Local prompt utility

Build a better FLUX Kontext edit prompt.

Choose the image-edit task, fill in concrete details, and copy a preservation-first instruction. Nothing is uploaded and no AI credit is used.

Structured fieldsReplace an empty prompt box with task-specific inputs.
No image uploadThe builder runs in your browser.
Copy anywhereUse the result in KontextFlux or another compatible editor.
01 · Select the edit

Replace text

Runs locally
02 · Copy the instruction38 characters

Build and copy the prompt entirely in your browser.

Prompt anatomy

State the change, then constrain everything else.

A useful image-edit prompt identifies one target, describes the replacement or transformation, and explicitly preserves unrelated composition, lighting, perspective, text, and identity.

Name the target

Use the exact visible words, a specific object, or a physical region such as the background behind the product.

Describe the requested state

Give the replacement text, scene, color and finish, object, or one custom change in concrete visual language.

Preserve unrelated details

Keep typography, material, identity, layout, perspective, reflections, and lighting unless the task requires changing them.

When to simplify

One prompt should express one visual decision.

Combining a new background, wardrobe, pose, headline, color palette, and camera angle in one edit makes it harder to know which change caused a failure.

Start with the highest-value edit. Compare the output, then use the next task only when another change is still needed.

Open the image editor when you are ready to apply the prompt to a source image.