How vigozhao/ai-visual-prompt-cookbook works
AI Visual Prompt Cookbook is a curated library of reusable JSON image-style systems, with a static gallery for discovering and copying them. The main user journey is browse or filter styles, inspect a style and its variables/examples, copy its JSON into an image-capable LLM, supply values, and generate an image prompt/result. The repository also builds searchable gallery data and crawlable style pages. Evidence is limited to the README and sampled excerpts; some browser-to-data wiring and unsampled page-generation details are not shown.
Main entry points: build-pages-mvp.py Gallery Builder · styles Style JSON Library · app.js Gallery UI · styles-data.js Gallery Style Data
Components 13
Style Catalog
- Style JSON Library
styles - Style Previews
styles - README Catalog
README.md
Gallery Experience
- Gallery UI
site/app.js - Gallery Style Data
site/styles-data.js - Style Detail Pages
site/styles
Gallery Generation
- Gallery Builder
scripts/build-pages-mvp.py - Search Vocabulary
scripts/search_terms.py - SEO Page Generator
scripts/seo_pages.py
Prompt Workflow
- User Values
Other components
- Visitor
- Image-capable LLM
- Generated Image
Connections 14
- Visitor → README Catalog: browses
README.md - Visitor → Gallery UI: browses
README.md - Gallery Builder → Search Vocabulary: uses vocabulary
scripts/build-pages-mvp.py - Gallery Builder → SEO Page Generator: calls generator
scripts/build-pages-mvp.py - Visitor → Style JSON Library: copies JSON
README.md - Visitor → User Values: provides values
README.md - Style JSON Library → Image-capable LLM: supplies style
README.md - User Values → Image-capable LLM: fills variables
README.md - Image-capable LLM → Generated Image: generates
README.md - Gallery UI → Gallery Style Data: reads data
- Gallery UI → Style Detail Pages: opens details
- Gallery Builder → Gallery Style Data: produces data
- Gallery Builder → Style Detail Pages: generates pages
- Style JSON Library → Style Previews: paired with previews