Where to Find the Full GPT-Image2 Prompt Template Library: Complete File Guide

The complete GPT-Image2 prompt template library lives in two core files within the freestylefly/awesome-gpt-image-2 repository: docs/templates.md for human-readable browsing and data/style-library.json for programmatic integration.

The freestylefly/awesome-gpt-image-2 repository maintains an industrial-grade collection of GPT-Image2 prompt templates categorized by use case. Whether you need to manually copy prompts or automate template selection, this guide provides the exact file paths, URLs, and code snippets to access the full library.

Primary Template Documentation in docs/templates.md

The master template file docs/templates.md contains the complete human-readable library. This markdown file organizes every prompt template into twelve distinct categories: UI, infographic, poster, product, brand, architecture, photo, illustration, character, scene, history, document, and other.

Each entry includes detailed prompt text, JSON-style usage examples, and common pitfalls to avoid. You can view this file directly in the GitHub web interface or fetch it programmatically for local reference.

Machine-Readable Style Library at data/style-library.json

For automated workflows and agent integrations, the repository exposes data/style-library.json. This structured JSON file maps template identifiers to their full metadata, including the prompt text, descriptive tags, and usage notes.

The JSON structure follows this schema:

{
  "tpl-ui": {
    "name": "UI Design",
    "prompt": "Detailed prompt text...",
    "tags": ["interface", "web", "mobile"]
  },
  "tpl-poster": {
    "name": "Poster Design",
    "prompt": "Detailed prompt text...",
    "tags": ["print", "marketing"]
  }
}

Both the web UI and the GPT-Image2 style-library skill consume this file to populate template selections dynamically.

How to Retrieve Templates Programmatically

You can fetch and parse the template library using standard HTTP requests. Below are practical implementations in Python and Bash.

Python: Loading Markdown and JSON Sources

import requests
import json

# Fetch human-readable documentation

md_url = "https://raw.githubusercontent.com/freestylefly/awesome-gpt-image-2/main/docs/templates.md"
md_text = requests.get(md_url).text
print(md_text[:500])  # Preview first 500 characters

# Access structured JSON library

json_url = "https://raw.githubusercontent.com/freestylefly/awesome-gpt-image-2/main/data/style-library.json"
library = requests.get(json_url).json()

# Retrieve specific template by key

ui_template = library["tpl-ui"]
print(f"Name: {ui_template['name']}")
print(f"Prompt: {ui_template['prompt']}")
print(f"Tags: {ui_template['tags']}")

Bash: Using curl and jq

# View markdown documentation in terminal

curl -L https://raw.githubusercontent.com/freestylefly/awesome-gpt-image-2/main/docs/templates.md | less

# Extract specific template metadata using jq

curl -sL https://raw.githubusercontent.com/freestylefly/awesome-gpt-image-2/main/data/style-library.json \
  | jq '.["tpl-poster"]'

These methods enable you to integrate the GPT-Image2 prompt template library into CI/CD pipelines, custom agents, or local development environments.

Agent Skill Integration

The repository includes an agent skill implementation documented in agents/skills/gpt-image-2-style-library/SKILL.md. This skill consumes data/style-library.json to automatically select appropriate templates based on user intent. Review this file to understand how the style library powers automated template selection workflows.

Additionally, the README.md file contains an "Industrial prompt templates and pitfalls guide" section that provides quick navigation links to both the markdown documentation and the JSON data source.

Summary

• docs/templates.md contains the complete human-readable template library with categories and usage guidelines.
• data/style-library.json provides machine-readable access to all templates with metadata and tags.
• Both files are accessible via direct GitHub URLs or raw content endpoints for programmatic retrieval.
• The agent skill at agents/skills/gpt-image-2-style-library/SKILL.md demonstrates production-grade integration patterns.

Frequently Asked Questions

Where is the main GPT-Image2 template file located?

The primary template file is docs/templates.md in the repository root. This markdown file contains all industrial-grade prompt templates organized by category, complete with JSON examples and common pitfalls.

How do I access templates programmatically?

Use the data/style-library.json file endpoint. Parse this JSON to access template objects by key (e.g., tpl-ui, tpl-poster), each containing the prompt text, name, and tags. Fetch via curl or HTTP libraries in Python, JavaScript, or any language.

What categories are covered in the template library?

The library organizes templates into twelve categories: UI, infographic, poster, product, brand, architecture, photo, illustration, character, scene, history, document, and other. Each category contains specialized prompts optimized for GPT-Image2 generation.

How does the agent skill use the template library?

According to agents/skills/gpt-image-2-style-library/SKILL.md, the agent skill reads data/style-library.json to match user requests against available templates. It returns the appropriate prompt text and context, enabling automated template selection without manual file browsing.

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