GarlicAI 🧄

Generate AI-powered carousel content and images from PDFs, text, and custom prompts directly from your terminal.

GarlicAI is an interactive CLI tool that helps creators, marketers, founders, and educators generate social-media carousel slides using OpenAI models and GPT-Image.

Features

• 📄 Convert PDFs into carousel slide content
• ✍️ Generate carousel copy from text prompts
• 🖼️ Generate slide images with GPT-Image
• 📋 Copy generated prompts to clipboard
• 🎨 Configurable themes and settings
• ⚡ Fast interactive terminal experience
• 🤖 Powered by OpenAI

Installation

Global Installation

npm install -g garlicai

Run with npx

npx garlicai

Requirements

• Node.js 18+
• OpenAI API Key

Setup

Create an OpenAI API key and export it as an environment variable:

export OPENAI_API_KEY="your_api_key"

Or create a .env file:

OPENAI_API_KEY=your_api_key

Usage

Start the CLI:

garlic

You will be guided through an interactive menu where you can:

• Generate carousel content from PDFs
• Generate carousel content from text
• Generate image prompts
• Generate slide images
• Copy prompts
• Update settings

Example Workflow

1. Convert PDF → Carousel

PDF
 ↓
AI Analysis
 ↓
Carousel Slides
 ↓
slide-1.txt
slide-2.txt
slide-3.txt
...

2. Generate Images

Carousel Slides
 ↓
Image Prompts
 ↓
GPT-Image
 ↓
slide-1.png
slide-2.png
slide-3.png
...

Development

Install dependencies:

bun install

Run in development mode:

bun run dev

Build:

bun run build

Publishing

Verify package contents:

npm pack --dry-run

Publish:

npm publish

Configuration

GarlicAI stores configuration in:

setup.json
theme.json

You can edit these files directly or through the CLI settings menu.

Output

Generated files are stored inside the gen/ directory:

gen/
 ├── slide-1.txt
 ├── slide-2.txt
 ├── slide-3.txt
 ├── slide-1.png
 ├── slide-2.png
 └── slide-3.png

Tech Stack

• TypeScript
• Bun
• OpenAI SDK
• GPT-Image
• Inquirer
• Chalk
• Ora

License

MIT

Author

Built with ❤️ by the GarlicAI team.

This is pdf file of existing carousle post and i want to recreate this post analyse this pdf visials and text content and give me image geration prompts from this json arreay format like ["first slide prompt", "second slide prompt", ...] — no markdown fences, no preamble, no extra text. INSTRUCTIONS: Dont include exsitng page branding or any other branding including text, profile picture, mention id