Skip to content

AI prompt optimizer

Paste a prompt you are about to send. You get a score out of 100, what is missing worst first, and a rewritten prompt.

The weak spots and fixes are tuned to this goal.

0 of 5,000 characters

No prompt handy? Try one:

What the optimizer checks

Most weak answers come from a prompt that leaves something out, and the AI fills the gap with a guess. The optimizer scores your prompt on the same seven things every time, so the same prompt gets the same score and you can see whether a change helped.

Goal20 pointsIs the result you want stated, and could you tell when you have it?
Audience15 pointsWho is the answer for?
Format15 pointsWhat shape should the answer come back in?
Constraints15 pointsWhat must it do, and what must it never do?
Context15 pointsDoes the AI have the background it needs, or will it guess?
Tone10 pointsHow should it sound?
Examples10 pointsIs there a sample of a good answer?

The full rubric, with what a failing and a passing prompt look like on each one, is on the prompt quality score page.

Before and after

A short request for a coding task, and the same request after the fixes:

Before

Write tests for this code.

After

Write Vitest tests for calcTotal() in src/utils/price.ts. Put them next to the file as calcTotal.test.ts. Cover an empty cart, one item, and a discount code that has expired. Match the style of src/utils/tax.test.ts. Each test should fail if the total is off by one cent.

Each added line answers one weak spot: which code, which framework and where the tests go, what to cover, what to copy, and what counts as done.

Optimizing a prompt for ChatGPT, Claude or Gemini

ChatGPT

ChatGPT fills gaps with confident filler. Most of its fixes are about the goal and the format: say what done looks like and what shape to answer in, and the long generic intro goes away.

Claude

Claude follows instructions closely, so vague ones cost you. Its fixes lean on constraints and context: what to leave alone, which files or facts to use, and when to stop and ask.

Gemini and others

The seven checks are about how the prompt is put together, so the same fixes carry across tools. A clear goal and stated limits help every model.

Questions

Is the prompt optimizer free?

Yes. You get 3 checks a day with no account and no card, and you see the score and every weak spot. The rewritten prompt is free too, with a free account.

What is the difference between a prompt optimizer and a prompt improver?

People use both names for the same job: take a prompt that gets weak answers and make it get better ones. This one also scores the prompt first, so you can see what changed and why instead of trusting a rewrite you cannot check.

Does it work for system prompts and agent tasks?

Yes. It checks system prompts, coding agent tasks and multi-step workflows as well as one-off questions. For CLAUDE.md and AGENTS.md files, use the CLAUDE.md checker instead.

Do you keep my prompt?

A check without an account is kept for 24 hours so your rewrite is waiting if you sign up. Signed-in checks are saved to your history. Do not paste passwords, keys or private customer data.

What happens after the fix?

When the AI finally gets a job right, you can save how it did it as a memory. Next time it looks that up instead of working it out again. Memories come with Pro.

Checking an agent file instead? Use the CLAUDE.md checker. Want more ways in, like building a prompt from scratch? Open the full checker.