agentleFS
r0undm1dn1ghty-starr0undm1dn1ghty-star/ds_content-radar/.claude/skills/youtube-research/SKILL.md

youtube-research

Skill0 starsChanged 35 days agoMITView on GitHub

Research high-performing YouTube videos in a niche WITHOUT TubeLab: yt-dlp scraping + local z-score outlier math (free). Identifies outlier videos, analyzes top 3 relevant videos with AI, and generates reports with actionable hook formulas.

  • Reads credentials

16 other projects use a version of this, 1 of them edited.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

8 rules. 0 are common, 4 are distinctive, and 4 are very specific to this project.

  • 0

    Common

    Most of the 13 edited versions we read keep these.

  • 4

    Distinctive

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    No API key needed: yt-dlp scrapes channels, outlier z-scores computed locally

    line 26
  2. Very specific:

    Read .claude/context/youtube-channel.md to get the channel ID.

    line 40
  3. Distinctive:

    Run the outlier search with both keyword sets:

    line 62
  4. Very specific:

    Read outliers.json and the user's niche from .claude/context/youtube-channel.md.

    line 84
  5. Distinctive:

    Transcript relevance: If transcript exists, check it mentions relevant topics

    line 88
  6. Distinctive:

    Direct niche priority: Prefer videos from direct keyword search over adjacent

    line 89
  7. Very specific:

    Write the filtered videos to {RUN_FOLDER}/filtered-outliers.json:

    line 93
  8. Very specific:

    Read {RUN_FOLDER}/outliers.json and {RUN_FOLDER}/video-analysis.json, then generate {RUN_FOLDER}/report.md.

    line 121

“Common” is measured: kept by at least 6 in 10 of the 13 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

YouTube Research

Research high-performing YouTube outlier videos, analyze top content with AI, and generate actionable reports.

Prerequisites

  • No API key needed: yt-dlp scrapes channels, outlier z-scores computed locally
  • GEMINI_API_KEY environment variable (for video analysis)
  • google-genai, yt-dlp, youtube-transcript-api Python packages

Workflow

Step 1: Create Run Folder

mkdir -p youtube-research/$(date +%Y-%m-%d_%H%M%S)

Step 2: Get Channel ID

Read .claude/context/youtube-channel.md to get the channel ID.

Step 3: Fetch Channel Videos

python scripts/get_channel_videos.py CHANNEL_ID --format summary

This returns JSON with the channel's video titles and view counts.

Step 4: Analyze Channel

Analyze the channel data to extract:

  • keywords: 4 search terms for the channel's direct niche
  • adjacent-keywords: 4 search terms for topics the same audience watches
  • audience: 2-3 profiles with objections, transformations, stakes
  • formulas: Reusable title templates

See references/channel-analysis-schema.md for the full schema and example output.

Step 5: Search for Outliers

Run the outlier search with both keyword sets:

python .claude/skills/youtube-research/scripts/find_outliers.py \
  --keywords "keyword1" "keyword2" "keyword3" "keyword4" \
  --adjacent-keywords "adjacent1" "adjacent2" "adjacent3" "adjacent4" \
  --output-dir youtube-research/{run-folder} \
  --top 5

This runs two searches:

  • Direct niche: keywords with 5K+ views threshold
  • Adjacent audience: adjacent-keywords with 10K+ views threshold

Output files:

  • outliers.json - All outliers normalized for video analysis
  • report.md - Basic markdown report
  • thumbnails/*.jpg - Video thumbnails
  • transcripts/*.txt - Video transcripts

Step 6: Filter Relevant Videos for Analysis

Read outliers.json and the user's niche from .claude/context/youtube-channel.md.

CRITICAL: Select MAX 3 videos that are most relevant to the user's niche. Filter by:

  1. Title relevance: Title contains keywords related to user's niche/topics
  2. Transcript relevance: If transcript exists, check it mentions relevant topics
  3. Direct niche priority: Prefer videos from direct keyword search over adjacent

Skip videos that are clearly outside the user's content style (e.g., entertainment/vlogs when user does tutorials).

