Chain-of-thought incantation
Models began reasoning through multi-step problems unprompted, and reasoning models made thinking a setting you control. Telling one to think step by step now works against that.
Scored against 2026, not 2023
Your prompt advice expired.
Take a deep breath. Think step by step. You are a world-class expert. That advice was real once — and models moved on. Prompto scores your prompt against how they actually work now, and names the habits quietly costing you quality.
Three runs a day, no account. Nothing to install.
38/ 100
4 issues · 2 critical
dated pattern
“Take a deep breath and think step by step.”
Inert on a model that already reasons through multi-step problems, and it spends context restating what the model does anyway.
Sound through 2023 · superseded as adaptive reasoning became standard
The differentiator
Nine habits that stopped working
Every prompt tool will tell you to be clearer and more specific. Prompto is the one that knows when a technique was good advice — and what changed. Each finding carries its own receipt.
Role inflation
A role still usefully sets voice and audience. As a quality lever it does nothing — and the superlatives crowd out the specifics that would actually have helped.
Emotional pressure and bribes
Those findings came from a narrower generation of models. Offering a $200 tip the model cannot receive communicates nothing, and eats context a real requirement could have used.
Assistant-turn prefill
Forcing a format by starting the model's reply for it. Structured output made it unnecessary, and it fails outright on models that will not accept a prefill.
Fixed thinking budgets
Pinning a reasoning length made sense when it was yours to set. Now it caps a model that would have allocated its own — paying for thinking you told it not to do.
Defensive boilerplate
Paragraphs of "do not hallucinate" and "be truthful" that read as diligence and function as padding. Length is a cost, not a virtue.
How it works
Paste, score, take the rewrite
Paste your prompt
No setup, no account, no extension. Bring your own API key if you'd rather it ran on your model — that's unlimited on a paid plan.
Read the score
Out of 100, across six dimensions — clarity, context, output contract, failure modes, examples, and model fit. Every issue names what to change and why it matters.
Score the rewrite
You get a paste-ready version back. Load it in and run it again — most tools stop before this bit, so you never find out whether the advice worked.
The rest of it
Built to come back to
Start with a pack, not a blank box
Answer two questions and get six prompts written for the way you work — marketing, engineering, analysis, support or operations. Each one written to the same rubric that scores yours.
A library that follows you
Save what works. Signed in, your prompts live on your account and follow you between browsers. Signed out, they stay in yours. Export the lot whenever you like.
What the length actually costs
Token counts on your prompt and the rewrite, priced per thousand runs. The rubric says padding is a cost rather than a virtue; this is the part that shows you the number.
A changelog, not a black box
Every pattern we check for is published with when it was sound and what changed. When the next technique expires, you'll see the date we added it.
Pricing
Two ninety-five a month
No seats, no trial countdown, no sales call. Cancel any time and keep it to the end of the period you paid for.
Free
A$0
- 3 runs a day with no account
- 10 runs a month once you sign up
- Starter packs and your library
- Every dated pattern we check for
Pro
Prompto Pro
A$2.95/ month, GST included
- Unlimited runs on your own API key
- 50 runs a month on the included model
- Cancel any time, access to period end
Find out what your prompt is carrying.
Three runs a day, free, no account. If it scores well, we'll tell you that too — and leave it alone.
Score a prompt