Facebook Carousel logo

Facebook Carousel

Community
roslove44
facebook-carousel

Creates the cards of a Facebook carousel ad at Meta's dimensions, as a series of coherent visuals. Use when the user wants a "Facebook carousel", a "carousel ad" or "several cards that swipe" for a project. Produces N .tsx templates (one card = one PNG) in a dedicated subfolder, sharing a common theme, aligned with brand.md. Requires a project with its brand.md in place.

Overview

Publisherroslove44
Repositorybrand-artisan
Skill namefacebook-carousel
Stars
5
Forks
0
Bundled files
Instructions only
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by roslove44 on GitHub. Read the source before you install it.

Installation

Install the Facebook Carousel AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/roslove44/brand-artisan/tree/main/skills/facebook-carousel
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/roslove44/brand-artisan.git /tmp/brand-artisan
mkdir -p .claude/skills
cp -r /tmp/brand-artisan/skills/facebook-carousel .claude/skills/facebook-carousel
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Facebook Carousel in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Facebook Carousel on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Facebook Carousel is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

facebook-carousel: cards for a Facebook carousel

Goal: produce the PNG assets of the N cards of a Facebook carousel ad, at the chosen Meta ratio, coherent with each other and aligned with the guidelines. A carousel is not N independent images: it is a series that tells something. Coherence across cards is the heart of the deliverable.

This is a specialization of new-template. Same conventions: the Template contract (from brand-artisan), assets through brand(), fonts discovered in fonts/, export default ... satisfies Template.

Status of the dimensions. The specs come from Meta's Ads Manager (Ads Guide + the carousel help pages); there is no organic post spec. The ratios are official Meta; the "1080×1080" is a convention (Meta documents 1024×1024 as the recommended minimum at 1:1).

0. Prerequisites (blocking)

The chain project -> template folder -> guidelines must exist:

  • Resolve <project> (from the arguments, otherwise ask).
  • Check templates/<project>/ and brands/<project>/brand.md.
  • Missing guidelines or project -> STOP: ask for /new-project <project> first. No visual without guidelines (CLAUDE.md rule).

1. Frame it (ask, don't guess)

  • Name of the carousel: a kebab-case slug, default carousel. It is the subfolder templates/<project>/<name>/. Check that it does not exist.
  • Narrative mechanic (see §3): a story, one card per product, a top N, a tutorial, before/after, or seamless. It dictates the whole design.
  • Number of cards: 2 to 10, sweet spot 3-5. If a card adds nothing, drop it.
  • Ratio: default 1:1 (1080×1080). Otherwise 4:5 (1080×1350) or 9:16 (1080×1920). The same for every card (mandatory).
  • Content of each card: a short message + its role in the series (hook, development, CTA). Plus the final CTA.
  • Read brands/<project>/brand.md: palette, type, logo variants, don'ts. If the project already holds a .tsx, read it as a reference.
  • Read brands/<project>/project.md if it exists: set the tone and the claims of the cards (don't invent figures or promises). If absent -> ask for the tone and the message rather than guessing.

2. Facebook constraints (to respect)

PointValueStatus
Number of cards2 to 10 (sweet spot 3-5)Official Meta
Ratio (identical throughout)1:1, 4:5 or 9:16 (3 % tolerance)Official Meta
Size 1:11080x1080 (Meta recommended min 1024)Convention / Meta
Size 4:51080x1350Official Meta
Size 9:161080x1920Official Meta
FormatPNG or JPGOfficial Meta
Weight / image≤ 30 MB (aim for < 1 MB)Official Meta
Headline / card≤ 40 chars (≤ 25 to avoid truncation)Official Meta
Primary text≤ 125 chars (lives outside the image)Official Meta
  • The same ratio on every card: mixing 1:1 and 4:5 is not supported.
  • The primary text and the headline/link are set in Ads Manager, not in the image: don't write everything into the visual.
  • The old 20 % text rule was dropped (2021), but an overloaded card is still illegible on mobile: one idea per card.

3. The spirit: arrange it so it lands

Narrative mechanics (pick ONE)

  • Sequential story: hook -> problem -> solution -> proof -> CTA. Each card moves the story forward.
  • One card = one product: a catalog, a range, variants. Constant product position and background; each card legible on its own.
  • Top N / tips: card 1 = the title of the list, cards 2..N = one item each, the last = the CTA. Number the cards.
  • Step-by-step tutorial: one step per card, each visual self-supporting (understandable without reading the text).
  • Before / after: strong contrast between the first and the last card.
  • Seamless (panoramic): see the dedicated warning below.

Card 1 = the hook (decisive)

It is the only one visible before the first swipe. A single message, 3-5 words, strong contrast, one element that creates curiosity (a figure with no context, a visual that "spills over", a question). Don't explain everything here: that kills the swipe.

