What is claude-image-generation?

This repository is a collection of Claude Agent Skills that teach different ways to generate images, then combine them into a working app. It is the companion code for a series of written tutorials, and it is built around one idea: instead of just calling an image-generation API, Claude does the planning and prompt writing itself, then runs the right tool and checks the result before handing you a finished file. It teaches image generation at three different levels, each trading off cost, realism, and control differently. Level 1 is a code-based design engine that uses Python libraries to draw gradients, shapes, and real typography without any image model at all, making it free and great for posters, quote cards, and geometric designs with crisp text. Level 2 renders an actual three-dimensional scene using Three.js and captures it as a single image, again with no image model involved, useful for product shots and styled backgrounds. Level 3 is the more familiar approach: it sends a prompt to a hosted diffusion model through Cloudflare Workers AI, which is better suited to photographic or freeform images like logos and illustrations. Beyond the three levels, the repository also includes an AI Storybook pipeline that chains several skills together into a small application. You give it a plain-English story, and it splits the story into scenes, generates one consistent illustration per scene using an external image model, adds narration audio for each scene using text-to-speech, and packages everything into a single self-contained HTML file that works offline and can be shared as-is, complete with swipe navigation and narrated illustrations. The project is organized as a set of skill folders that plug into Claude Code, along with example images and a full example storybook. Level 1 and 2 skills need no external API and run for free and locally, while Level 3 and the storybook pipeline require API keys for services like Cloudflare Workers AI and other providers used for illustration and narration.