1. Describe the task

What should exist when the model is finished? One outcome.

The expertise or perspective to answer from. Optional.

Background the model cannot guess: audience, situation, what has been tried.

Material the model should use. It is wrapped in tags so it is not confused with instructions.

Steps to follow, one per line. They become a numbered list.

Limits and things to avoid, one per line.

Format, length and structure of the answer.

How a good answer would be judged, one per line.

2. Your prompt

0 words · editable · updates as you type

Built in your browser from a fixed template. No AI model is called and nothing you type is sent to our servers.

How it works

Fill in what you know

Only the objective is needed. Every other field you complete removes a decision the model would otherwise make for you.

Sections are assembled for you

Inputs become labelled sections in a consistent order. Pasted material is wrapped in tags so the model treats it as data, and a rule against invented facts is always added.

Edit, copy, test

The output is editable before you copy it. Run it, read the result, and adjust the constraint or format that was not followed.

Prompt Optimizer →Prompt Comparison →Browse ready-made prompts →