Sign in to ProperPrompt

Passkeys only: your face, fingerprint or device PIN. No passwords to remember, nothing for us to leak.

Messy prompt in.
Proper prompt out.

The PromptPerfect successor. Per-platform AI prompts that actually work.

Paste what you would normally type. We restructure it into an AI-optimised prompt, tailored to Claude, ChatGPT, Gemini or Grok, because each one reads prompts differently.

Polish my prompt

5 free rewrites a day. No account needed.

Claude ChatGPT Gemini Grok

Polish a prompt

Start from one of these:

ProperPrompt → Claude
Your polished prompt appears here. Pick a starting point above or paste your own.

 

Why per-platform matters

One prompt does not fit all

Claude reads XML-tagged sections best. ChatGPT follows literal, strict schemas. Gemini wants numbered steps and an explicit output format. Grok wants a stated role and a straight answer on whether to go and look something up. These are vendor-documented differences, so we rewrite for each rather than just tidying your text.

Structured extraction

We pull out your goal, constraints, deadline, acceptance criteria and what only you can decide, then rebuild the prompt around them.

No interrogation

We never ask you 10 questions back. Gaps are inferred, and true unknowns are marked [FILL IN: …] so you can complete them in seconds.

Works in your language

Pick a language and the polished prompt comes back in it, headers and all. Fifteen languages, included on every plan.

In the browser you already use

The extension puts a Polish button straight into Claude, ChatGPT, Gemini and Grok, so you never leave the tab you are working in.

PromptPerfect retires 1 Sep 2026

Looking for a successor? ProperPrompt is the cross-model replacement, built for the 2026 model line-up. Read the migration guide.

Pricing

Free

£0

  • 5 rewrites a day
  • One platform at a time
  • No account needed
Start now

Lifetime

£79 once

  • Everything in Pro
  • One payment, no renewal
  • Early access to new features

Why it works: honest numbers

+30%

Anthropic reports its prompt improver lifted accuracy by 30% on a multilabel classification eval. Structure alone did that.

Source: Anthropic ↗

17.9% → 58.8%

The Chain-of-Thought paper showed structured step-by-step prompting more than tripled GSM8K maths accuracy on PaLM 540B.

Source: Wei et al., 2022 ↗

We cite real published results and will not promise your prompts improve by a fixed %. Structure reliably helps; how much depends on your task.

Get product updates

New platforms, new languages, and the extension when it lands in the store. No more than one email a month.

FAQ

Do I need an account?

No. The free tier works instantly, with 5 rewrites a day per device. Pro adds unlimited use and history.

Is my prompt stored?

Free-tier prompts are processed and discarded. Pro prompt history is opt-in and deletable any time.

What is actually different per platform?

Claude gets XML-tagged sections, which is Anthropic's documented best practice. ChatGPT gets a literal strict schema with capitalised headers. Gemini gets numbered steps plus an explicit output-format block. Grok gets a stated role, markdown sections and an explicit line on whether to search for live information. Same intent, four genuinely different rewrites.

Can I use it in my own language?

Yes. Pick your language from the globe in the top bar and the polished prompt comes back in that language, section headers included. Code, file paths and XML tags stay as they are. It is included on the free tier too.

I used PromptPerfect. Why switch?

PromptPerfect shuts down on 1 September 2026. ProperPrompt is built as its cross-model successor: newer models, per-platform tailoring, simpler pricing. The migration guide walks through moving your prompts over.

Which AI does the rewriting?

Frontier models from Anthropic and Moonshot, behind a prompt-engineering template refined on real production prompts. Which one runs depends on your plan, and the template is the part that matters.