What Are the 12 Content Categories in the GPT-Image2 Prompt Library?

The GPT-Image2 prompt library organizes its templates into 12 primary content categories—including UI & Interfaces, Photography & Realism, and History & Classical Themes—defined in data/style-library.json in the freestylefly/awesome-gpt-image-2 repository.

The GPT-Image2 content categories provide a taxonomy for classifying prompt templates by thematic use case. These classifications enable both manual browsing and programmatic filtering of the 250+ templates available in the style library.

Overview of the Category System

In [data/style-library.json](https://github.com/freestylefly/awesome-gpt-image-2/blob/main/data/style-library.json) at lines 32–88, the "categories" array defines the core taxonomy. Each category object contains:

• id: Unique identifier with cat- prefix (e.g., cat-ui)
• title: Bilingual labels in English and Chinese
• description: Scope definition for the category
• anchor: Markdown anchor for documentation navigation
• templateAnchor: Direct link to the template gallery section

These definitions power the navigation tiles rendered in README.md (lines 76–90) and enable the agent skill at agents/skills/gpt-image-2-style-library to process category-based queries.

The 12 Primary GPT-Image2 Content Categories

The library groups prompts into the following 12 core thematic buckets:

Category ID	English Name	Chinese Name	Description
cat-ui	UI & Interfaces	UI 与界面	Apps, websites, dashboards, social screenshots, and product interfaces.
cat-infographic	Charts & Infographics	图表与信息可视化	Infographics, knowledge maps, technical explainers, and structured diagrams.
cat-poster	Posters & Typography	海报与排版	Event posters, covers, type-driven visuals, and strong layout compositions.
cat-product	Products & E-commerce	商品与电商	Product shots, detail pages, packaging, selling points, and ads.
cat-brand	Brand & Logos	品牌与标志	Logos, identity systems, brand touchpoints, and campaign visuals.
cat-architecture	Architecture & Spaces	建筑与空间	Architecture renders, interiors, city maps, and spatial concepts.
cat-photo	Photography & Realism	摄影与写实	Portraits, phone photography, film texture, and commercial photography.
cat-illustration	Illustration & Art	插画与艺术	Illustration, art styles, material experiments, and decorative images.
cat-character	Characters & People	人物与角色	Character design, pose references, cards, and 3D toys.
cat-scene	Scenes & Storytelling	场景与叙事	Storyboards, narrative scenes, livestream frames, and world-building.
cat-history	History & Classical Themes	历史与古风题材	Classical scrolls, historical figures, traditional themes, and poetry visuals.
cat-document	Documents & Publishing	文档与出版物	White papers, manuals, encyclopedic plates, and publishing layouts.

The "Other" Classification

Beyond the 12 primary categories, the library includes a cat-other entry for Other Use Cases (其他应用场景). This catch-all category houses creative experiments, special tasks, and mixed workflows that do not fit the standard taxonomy.

Accessing Categories Programmatically

You can query these categories using the gpt-image-2-style-library package. The following snippets demonstrate how to load the library and filter by content category.

List All Categories

import { loadStyleLibrary } from 'gpt-image-2-style-library';

(async () => {
  const lib = await loadStyleLibrary();               // loads data/style-library.json
  const categories = lib.categories.map(c => ({
    id: c.id,
    titleEn: c.title.en,
    titleZh: c.title.zh,
    descriptionEn: c.description.en,
  }));
  console.log(categories);
})();

Filter Templates by Category

To retrieve only templates for Charts & Infographics:

import { loadStyleLibrary } from 'gpt-image-2-style-library';

(async () => {
  const lib = await loadStyleLibrary();
  const chartTemplates = lib.templates.filter(t => t.category === 'Charts & Infographics');
  console.log(chartTemplates.map(t => t.title.en));
})();

Agent Skill Integration

When working with Claude Code or Codex agents, reference categories directly via the skill:

Use gpt-image-2-style-library to list all "Architecture & Spaces" templates.

The skill queries the categories array and returns matching template IDs and titles for generation requests.

Key Source Files

• data/style-library.json (lines 32–88): Central definition of the 12 content categories and their metadata schema.
• README.md (lines 76–90): Visualizes categories as navigation tiles linking to template sections.
• docs/templates.md: Documents individual templates and references categories via the "category" field.
• agents/skills/gpt-image-2-style-library/SKILL.md: Exposes the category taxonomy for LLM-driven filtering.

Summary

• The library defines 12 primary GPT-Image2 content categories spanning UI design, photography, illustration, and historical themes.
• Category metadata resides in data/style-library.json with bilingual support and navigation anchors.
• The cat- prefix convention (e.g., cat-photo, cat-architecture) ensures consistent referencing across templates and documentation.
• Programmatic access is available via the gpt-image-2-style-library package using standard array filtering methods.

Frequently Asked Questions

How do I add a custom content category to the library?

To add a new category, append an object to the "categories" array in data/style-library.json containing id, title (en and zh), description, anchor, and templateAnchor fields. Use the cat- prefix for the ID to maintain consistency with existing categories like cat-ui and cat-photo. Update docs/templates.md to document the new classification.

Can I filter templates by multiple categories at once?

The gpt-image-2-style-library package filters templates by exact string match on the "category" field. To query multiple categories, pass an array of category names to a filter function or execute parallel queries and merge the results. The underlying JSON structure in data/style-library.json stores the category value as a string in each template object.

What is the difference between anchor and templateAnchor?

The anchor field provides a Markdown link anchor for the category's general documentation section, while templateAnchor links directly to the specific template gallery page for that category. In data/style-library.json, these fields enable the homepage tiles in README.md to route users to precise locations within the documentation hierarchy.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}