agentleFS
retrogtxretrogtx/ai-marketing-videos/skills/ad-copywriting/SKILL.md

ad-copywriting

Skill4 starsChanged 54 days agoMITView on GitHub

Write short-form video ad scripts, hooks, storylines, captions and CTAs for any app or product, and keep them inside platform ad policy (Meta, TikTok, YouTube). Includes safe-zone and caption specs for 1080x1920. Use when drafting or reviewing ad copy, VO scripts, on-screen text or end cards.

3 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

6 rules. About 0 are common, 6 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Put a price on creative unless the landing page matches exactly and the user wants it.

    line 47
  2. Distinctive:

    Use competitor brand marks, or public figures without a clear right to.

    line 50
  3. Distinctive:

    Don't imply past results predict future ones, or that a regulator endorses you.

    line 56
  4. Distinctive:

    Top: keep the top 14% clear (y < 269), for the name and "Sponsored" label.

    line 61
  5. Distinctive:

    Bottom: keep the bottom 35% clear (y > 1248), for the CTA button and caption.

    line 62
  6. Distinctive:

    Sides: keep 6% clear on each side (x < 65, x > 1015).

    line 63

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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

Ad copywriting

Storylines

Draft 4–6 storylines before writing any script. For each one, give:

  • a title,
  • a one-line angle,
  • the psychology it uses (curiosity gap, insider knowledge, pattern interrupt, identity or aspiration, before/after, goal gradient),
  • the proof it rests on.

Let the user choose. Expect some to be replaced; don't argue for rejected ones.

Angles that tend to work:

  • Borrowed attention: build around well-known, trendy companies or events the audience already cares about. Use real names and facts as story facts, sourced from your own content.
  • The money or cause trail: "You can't buy X. But you can own/use/learn what powers it."
  • The overlooked signal: "Everyone watches X. Almost nobody watched Y."
  • Identity outcome: "Six months from now, you could be the one your friends ask."

Script shape (30–40s, about 170–190 wpm)

  1. Hook (0–2s): topic plus outcome, instantly clear. A pattern interrupt or a concrete fact.
  2. Tension: what most people miss, or the pain.
  3. Insight or payoff: the story's reveal.
  4. What you get: "[Product] is a [category] that [does X], with [concrete thing 1] and [concrete thing 2]." Say it plainly and show it on the real UI.
  5. Secondary value, e.g. "Plus a course so you can do it yourself."
  6. CTA line: one verb plus the brand plus the outcome. "Join X. Become the one who knows."

Store scripts as JSON, one object per line: {id, text, pause}. Keep a markdown doc next to it with the approved scripts, a source per factual claim, and a compliance note.

Hooks

  • The first 2 seconds must say what this is about and why the viewer should care.
  • Shapes that work: specific receipt, myth-bust, stop-doing, curiosity gap, POV.
  • The first frame must be visually distinct. Use at least 2 cuts in the first 3 seconds, and never open on a logo.
  • Curiosity hooks suit cold traffic; problem/solution suits retargeting.
  • Measure hook rate as 3-second views ÷ impressions. Under 25% means the hook is failing.

Never

  • Deny a real part of the offer to sound humble. If the product gives recommendations, say so.
  • Make performance, income, health or results claims you can't substantiate, or guarantees. Performance claims are also a platform-policy problem.
  • Put a price on creative unless the landing page matches exactly and the user wants it.
  • Pitch free tiers or trials unless the user asks.
  • Show jargon labels, internal feature names, or on-screen sentences that duplicate the VO.
  • Use competitor brand marks, or public figures without a clear right to.

Regulated categories (finance, health, credit, housing, employment)

  • Frame the product as research or education, not advice. Put a disclaimer on the end card, e.g. "Research and education. Not investment advice."
  • Platforms review the landing page too, so the ad and the page must agree.
  • Don't imply past results predict future ones, or that a regulator endorses you.
  • Declare a special ad category when it applies.

Captions and safe zone (1080x1920, Meta Reels/Stories 2026)

  • Top: keep the top 14% clear (y < 269), for the name and "Sponsored" label.
  • Bottom: keep the bottom 35% clear (y > 1248), for the CTA button and caption. A disclaimer pushes this to 40%.
  • Sides: keep 6% clear on each side (x < 65, x > 1015). The right rail icons go deeper, so keep x > 960 empty below mid-screen.
  • Key visuals go inside x 65–1015, y 250–1150. The caption band is about y 1170–1310, and nothing important sits under it.
  • Caption style: 2–4 words per chunk, bold sans, white with a stroke or dark plate, 56–80px. Keep one fixed anchor for the whole ad and make sure the caption text never duplicates on-screen titles.
  • Whisper spelling fixes: fix only real mis-hearings of brand names. Over-eager rules break normal words (e.g. "everyone" became "every one").

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.