Skip to content

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Meta Ads Generator

A Claude Code Skill by @tenfoldmarc

Drop a product URL. Get four finished, on-brand Meta ad creatives. The skill scrapes the brand's voice and palette, pulls real customer reviews and competitor ads, builds a psychology pillar map, drafts the concepts, and renders them through your favorite image model.

Backend-agnostic — works with Higgsfield MCP, kie.ai, OpenAI, or any OpenAI-compatible image API.


What It Does

  1. You paste a product URL (e.g. https://www.goodles.com)
  2. The skill extracts the brand's voice, palette, fonts, and product imagery from the site
  3. It scrapes 30-50 customer reviews and 5-10 competitor ads in the same category
  4. It builds a "psychology pillar map" — the 4-6 emotional drivers behind why people buy this thing
  5. It drafts 4 ad concepts (hook + headline + subhead + CTA + visual direction) and shows them to you for approval
  6. After you approve, it generates 4 finished ad images through your chosen image backend, using your saved reference ads as visual inspiration
  7. It opens the images in Preview so you can review and stage the keepers

Requirements

  • A Mac, Linux, or Windows computer
  • Claude Code installed and working
  • One of the following image generation options:
    • Higgsfield MCP installed in Claude Code (best quality, no API key needed)
    • A kie.ai API key — sign up at kie.ai
    • An OpenAI API key — get one at platform.openai.com/api-keys
    • Any OpenAI-compatible image API (Together, Fireworks, OpenRouter, self-hosted, etc.)

Don't worry about connecting these manually — the skill walks you through everything on first run.


Install

Step 1 — Open your terminal

Mac: Press Command + Space, type Terminal, hit Enter. Windows: Press Win + R, type cmd, hit Enter. (If you have Git Bash, use that instead.) Linux: Open your terminal app.

Step 2 — Run this command

Copy-paste this entire line and hit Enter:

git clone https://github.com/tenfoldmarc/meta-ads-generator-skill ~/.claude/skills/meta-ads-generator

You'll see some text scroll by. Wait until it finishes (takes a few seconds).

Step 3 — Open Claude Code

In the same terminal window, type:

claude

Hit Enter. Claude Code will open.

Step 4 — Run the skill

Type:

/meta-ads-generator

Hit Enter. On your first run, the setup wizard walks you through picking an image backend, pasting your API key (if needed), and pointing to a folder where you'll save reference ads you love.


The Reference Ads Folder (read this — it matters)

The skill works dramatically better when it can see ads you actually like.

During setup it creates a folder (default: ~/meta-ads-generator-references/). Drop 5-15 PNG or JPG screenshots of ads you love in there — yours, competitors', or anything you find inspiring. The skill passes them to your image model as visual references every time you run it.

Where to grab good reference ads:

  • Meta Ad Library — search competitors in your niche
  • Foreplay or Motion — paid ad swipe tools
  • Screenshots from your Instagram / Facebook feed
  • Your own past winning ads

You can add or remove references anytime — the skill always reads what's currently in the folder at run time.


Usage

Generate ads for an e-commerce brand:

/meta-ads-generator https://www.goodles.com

Generate ads for a SaaS product with a specific goal:

/meta-ads-generator

Then answer the prompts. Goal = lead, audience = solo founders, etc.

Switch image backend later:

re-run meta-ads-generator setup

Updating

To get the latest version:

cd ~/.claude/skills/meta-ads-generator && git pull

Built By

@tenfoldmarc — Follow for daily AI automation walkthroughs. Real systems, not theory.

About

One-prompt Meta ad generator. Product URL → 4 finished on-brand ad creatives. Backend-agnostic (Higgsfield MCP / kie.ai / OpenAI / any OpenAI-compatible API).

Resources

Stars

19 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors