agentleFS
TubeLabHQTubeLabHQ/youtube-research-skills/skills/script-breakdown/SKILL.md

script-breakdown

Skill91 starsChanged 11 days agoMITView on GitHub

Use when the user wants to break down a YouTube video's hook, structure or beat sheet, see how the top videos in a niche or channel are built, or outline a new script. Requires the TubeLab MCP.

2 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 7 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If it runs long, cut whole rows.

    line 49
  2. Very specific:

    Do not search without publishedAtFrom and classificationQuality at "neutral" or above, or the results fill with low-effort videos.

    line 55
  3. Distinctive:

    Do not pick videos by raw views.

    line 56
  4. Distinctive:

    Do not paraphrase the hook.

    line 57
  5. Distinctive:

    Do not name a beat by its topic.

    line 58
  6. Distinctive:

    Do not stop at the first payoff.

    line 59
  7. Distinctive:

    Do not call a layout a pattern from one video.

    line 60
  8. Distinctive:

    Do not trust numbers in captions.

    line 61

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “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

Script Breakdown

Overview

How a video holds attention, mapped from its transcript into a structure the user can reuse. Strong videos cycle through promise, stakes, re-hooks and small payoffs before the final payoff. Showing where each one lands is the point. The hook is the exact first 30 seconds, quoted. Transcripts are free to read.

Workflow

  1. Clarify scope only if needed. Read 1 to 3 videos. If the user names a channel or a niche instead, pick the videos yourself. For a channel, take its best videos of the last three months by ratio, which is a free read. For a niche, take the top outliers from one search. Say which you picked and why.
  2. Tag every beat in order with one type: hook, promise, stakes, open loop, proof, re-hook, mini-payoff, payoff, or call to action. Types repeat, and a long video has 10 to 20 beats.
  3. Turn the beats into a layout someone could reuse. With several videos, keep only the parts they share.

Output Format

# 🎬 Script Breakdown: {video, channel or niche}

- **Main finding:** {one sentence}
- **Videos picked:** {only if you chose them: which, and why in one sentence}

## [{title}]({video url}) · [{channel}]({url}) · {views, ratio, length}
- **Hook (0:00 to 0:30):** "{word for word}"
- **What it does:** {one sentence}

| Time | Beat | What happens |
|---|---|---|
| [{m:ss}]({link to that moment}) | {type} | {a few words or a short quote} |

## 🗺️ Layout
Hook 0:00 → Promise {m:ss} → ... → Call to action {m:ss}

## 📝 Script Skeleton
{Only if the user gave a topic. Build it on the layout.}
  • Length: about 700 words per video.
  • Language: plain words and short, complete sentences. Nothing compressed.
  • Bullets: one fact each, under 25 words, except quotes and verbatim titles. No extra sections or bullets.
  • If it runs long, cut whole rows. Never cut the hook, and never squeeze sentences.
  • Links: link every video, channel, moment, brand and offer you cite, built from IDs, handles and links in the data. Never guess one.
  • Quotes: the hook and short lines only. Never paste the full transcript.

Common Pitfalls

  • Do not search without publishedAtFrom and classificationQuality at "neutral" or above, or the results fill with low-effort videos.
  • Do not pick videos by raw views. A high ratio shows the structure worked for that channel, not just that the channel is big.
  • Do not paraphrase the hook. The exact words are the finding.
  • Do not name a beat by its topic. Name what it does to keep the viewer watching.
  • Do not stop at the first payoff. Late re-hooks are what hold a long video.
  • Do not call a layout a pattern from one video.
  • Do not trust numbers in captions. Check them against the description, and mark unclear words [unclear].

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.