Prompt Optimizer

What it is for

Most prompts that fail do not fail because the model is weak. They fail because they ask for something the model cannot infer — an audience, a format, a boundary — and the model fills the gap with an average guess. "Write about our product launch" produces average copy because it is an average brief.

The optimizer restructures a prompt into five parts that between them close the common gaps: who the model should be, what it needs to know, what it should do, how the answer should be shaped, and what it must avoid.

How it works

Paste a prompt and the tool reads it for the five components, flags which are missing, and rebuilds it with the gaps made explicit. The before-and-after score is a rough measure of specificity, not of quality — a well-structured prompt asking for the wrong thing still scores well.

Nothing is sent anywhere. The analysis runs in your browser, so you can paste work material without thinking about it.

When to use it — and when not to

Use it when a prompt you rely on has started producing vaguer output than it used to, when you are converting a one-off prompt into something reusable, or when you are handing a prompt to someone else and it needs to work without you standing over it.

Do not use it on prompts that are already working. Restructuring a prompt that produces good output is a reliable way to make it worse, and the score is not a reason to change something that does its job.

Things worth knowing

The single highest-leverage addition is usually constraints — what the model should not do. "Do not use marketing language", "do not invent statistics", "if the answer is not in the provided text, say so". Models comply with negative instructions more reliably than most people expect, and those three lines fix more bad output than any amount of role-play framing.