Every premium and members-only resource for $9.99/month, as a no-commitment subscription. See the plans

PalateKit

Menu

Request a resource

Describe the resource you are missing in one sentence. We will email you when it is published.

0 / 280

Unlock PalateKit

Unlock everything

Plan

Paying means accepting our terms of sale.

Secure payment by Stripe. Receipts and invoices sent by email.

Welcome back

Sign in to install, download or save this resource. A free account is enough for free resources.

Locked Portrait Prompt in JSON v1.0.0

Use this resource to build a locked image prompt from reference photos

New

  • 510 downloads
  • 239 likes
  • 119 saves

Free

Install with the CLI

In a terminal, from the folder you want:

npx --yes palatekit@latest install locked-portrait-prompt

With 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 login

The agent studies your reference images, picks an anchor image, pins every variable (face, hair, outfit, pose, lighting, camera) and outputs one JSON prompt without the smooth "AI look". It can also revise the JSON when a generated image drifts from the reference. You get a stable original character across all your generations.

Updated November 2, 2026

Excerpt of a JSON prompt describing the face, outfit, lighting and camera of a portrait.

Details

Generates with
Générateur d'images IA (tout modèle acceptant un prompt)
Licence
MIT

Similar resources

View all
  • Free

    Video

    App window with zoom

    • 365365 downloads
    • 125125 likes
    • 188188 saves
    Works with
    • Claude Code
    • Codex
    • Cursor
    • and 3 more
    + HyperFrames
  • Free

    Video

    One big number

    • 463463 downloads
    • 223223 likes
    • 100100 saves
    Works with
    • Claude Code
    • Codex
    • Cursor
    • and 3 more
    + HyperFrames
  • PRO

    Video

    Three bars head to head

    • 348348 downloads
    • 133133 likes
    • 182182 saves
    Works with
    • Claude Code
    • Codex
    • Cursor
    • and 3 more
    + HyperFrames
  • PRO

    Video

    Before / after slider

    • 322322 downloads
    • 248248 likes
    • 238238 saves
    Works with
    • Claude Code
    • Codex
    • Cursor
    • and 3 more
    + HyperFrames
  • PRO

    Video

    Three-card carousel

    • 418418 downloads
    • 157157 likes
    • 101101 saves
    Works with
    • Claude Code
    • Codex
    • Cursor
    • and 3 more
    + HyperFrames
  • PRO

    Video

    Chat exchange

    • 564564 downloads
    • 127127 likes
    • 161161 saves
    Works with
    • Claude Code
    • Codex
    • Cursor
    • and 3 more
    + HyperFrames
  • PRO

    Video

    Checklist ticking off

    • 442442 downloads
    • 233233 likes
    • 160160 saves
    Works with
    • Claude Code
    • Codex
    • Cursor
    • and 3 more
    + HyperFrames

What you get

  • SKILL.md with locking rules, a variable audit, counter-steering of model defaults and de-slop steps (French version translated)
  • A full JSON schema to extend
  • An iteration mode to fix drift from the reference

Good fit for

  • Keeping the same brand character across a series of visuals
  • Describing a reference portrait precisely for an image model
  • Fixing a prompt whose images drift between generations

Not for

  • Identifying or naming a real person
  • Getting free-form text: the output is JSON only
  • Generating the image itself (the skill only writes the prompt)

Tags

  • portrait
  • prompt
  • json
  • consistency
  • character

About

You say

"Here are three reference photos of my character. Write the locked JSON prompt, using image 2 as the anchor."

Excerpt of a JSON prompt describing the face, outfit, lighting and camera of a portrait.

Skill published by agentara under the MIT license.

Show more

Prerequisites

You bring
One or more reference images and An image generation model to use the prompt

How to use it

  1. Provide one or more images, at least one with a person
  2. Name the anchor image if you have one
  3. Paste the JSON into your image tool, compare, and ask for a revision if needed

Content preview

Files

  • LICENSE1 KB
  • NOTICE-PALATEKIT.md0.4 KB
  • SKILL.md19.7 KB

Origin and licence

Modified by PalateKit

Author
agentara
Source
github.com/agentara/skills/tree/950e1bfb11e1df67b46df19804ed85ada9012c33/skills/aigc/portrait-clone
Licence
MIT

Notice

Version modifiée de « portrait-clone » (agentara, MIT) : traduction française et fiche par PalateKit.

User reviews

–

No ratings yet

Loading reviews…

Delete your review?

Your rating and comment will be removed from the page.

Frequently asked questions

Can I clone a real person's face?

The skill targets a consistent original character: it forbids naming the person or comparing them to a celebrity. Respect image rights before using someone else's photo.

Why is the output JSON only?

That is the skill's rule, so the prompt can be used directly. JSON values stay in English.