image-generation: AI Image Generation for Claude Code Developers

AI image generation has become an essential tool for developers building creative applications, especially when working with video creation, educational content, or storytelling projects. The image-generation skill pack brings two powerful, production-ready image generation capabilities directly into your Claude Code workflow — all completely free and open source under the MIT license.

Two Complementary Image Generation Capabilities

The image-generation pack isn't just one tool — it's a carefully curated combination of two specialized services that work together seamlessly:

Gemini NanoBanana Image Generation

This skill leverages Google's Gemini NanoBanana model to generate high-quality AI images from text prompts. Unlike generic image generators, NanoBanana is optimized for specific use cases like educational content, book illustrations, and poetry visuals. It automatically applies style prefixes (Ghibli, Educational, etc.) and generates images in the perfect 9:16 portrait ratio for Shorts and vertical video formats.

The implementation is clean and developer-friendly, with a TypeScript service class that handles all the complexity:

class ImageGenerationService {
  async generateImages(
    options: {
      prompt: string;
      numberOfImages: number;
      aspectRatio: string; // "9:16" for portrait
    },
    videoId: string,
    sceneIndex: number
  ): Promise<{ success: boolean; images: Array<{ data: Buffer }> }>;
}

Key features include automatic style prefixing, character reference mode support, and efficient API usage patterns that minimize costs while maximizing quality.

Seedance Image-to-Video Conversion

Once you have your static images, the next step is bringing them to life. The Seedance integration converts your AI-generated images into smooth, professional video clips using the Seedance 1.5 Fast model hosted on Atlas Cloud. At just $0.11 per clip, this is one of the most cost-effective ways to add motion to your content.

The Seedance service is designed for reliability and quality, with critical safeguards built in:

class AtlasCloudClient {
  async generateVideo(options: {
    prompt: string;        // 영상 모션 설명
    image: string;         // base64 또는 URL
    duration: number;      // 초 단위 (5~10)
    orientation: string;   // "portrait" (9:16)
    cameraFixed: boolean;  // true 권장 (할루시네이션 방지)
  }): Promise<{ url: string }>;
}

Why This Matters for Claude Code Developers

For developers using Claude Code to build automation workflows, having reliable, programmable image generation capabilities is transformative. Instead of manually creating assets or relying on external services with complex APIs, you can now generate exactly what you need, when you need it, with simple function calls.

The image-generation pack is designed with real-world development constraints in mind:

• Production-ready architecture: Clean TypeScript interfaces with proper error handling and type safety
• Cost optimization: Built-in safeguards to minimize API usage while maintaining quality
• Easy integration: Simple copy-and-paste installation with clear dependency management
• Vertical optimization: Specifically tuned for 9:16 portrait format used in Shorts and vertical video
• Open source transparency: Full MIT license means you can inspect, modify, and extend the code

Getting Started in 3 Steps

Adding these capabilities to your Claude Code environment takes less than a minute:

1. Add the marketplace: /plugin marketplace add https://github.com/clickaround/marketplace
2. Install the pack: /plugin install image-generation@clickaround
3. Start generating: Use the ImageGenerationService and AtlasCloudClient classes in your projects

The complete source code is available for inspection and modification, making it easy to adapt to your specific needs.