black-forest-labs/flux — reverse-engineered prompt

Reverse engineered prompt

GitHub

Build me a simple local app for running FLUX image generation and editing from this repo.

I want a friendly interface where I can type a prompt, choose one of the available FLUX models, generate an image, and save the result. It should also support the editing style workflows this repo mentions, like fill or inpainting, image variation, structural guidance, and Kontext image editing when the right model is selected. Please include clear setup steps for Python 3.10, installing the repo, and any Hugging Face or BFL API key setup that’s needed.

Keep it practical and minimal. Use the existing demo files and inference code instead of reinventing everything. Add sensible error messages for missing models, missing keys, or GPU problems. If commercial usage tracking is enabled, expose that as a simple checkbox or option and document what it does. Look up the current FLUX docs online if you need to.

Want more depth? Deep Reverse