Complete Guide to Prompt Template Categories in Awesome-GPT-Image-2

The Awesome-GPT-Image-2 repository organizes its visual generation assets into 13 distinct prompt template categories, each providing domain-specific structural schemas and automation-friendly formats for AI image generation.

The freestylefly/awesome-gpt-image-2 project implements a prompt-as-code architecture designed to standardize AI image generation workflows. These prompt template categories group reusable assets by visual style and functional purpose, enabling developers and designers to select pre-validated structural skeletons that minimize ambiguous outputs.

The 13 Prompt Template Categories

The repository defines 13 distinct categories in docs/templates.md, each anchored to a specific section ID and indexed in data/style-library.json. Each category contains one or more industrial-grade templates available in both human-readable plain-text and structured JSON formats.

• UI & Interfaces (docs/templates.md#tpl-ui): App screens, web dashboards, and mobile mock-ups requiring high-fidelity layouts with readable text constraints.

• Charts & Infographics (docs/templates.md#tpl-infographic): Data visualizations, knowledge maps, and statistical graphics with structured information hierarchies.

• Posters & Typography (docs/templates.md#tpl-poster): Event posters, cover art, and text-driven compositions emphasizing typographic hierarchy.

• Products & E-commerce (docs/templates.md#tpl-product): Product-shot layouts, packaging designs, and promotional advertisements.

• Brand & Logos (docs/templates.md#tpl-brand): Visual identity systems, logo variations, and comprehensive brand asset generation.

• Architecture & Spaces (docs/templates.md#tpl-architecture): Building renders, interior designs, and city-scape conceptual visualizations.

• Photography & Realism (docs/templates.md#tpl-photo): Portraits, lifestyle shots, and cinematic stills requiring photorealistic output.

• Illustration & Art (docs/templates.md#tpl-illustration): Painterly styles, decorative illustrations, and experimental artistic workflows.

• Characters & People (docs/templates.md#tpl-character): Character designs, avatars, and figurative studies with consistent anatomical constraints.

• Scenes & Storytelling (docs/templates.md#tpl-scene): Narrative frames, storyboard panels, and world-building visual sequences.

• History & Classical Chinese Themes (docs/templates.md#tpl-history): Traditional scrolls, historic portraits, and cultural motifs specific to classical aesthetics.

• Documents & Publishing (docs/templates.md#tpl-document): Manuals, white-papers, and editorial spreads with structured page layouts.

• Other Applications (docs/templates.md#tpl-other): Experimental, mixed-workflow, or niche use-cases not covered by standard classifications.

Each category includes an "avoid-pitfalls" checklist that enforces domain-specific constraints, such as mandating readable text for UI screenshots or ensuring cultural accuracy for historical themes.

Template Formats and JSON Schemas

Every prompt template category supports dual-format implementations to accommodate both manual editing and programmatic automation.

The JSON variants follow strict schemas designed for direct API consumption. The UI & Interfaces category, for example, uses this agent-ready structure:

{
  "type": "UI Screenshot",
  "platform": "iOS",
  "product": "Fitness App",
  "layout": "Card‑based feed with bottom tab bar",
  "style": {
    "theme": "Dark Mode",
    "primary_color": "Neon Green",
    "typography": "Clean sans‑serif"
  },
  "content": {
    "header": "Today's Activity",
    "cards": [
      {"title": "Running", "data": "5.2 km", "button": "Start"},
      {"title": "Calories", "data": "340 kcal"}
    ]
  },
  "constraints": "High fidelity, readable text, 9:16 aspect ratio"
}

Source: docs/templates.md – UI template section

Plain-text templates prioritize human readability while maintaining structured fields. The Charts & Infographics category uses this format:

# Charts & Infographics – plain‑text version

生成[主题：明确、具体，避免宽泛。例如：“老年人日常健康管理指南”]信息图，目标读者为[人群:细化人群特征，如年龄段、职业、兴趣等]。
结构：标题区 + [3‑5]个模块（每模块含图标、短标题、1‑2句说明）。
图表类型：[流程图/对比图/关系图/时间线]。
风格：[专业报告/科普插画/儿童教育等]，主色[颜色]，背景[浅色/深色]。
输出：信息层级清晰、可读性高的中文信息图。

Source: docs/templates.md – Infographic template section

Source Code Architecture and Implementation

The prompt template system relies on four core files that connect category definitions to generation endpoints:

• docs/templates.md: Contains the canonical list of all 21+ industrial templates with category anchors (e.g., #tpl-ui, #tpl-infographic) and constraint specifications.

• data/style-library.json: Indexes template anchors used by the gallery UI to filter cases by category, enabling rapid template lookup without parsing the full documentation.

• README.md: Provides the Category Overview section with human-friendly navigation links and visual summaries of each category's purpose.

• src/apimartClient.js: Implements the submitPlatformGeneration API wrapper that transmits JSON template payloads to the APIMart backend, handling authentication and response parsing for automated workflows.

When generating images programmatically, agents reference data/style-library.json to locate the appropriate template anchor, retrieve the full schema from docs/templates.md, and submit structured payloads via src/apimartClient.js.

Summary

• The awesome-gpt-image-2 repository defines 13 prompt template categories spanning UI design, data visualization, branding, and historical art styles.

• Each category provides dual-format templates: JSON schemas for programmatic access and plain-text versions for manual editing.

• Template anchors are indexed in data/style-library.json and fully documented in docs/templates.md with domain-specific constraint checklists.

• The src/apimartClient.js module enables direct API integration, consuming JSON templates for automated generation workflows.

• Categories enforce modularity and domain constraints to reduce hallucination and ensure consistent visual output.

Frequently Asked Questions

How do I choose between JSON and plain-text prompt templates?

Select JSON templates when integrating with automated agents or APIs such as the submitPlatformGeneration function in src/apimartClient.js, as they provide structured typing and validation. Use plain-text templates for manual prompt engineering or when you need to quickly modify descriptive constraints without editing schema syntax.

What is the difference between a template category and an individual template?

A template category (such as "UI & Interfaces") represents a visual domain containing multiple specialized templates, while an individual template is a specific structural scaffold (plain-text or JSON) designed for a narrow use case within that domain. The repository organizes 21+ individual templates across the 13 high-level categories.

How are prompt template categories implemented in the codebase?

Categories are implemented as markdown anchors in docs/templates.md (e.g., #tpl-product) and as JSON indices in data/style-library.json. The README's Category Overview section provides human navigation, while src/apimartClient.js processes the JSON payloads associated with each category anchor.

Can I use these templates with my own AI image generation service?

Yes. The plain-text templates can be copied directly into any chat interface or API prompt field. The JSON templates require integration with an API client that accepts structured payloads, similar to the submitPlatformGeneration method implemented in src/apimartClient.js. Both formats include platform-agnostic constraint specifications that work with GPT-Image-2 and compatible diffusion models.

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