICAL: If no topic is provided in $ARGUMENTS, you MUST ask the user what topic they want to research before proceeding. Do NOT guess or assume a topic.

Usage

  1. Activate the virtual environment and run the search script:
cd /home/user/Sneha && source .venv/bin/activate && python .claude/skills/yt-research/scripts/yt_search.py --query "<TOPIC>" --max-results <N> --sort <relevance|date>

Parameters

  • --query / -q: The search topic (REQUIRED - ask user if not provided)
  • --max-results / -n: Number of videos to return (default: 10)
  • --sort: Sort order
    • relevance (default): Most relevant results
    • date: Most recent uploads (use when user says "latest", "trending", "newest", or "recent")

Interpreting User Requests

  • "Find 25 latest videos on X" → --query "X" -n 25 --sort date
  • "Search for videos about X" → --query "X" (defaults are fine)
  • "25 trending videos on X" → --query "X" -n 25 --sort date

Output

The script outputs a JSON array to stdout. Parse it and present results as a formatted table with columns: #, Title, Author, Views, Duration, URL.

Example Output

[
  {
    "title": "Video Title",
    "url": "https://www.youtube.com/watch?v=...",
    "author": "Channel Name",
    "views": 123456,
    "duration": "12:34",
    "upload_date": "20250401"
  }
]

Format views with commas (e.g., 123,456) and upload_date as readable dates (e.g., "Apr 1, 2025").