FreeVideo
App window with zoom
- 365365 downloads
- 125125 likes
- 188188 saves
Works with
+ HyperFrames- Claude Code
- Codex
- Cursor
- and 3 more
Recover the Prompt of a Reference Image v1.0.0
Use this resource to recreate the look of a reference image with a generation prompt
New
- 422 downloads
- 177 likes
- 210 saves
$4.99
Lifetime access to this resource, with its updates.
Install with the CLI
In a terminal, from the folder you want:
npx --yes palatekit@latest install reverse-engineer-image-promptWith no option, the skill goes to ~/.agents/skills/; add --agent claude-code (or codex, cursor…) to target an agent.
First time: connect the CLI with
npx --yes palatekit@latest loginThe agent examines your image (photo, illustration, 3D render, product shot, poster) and extracts the visual anchors that matter: composition, light, colour, materials, depth layers, mood. It delivers a positive and a negative prompt ready to paste into a generator. Brands and logos are translated into shapes and colours instead of protected names.
Mostly picked by
Updated November 2, 2026
Previews
Details
- Install guides
- Generates with
- GPT Image 2
- Licence
- MIT
Similar resources
View all- Free
FreeVideo
One big number
- 463463 downloads
- 223223 likes
- 100100 saves
Works with+ HyperFrames- Claude Code
- Codex
- Cursor
- and 3 more
PROVideo
PalateKit Motion, the library
- 338338 downloads
- 198198 likes
- 249249 saves
Works with+ HyperFrames- Claude Code
- Codex
- Cursor
- and 3 more
PROVideo
Three bars head to head
- 348348 downloads
- 133133 likes
- 182182 saves
Works with+ HyperFrames- Claude Code
- Codex
- Cursor
- and 3 more
PROVideo
Before / after slider
- 322322 downloads
- 248248 likes
- 238238 saves
Works with+ HyperFrames- Claude Code
- Codex
- Cursor
- and 3 more
PROVideo
Three-card carousel
- 418418 downloads
- 157157 likes
- 101101 saves
Works with+ HyperFrames- Claude Code
- Codex
- Cursor
- and 3 more
PROVideo
Chat exchange
- 564564 downloads
- 127127 likes
- 161161 saves
Works with+ HyperFrames- Claude Code
- Codex
- Cursor
- and 3 more
PROVideo
Checklist ticking off
- 442442 downloads
- 233233 likes
- 160160 saves
Works with+ HyperFrames- Claude Code
- Codex
- Cursor
- and 3 more
What you get
- SKILL.md: method, medium boundaries and output format
- General visual analysis framework (analysis-framework.md)
- Per-subject guides: photo, product, character, typography and more
- Dedicated rules for illustration styles
Good fit for
- Reproducing the mood of an inspiration image for a campaign
- Building a prompt library from brand visuals
- Understanding how a poster or product photo is constructed
Not for
- Extracting text from an image (OCR)
- Faithfully copying a protected artwork or logo
- Getting a plain description of an image
Tags
- prompt
- reverse engineering
- image
- reference
- style
About
You say
"Here is a product photo I like: give me a prompt to recreate the same style with my own product."
Skill published by wuyoscar under the MIT license.
Show moreShow less
Prerequisites
- You bring
- A reference image and The target generation model (optional)
How to use it
- Install the skill in your agent's skills folder
- Attach the reference image and name the target generation model
- Paste the positive and negative prompts into your generator and adjust
Content preview
Files
- LICENSE1 KB
- NOTICE-PALATEKIT.md0.4 KB
- references/analysis-framework.md6.1 KB
- references/category-guides.md7.8 KB
- references/illustration-style.md5.6 KB
- SKILL.md7.3 KB
Origin and licence
Modified by PalateKit
- Author
- wuyoscar
- Source
- github.com/wuyoscar/gpt-image2-skill/tree/9f8aa1a44a4a5c0f5a5d6e45eed564b3e2d5b87f/skills/get-prompt-from-image
- Licence
- MIT
Notice
Version modifiée de « get-prompt-from-image » (wuyoscar, MIT) : traduction française et fiche par PalateKit.
User reviews
–
No ratings yet
Loading reviews…
Frequently asked questions
Why is there a Chinese prompt?
By default the skill outputs a Chinese and an English version; just ask for French or English only to change that.
Is my image stored or sent anywhere?
The skill ships no scripts: the agent reads the image you provide and the skill itself sends nothing.