image-gen

Image-Gen - Claude AI Image Generation Skill

Skill Overview

Image-Gen is a powerful Claude skill that lets you directly call multiple AI image generation APIs through conversation. It supports text-to-image generation, reference image editing, bulk generation, and more.

Use Cases

1. Generating Illustrations for Content Creation

Quickly generate cover images and accompanying visuals for blog posts and social media content. Supports multiple aspect ratios and sizes to fit the needs of different platforms.

2. Product Prototyping and Asset Creation

Use the reference image feature to generate variations or modify styles based on existing assets, enabling fast iteration on product design and marketing materials.

3. Bulk Image Production

Generate multiple images at once—ideal for scenarios that require a large amount of visual assets, such as batch product display images or test data images.

Core Features

Multi-Provider API Integration

A unified interface supports four major image generation services: OpenAI, Google Imagen, Alibaba Tongyi Wanxiang (DashScope), and Replicate. You can switch flexibly based on your needs without having to separately learn how to call each platform.

Intelligent Image Configuration

Supports custom aspect ratios (16:9, 9:16, 1:1, etc.) and quality presets (normal, 2k). Automatically handles platform-specific size differences so the generated images match your use case precisely.

Reference Image Editing

Supports uploading reference images for modification and re-creation. Compatible with Google multimodal models and OpenAI editing features, making it suitable for fine-tuning existing images or performing style transfer.

FAQs

Which image generation APIs are supported by image-gen?

Image-Gen integrates four major AI image generation services: OpenAI’s GPT Image model, Google’s Imagen and Gemini multimodal models, Alibaba Cloud’s Tongyi Wanxiang (DashScope), and Replicate’s open-source model ecosystem. You need to configure the corresponding API keys before use.

How do I generate a new image using a reference image?

Use the --ref parameter to specify the reference image path. For example: --ref source.png. The reference image feature supports Google multimodal models (e.g., gemini-3-pro-image-preview) and OpenAI image editing. The system will automatically select a compatible provider, or you can manually specify --provider google or --provider openai.

How do I configure the default image generation model?

Create an EXTEND.md configuration file to set default options. You can configure the default provider, default model, default quality, default aspect ratio, and more. Configuration priority is: CLI parameters > EXTEND.md > environment variables. During the first use, the system guides you through the setup process; afterward, generated images automatically use the preset model and parameters.