Find the prompt.
Keep creating.

Which images can I read?

AUTOMATIC1111 and Stable Diffusion

Use an original PNG, JPEG or WebP carrying supported generation settings. The reader can recover the saved prompt, negative prompt and model name when they are included.

ComfyUI, including supported Flux workflows

Use a PNG with its saved workflow. isGenAI follows the steps connected to the saved image to find its prompt. Complex workflows can combine several prompts or image inputs, so some have no single prompt to copy.

InvokeAI

Use an original PNG carrying its creation settings. Saved prompt, negative prompt and model details are supported. Files with different main and style prompts need the full workflow rather than a shortened prompt.

Why was no saved prompt found?

Screenshots and social previews often contain only the picture. An export may also remove its saved details, or use a format this reader does not support. Try the original file from its creator.

A missing prompt does not tell you whether a person or AI made the picture. For other clues, use the AI image detector.

Is this the original prompt or a new suggestion?

This tool reads saved text from the file. It does not recreate a prompt from the picture. Saved details can be edited, so they record what is in this copy of the file.

Try a sample before your own image

Download our sample with a saved prompt and the same picture without one, then upload each above. One returns saved text; the other returns no saved prompt.

These are controlled examples: we drew the picture in code and added a sample prompt to one copy. Read how the example works.

Can I use it in my own workflow?

Developers can extract saved prompts through the API with no reconstruction call. See the saved-prompt API example for a runnable Node.js script.