image-prompt-reverse skill
A skill that turns a picture into a generator prompt: frame breakdown, an enhanced version, three variants and platform editions.
- 61 ★ on the repository
- Last commit
- MIT
Minimal access
Why this level
- The skill only analyzes an image and writes prompt text; it has no external calls or file writes except test rendering in WorkBuddy.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/luozhilzh-image-prompt-reverseDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewMinimal accessThe tool runs nothing and makes no network calls.
- The skill only analyzes an image and writes prompt text; it has no external calls or file writes except test rendering in WorkBuddy.
This is third-party code. Review the repository files before installing.
What it does
The skill analyzes an image by dimensions (composition, light, color, materials, style) and builds a prompt from them. Different genres have their own schemes: logo and lettering, landscape, portrait, illustration, product, food, interior, fashion and cars. A fallback scheme covers the rest. The output is a pack: an analysis block, an "as in the picture" version, an enhanced version with 14 directions (cinematic, film, commercial gloss, cyberpunk, Chinese style), three variants shifting composition, light or style, editions for Midjourney, DALL·E, Stable Diffusion and Flux, negative constraints and an originality boundary. The agent asks one or two questions: goal, platform and direction.
Who it is for. Designers and content creators who need to reconstruct a prompt from a picture they like and get variants for their platform.
Good fit when
- A reference exists and a prompt must reproduce its composition and light.
- An enhanced cinematic or film-style version is needed for an ad image.
- The same frame must be prepared for Midjourney and Flux at once.
- Three variants are needed: a different angle, other lighting, another style.
Not a fit when
- A one-to-one copy of someone else work is needed: the pack has an originality boundary that requires differences.
- The picture itself is needed: the skill writes text, and test rendering exists only in WorkBuddy.
- The image must not be shown to a model: the agent reads the frame to break it down.
Example request
Break this picture down and give a Midjourney prompt: as in the reference plus an enhanced version in a cinematic style.Limitations
The skill is written in Chinese, with an English README page. The agent must be able to read images. Step 8, a test image render, is available only in WorkBuddy, needs permission and a built-in ImageGen, and is skipped in other agents. For Claude Code the folder is copied to `~/.claude/skills/`, in Cursor the contents go into an `.mdc` rule, in Codex they are added to `AGENTS.md`. The license is MIT. Results depend on the generation model, and the three variants give directions rather than guaranteed frames.
How to disable. Delete the `image-prompt-reverse` folder from the skills directory (`~/.claude/skills/`, `~/.workbuddy/skills/`) or remove the pasted rule from Cursor and `AGENTS.md`.
FAQ
How do I install the skill in Claude Code?
With `cp -r image-prompt-reverse ~/.claude/skills/`. The agent finds `SKILL.md` itself, and in a new session you only send an image.
How do I connect it to Cursor?
Create `.cursor/rules/image-prompt-reverse.mdc`, paste the `SKILL.md` text and reference `references/*.md`. Use rule type `always` or `auto`.
Which platforms are supported?
Midjourney, DALL·E, Stable Diffusion and Flux: each has its own prompt edition.
What is the enhancement layer?
A library of 14 directions (cinematic light, film, commercial gloss, cyberpunk, Chinese style and others) that can be combined.
How many questions does the skill ask?
A minimum: one or two about goal, platform and enhancement direction.
Discussion
No questions or reviews yet.
Related
Skillsimage-prompt-reverse analyzes an uploaded picture and writes a generator prompt together with a negative-term list.
A Claude Code skill that compiles a vague request into a finished gpt-image-2 prompt and checks it with a validator script
A Codex skill that turns a photograph into a vertical composition with an abstract memory panel and an English title
Nuyoah Image Reverse Prompt
nuyoah-image-reverse-prompt
A skill that breaks a picture into frame, light, color and materials and assembles a ready Chinese prompt for image generation.
Female Portrait Director
Female Portrait Prompt Director Skill
A skill that turns a short set of parameters into a photography-directed portrait prompt for an adult woman, plus negative constraints
A skill that searches a library of 10,000+ image prompts, shows sample images and tailors the chosen prompt to your article or script
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| image-prompt-reverse (this card) | no | 61 | |
| image-prompt-reverse | no | 905 | a different project with the same name; this adds a 14-direction enhancement layer and three variants |
| Nuyoah Image Reverse Prompt | no | 104 | also reconstructs a prompt from an image; this adds editions for four platforms and an enhancement layer |
README in short
The project page lists features and loading the skill into WorkBuddy, Claude Code, Cursor, Codex and any agent with long instructions. The principle is one: the agent reads `SKILL.md` and the `references` files.