Coherence across cards (the heart of it)

  • Identical palette and type on every card (a single accent that varies if needed).
  • The logo in a fixed, discreet position (the same corner throughout), never as the centerpiece: enlarging it on every card is a common mistake.
  • Constant placements: if the headline is at the bottom, it is always at the bottom.
  • A visual through-line: an element or a color that links the cards; a shade that evolves from card to card signals the progression.
  • Each card legible on its own (a user can enter in the middle).

Rising CTA

Soft at the start ("Find out more"), direct next, conversion on the last card ("Request a quote"). Each card has its own link in Meta.

Mistakes to avoid

Card 1 too explanatory; cards siloed with no through-line; repeating the same message; text too dense; mixing ratios; logo too big.

4. Write the cards

Structure: one subfolder per carousel, one .tsx per card, plus a shared theme.ts.

templates/<project>/<name>/
  theme.ts        <- shared palette + layout (pure TS, NO JSX)
  card-1.tsx      <- one card = one PNG
  card-2.tsx
  card-3.tsx
  • theme.ts factors out the coherence: palette taken from the guidelines, ratio/SIZE, margins. Pure TS with no JSX: discovery only loads .tsx files, so theme.ts is ignored as a template. Do not create a frame component as a .tsx in this folder: it would be discovered as a fake card and break the build. The coherence travels through the constants; the little bit of frame JSX gets repeated in each card (they differ anyway).
  • Naming and order: the sort is natural, so card-2 comes before card-10. Number the cards plainly, no padding needed.
  • Colors as constants (from theme.ts), assets through brand("<project>/..."), fonts discovered in fonts/; a missing font -> the "missing font" appendix in new-template, never silently.

theme.ts (skeleton)

ts
import type { CSSProperties } from "react";

export const SIZE = { width: 1080, height: 1080 }; // 1:1 carousel

// <Project> guideline palette (from brands/<project>/brand.md).
export const INK = "#......";
export const ACCENT = "#......";
export const PAPER = "#......";

// Frame shared by every card: same background, same safe area.
export const frame: CSSProperties = {
	width: "100%",
	height: "100%",
	display: "flex",
	flexDirection: "column",
	justifyContent: "space-between",
	padding: 96, // safe area (~10%)
	backgroundColor: INK, // opaque background, identical everywhere
};

card-1.tsx (skeleton)

tsx
import type { ReactNode } from "react";
import type { Template } from "brand-artisan";
import { SIZE, frame, PAPER, ACCENT } from "./theme";

function render(): ReactNode {
	return (
		<div style={frame}>
			{/* logo in a fixed position (same corner on every card) */}
			{/* the hook: a single message, strong contrast */}
			{/* progress marker (e.g. 1/3) */}
		</div>
	);
}

export default { size: SIZE, render } satisfies Template;

(With no title, the PNG comes out under the file's name: out/<project>/<name>/card-1.png.)

Warning: seamless mode (panoramic)

Facebook inserts gutters (a border plus a shadow) between the cards, so a panoramic image is never perfectly continuous on screen. If the user wants seamless with this model (N files):

  • Design a mural of N×1080 × 1080 in your head, and offset the background by -1080*(i-1) px in card i (each card renders its own slice).
  • No critical element (text, a face, the logo) within about 100 px of a left or right edge: the gutter would cut it.
  • In Ads Manager, turn off automatic optimization ("show the best-performing cards first") and upload left to right, otherwise Meta reorders them and destroys the mural. Seamless is demanding: by default, prefer a per-card mechanic (product, story, top N) that tolerates the gutters.

5. Verify

  • npm run typecheck -> green.
  • npm run build -> writes out/<project>/<name>/card-*.png.
  • Check each PNG: exact dimensions for the chosen ratio, the same ratio throughout, opaque background.
  • Check the visual coherence across the series (palette, type, logo position, through-line) and that each card is legible on its own.
  • Preview: npm run dev then /<project>/<name> (lists the carousel's cards).

Success criterion: N cards (2-10) at the same Meta ratio, opaque, visually coherent, card 1 catchy, the last card carrying the CTA, and using only colors and type from brand.md.

Frequently asked questions

What does the Facebook Carousel AI skill do?

Creates the cards of a Facebook carousel ad at Meta's dimensions, as a series of coherent visuals. Use when the user wants a "Facebook carousel", a "carousel ad" or "several cards that swipe" for a project. Produces N .tsx templates (one card = one PNG) in a dedicated subfolder, sharing a common theme, aligned with brand.md. Requires a project with its brand.md in place.

Why use Facebook Carousel on TypingMind?

Because you install it once and use it with any model. Facebook Carousel is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Facebook Carousel in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/roslove44/brand-artisan/tree/main/skills/facebook-carousel. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Facebook Carousel?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Facebook Carousel?

As many as you like. As long as a model supports skills, you can use Facebook Carousel with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Facebook Carousel AI skill free?

Yes. It is published on GitHub by roslove44 under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