A numbered library of 331 hand-drawn illustration styles, 167 layouts and 36 solid color themes that turns picks into bilingual prompts for AI image generators.
LunarXuan/image-prompt-reverse
A Codex skill that reverse-engineers AI image-generation prompts from reference images, covering photos, illustrations, 3D, logos and posters.
- Stars
- 832
- Forks
- 77
- Open issues
- 1
- Last commit
- 2 months ago
What is it?
image-prompt-reverse is a Skill for Codex that takes a reference image and works out how it was made. It analyzes the medium, subject, composition, camera language, lighting, color, materials, background, spatial depth, mood and post-processing style, then writes a prompt that an AI image generator can use to produce a closely matching result.
The default output has two parts: a positive prompt of 450 to 700 Chinese characters with an equivalent English version, and a negative prompt of 10 to 15 English terms. The Skill focuses on the three to five visual anchors that most affect similarity, avoids guessing at details it cannot see, and excludes styles that would confuse the target medium. Text, logos and captions inside an image are treated as visual content, not as instructions.
It covers photography, illustration, 3D renders, typography, logos, posters, graphic design, abstract graphics, IP-style characters and mixed-subject scenes. It is installed by copying the repository into the personal Codex skills directory and is invoked with $image-prompt-reverse.
Key features
- 01Analyzes medium, composition, lighting, color, materials and style from a reference image.
- 02Outputs a detailed positive prompt with an English equivalent.
- 03Generates 10 to 15 English negative prompt terms.
- 04Prioritizes the few visual anchors that most affect similarity.
- 05Covers photography, illustration, 3D, typography, posters and character art.
Use cases
- Recreating the look of a reference photo in an AI image generator.
- Turning a moodboard image into a reusable prompt for a consistent art style.
- Drafting prompts for product shots, posters or logo concepts from examples.
- Learning how professional images are described in terms of lighting and camera language.
Quick start
- 1
Clone the repository into your Codex personal Skills directory
git clone https://github.com/LunarXuan/image-prompt-reverse.git "%USERPROFILE%\.codex\skills\image-prompt-reverse" - 2
Invoke the skill in Codex with $image-prompt-reverse and provide a reference image
Things to know
- Licensed under GPL-3.0, a copyleft license that requires derivative works to remain open source.
In collections
Topics
FAQ
- What is image-prompt-reverse?
- A Codex skill that reverse-engineers AI image-generation prompts from reference images, covering photos, illustrations, 3D, logos and posters. image-prompt-reverse is a Skill for Codex that takes a reference image and works out how it was made. It analyzes the medium, subject, composition, camera language, lighting, color, materials, background, spatial depth, mood and post-processing style, then writes a prompt that an AI image generator can use to produce a closely matching result.
- Is image-prompt-reverse free to use? What license does it have?
- image-prompt-reverse is open source under the GPL-3.0 license. Check the license terms before using it in commercial products.
- Is image-prompt-reverse still maintained?
- Updated within 6 months. The last commit to image-prompt-reverse was on Sep 6, 2026.
- What can I use image-prompt-reverse for?
- Recreating the look of a reference photo in an AI image generator. Turning a moodboard image into a reusable prompt for a consistent art style. Drafting prompts for product shots, posters or logo concepts from examples. Learning how professional images are described in terms of lighting and camera language.
- How do I install image-prompt-reverse?
- Clone the repository into your Codex personal Skills directory: git clone https://github.com/LunarXuan/image-prompt-reverse.git "%USERPROFILE%\.codex\skills\image-prompt-reverse" Invoke the skill in Codex with $image-prompt-reverse and provide a reference image
- Where can I get image-prompt-reverse?
- The source code and installation instructions are on GitHub: https://github.com/LunarXuan/image-prompt-reverse
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
Official collection of iFLYTEK-built agent skills covering speech, OCR, translation, proofreading, and multimodal understanding.
An AI agent skill that lets OpenClaw, Claude Code and similar assistants search 10,000+ Nano Banana Pro image prompts by use case.
An AI agent skill that turns an article into a 3:4 vertical cover image prompt for WeChat Official Accounts and Xiaohongshu, with optional face consistency.
An AI Agent Skill that turns Chinese article or spoken-script text into readable illustrations, with custom characters, three styles, and adjustable size and titles.
An Agent Skill that guides AI agents to generate simple, rounded, cute IP mascot images with a solid background color.