SKILL/TREE

YouTube long form script outliner for 8-12 minute videos

Paste your topic and target audience — get a first-15-second hook, a 5-beat retention-paced outline with timestamps, shootable B-roll prompts, and a 3-part CTA strategy.

$3.99
Works in Claude (Pro/Max/Team) or ChatGPT (Plus/Team) · paid AI plan required
Both Claude + ChatGPT packages included — you can't pick the wrong one.
Updated
Included
01Writes the first 15 seconds word-for-word — the window where viewers decide to stay or swipe
025 beats with timestamps tuned to the 30% and 60% drop zones where retention graphs cliff-dive
03One pattern interrupt per beat — B-roll cut, text reveal, rhetorical question, camera angle change
04Shootable B-roll prompts ("close-up of hands typing on a laptop, no face" — not "show working")
053-part CTA strategy: mid-roll soft ask, end-screen primary, plus a binary comment prompt to pin
06Outputs a dark-mode HTML artifact with timestamp cards, B-roll callouts, and pattern interrupt pills — print-friendly

What it does

Most creators know their topic cold but wing the structure — and YouTube's analytics punish it. Audience retention graphs that cliff-dive at the 3-minute mark aren't a content problem, they're a pacing problem. This YouTube long form script outliner solves it. Give it a topic and a one-sentence description of your audience ("intermediate Python devs who hate filler" beats "developers"). It returns a complete filming blueprint: a first-15-second hook in one of four archetypes (counterintuitive claim, open loop, urgent problem, stakes hook), five beats with timestamp ranges distributed across 8-12 minutes, one pattern interrupt per beat targeting the 30% and 60% drop zones, specific shootable B-roll prompts you can hand to an editor, and a three-part CTA strategy — mid-roll soft ask, end-screen primary, optional pinned comment prompt. The whole thing takes under two minutes and outputs a dark-mode HTML artifact you can film against. It runs on `retention-patterns.md` — a bundled reference encoding loop mechanics (opening/closing rules, why never to leave a loop open past the 90% mark), pattern interrupt timing, the four hook archetypes by content type, and B-roll writing conventions (shootable, specific, relevant). That's the framework creator coaches package as $200 templates. You're not getting a fill-in-the-blank prompt — you're getting structured output informed by what actually keeps people watching.

Frequently asked

How do I keep YouTube viewers watching past the 3 minute mark?
Pace your video around the documented drop zones at the 30% and 60% timestamps. This skill builds the outline around them — Beat 2 lands a substantive payoff before the 30% drop, Beat 3 stacks pattern interrupts through the 60% zone, and at least one secondary loop stays open from Beat 1 through Beat 4. The bundled `retention-patterns.md` explains the mechanics.
Does this write the full script or just the outline I still have to write?
Structured outline — the first 15 seconds is scripted word-for-word (opening line + visual direction + promise statement), each of the 5 beats gets 3-5 concrete talking points (not vague topics), B-roll prompts per beat, and the end-screen CTA line is written verbatim. The middle talking points are bullets you film against. Ask it to expand any beat into full sentences and it will.
Can I use this for tech tutorials, finance, cooking, or only general vlogs?
Any niche. Tell it the topic and audience plainly ("async/await for intermediate Python devs, direct technical style, no fluff") and the talking points use real domain vocabulary — event loop, coroutines, asyncio.gather — not generic descriptions. Pattern interrupts adjust too: code-on-screen reveals for technical content, plate close-ups for cooking.
Why not just ask Claude or ChatGPT to outline a YouTube video for free?
Generic prompts produce generic outlines — usually a 5-bullet list with no timestamps, no pattern interrupts, no hook mechanics, and a "like and subscribe" CTA. This skill bakes in the 4 hook archetypes, the 30%/60% drop zones, the loop-and-close discipline, and the goal-matched CTA framework as a bundled reference file the skill consults before writing.
Does it work for Shorts or 20+ minute long-form?
Sweet spot is 8-12 minute long-form where retention pacing matters most. For a 20-minute deep dive the structure still applies — you expand each beat. For a 60-second Short the framework is overkill; a different tool is the right call.
Is this in the subscription or a separate buy?
Standard-tier skill at $3.99. If your subscription covers standard skills, you have access. Otherwise it's a one-time buy that runs in Claude or ChatGPT with no install.

Install — no Terminal required

After checkout you land on a page with a one-click download and a pre-built install prompt. Pick the AI you already pay for — both packages are included, so you can't pick wrong.

What you'll see the second your payment clears
Your license key
ST-XXXX-XXXX-XXXX · permanent
Pre-built install prompt⧉ Copy
ChatGPT package (ZIP)⬇ Download

No setup work before you buy — copy, paste, done.

Have Claude Pro, Max, or Team?
Copy the prompt, paste into Claude

Copy the pre-built install prompt from the success page, paste it into any Claude Code chat, and Claude installs the skill itself — about 10 seconds. Using the app instead? Drop the downloaded folder into a Claude Project's knowledge and ask Claude to use it.

Have ChatGPT Plus or Team?
Build a Custom GPT — ~2 minutes
  1. Click Download ChatGPT package (included with every purchase).
  2. Unzip it — inside is a SETUP.md and a knowledge/ folder.
  3. Go to chatgpt.com → your profile → My GPTs Create a GPTConfigure.
  4. Paste the Name, Description, and Instructions from SETUP.md into the matching fields.
  5. Under Knowledge, upload every file in the knowledge/ folder.
  6. Click Create/Save, open your GPT, and describe your task in plain English.

A paid plan on either platform is required — Claude (Pro, Max, or Team) or ChatGPT (Plus or Team). Comfortable with Terminal? A one-line npx skilltree-network install path is available too.