Write the filtered videos to {RUN_FOLDER}/filtered-outliers.json:

{
  "outliers": [/* max 3 relevant videos */],
  "filter_reason": "Selected based on relevance to [user's niche]"
}

Step 7: Analyze Top Videos with AI

python3 .claude/skills/video-content-analyzer/scripts/analyze_videos.py \
  --input {RUN_FOLDER}/filtered-outliers.json \
  --output {RUN_FOLDER}/video-analysis.json \
  --platform youtube \
  --max-videos 3

Extracts from each video:

  • Hook technique and replicable formula
  • Content structure and sections
  • Retention techniques
  • CTA strategy

See the video-content-analyzer skill for full output schema and hook/format types.

Step 8: Generate Final Report

Read {RUN_FOLDER}/outliers.json and {RUN_FOLDER}/video-analysis.json, then generate {RUN_FOLDER}/report.md.

Report Structure:

# YouTube Research Report

Generated: {date}

## Top Performing Hooks

Ranked by engagement. Use these formulas for your content.

### Hook 1: {technique} - {channelTitle}
- **Video**: "{title}"
- **Opening**: "{opening_line}"
- **Why it works**: {attention_grab}
- **Replicable Formula**: {replicable_formula}
- **Views**: {viewCount} | **zScore**: {zScore}
- [Watch Video]({url})

[Repeat for each analyzed video]

## Content Structure Patterns

| Video | Format | Pacing | Key Retention Techniques |
|-------|--------|--------|--------------------------|
| {title} | {format} | {pacing} | {techniques} |

## CTA Strategies

| Video | CTA Type | CTA Text | Placement |
|-------|----------|----------|-----------|
| {title} | {type} | "{cta_text}" | {placement} |

## All Outliers

### Direct Niche
| Rank | Channel | Title | Views | zScore |
|------|---------|-------|-------|--------|
[List direct niche outliers]

### Adjacent Audience
| Rank | Channel | Title | Views | zScore |
|------|---------|-------|-------|--------|
[List adjacent outliers]

## Actionable Takeaways

[Synthesize patterns into 4-6 specific recommendations based on video analysis]

Focus on actionable insights. The "Top Performing Hooks" section with replicable formulas should be prominent.

Quick Reference

Full pipeline:

RUN_FOLDER="youtube-research/$(date +%Y-%m-%d_%H%M%S)" && mkdir -p "$RUN_FOLDER" && \
python .claude/skills/youtube-research/scripts/find_outliers.py \
  --keywords "k1" "k2" "k3" "k4" \
  --adjacent-keywords "a1" "a2" "a3" "a4" \
  --output-dir "$RUN_FOLDER" --top 5

Then filter outliers for niche relevance (max 3), run video analysis, and generate the report.

Script Reference

get_channel_videos.py

python .claude/skills/youtube-research/scripts/get_channel_videos.py CHANNEL_ID [--format json|summary]
ArgDescription
CHANNEL_IDYouTube channel ID (24 chars)
--formatjson (full data) or summary (for analysis)

find_outliers.py

python .claude/skills/youtube-research/scripts/find_outliers.py --keywords K1 K2 K3 K4 --adjacent-keywords A1 A2 A3 A4 --output-dir DIR [options]
ArgDescription
--keywordsDirect niche keywords (4 recommended)
--adjacent-keywordsAdjacent topic keywords (4 recommended)
--output-dirOutput directory (required)
--topVideos per category (default: 5)
--daysDays back to search (default: 30)
--jsonAlso save raw JSON data

Output: outliers.json, report.md, thumbnails/, transcripts/

Scoring Algorithm

Videos ranked by: zScore × recency_boost

  • zScore: How much video outperforms its channel average
  • recency_boost: 1.0 for today, decays 5%/day (min 0.3×)

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.