Skip to content
This repository was archived by the owner on May 19, 2026. It is now read-only.

rzem-ai/flux1-prompt-engineer

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 
 
 

Repository files navigation

FLUX.1 Prompt Engineer Skill Package

License: MIT FLUX.1 Claude Skill

Overview

When installed, Claude will automatically use this skill whenever users request AI-generated images.

What's Inside

SKILL.md (Main Instructions)

  • Core workflow for generating three prompt variants
  • Prompt construction principles (specificity, layering, dynamics)
  • Anti-patterns to avoid
  • Quick examples
  • References to advanced techniques

Reference Files (Loaded on demand)

  1. layering-hierarchy.md - Foreground/middle/background composition techniques
  2. contrasts.md - Visual impact through color and aesthetic contrasts
  3. text-rendering.md - Font selection, styles, and text integration
  4. camera-settings.md - Technical parameters (cameras, lenses, ISO, aperture)
  5. materials-textures.md - Transparent materials, reflections, atmospheric effects
  6. advanced-techniques.md - Artistic references, concept blending, temporal narratives

How It Works

  1. User requests an image → Claude automatically triggers this skill
  2. Claude analyzes the request → Identifies key elements, mood, style
  3. Claude generates three variants → Each in a code block for easy copying
  4. Advanced requests → Claude reads relevant reference files as needed

Key Features

  • Progressive disclosure - Only loads detailed references when needed
  • Token-efficient - Concise main instructions, comprehensive references
  • Three variants - Gives users options for different interpretations
  • Best practices - Incorporates all techniques from your original guides
  • Anti-pattern awareness - Avoids Stable Diffusion syntax and common mistakes

Installation

Upload the flux-prompt-engineer.zip file to Claude as a custom skill. Once installed, simply ask Claude to create images and it will automatically use this skill to generate optimized FLUX.1 prompts.

Example Usage

User: "Create a cyberpunk street scene"

Claude will output three code blocks:

[Detailed prompt with layering, technical specs, mood]
[Alternative interpretation with different angle/time]
[Creative variation with unique elements]

Structure

flux-prompt-engineer/
├── SKILL.md (4.5K - always loaded)
└── references/ (23K - loaded on demand)
    ├── layering-hierarchy.md
    ├── contrasts.md
    ├── text-rendering.md
    ├── camera-settings.md
    ├── materials-textures.md
    └── advanced-techniques.md

Configuration Notes

  • Optimized for FLUX.1 [dev] variant
  • Default settings:
    • Guidance Scale: 3.5
    • Steps: 50
    • Dimensions: 1024x1024

About

A FLUX.1 text-to-image generation skill converts natural language requests into optimized prompts that produce high-quality AI-generated images by applying expert techniques for composition, style, lighting, and technical parameters.

Resources

License

Stars

1 star

Watchers

0 watching

Forks

Packages

 
 
 

Contributors