Extensions have entered the AI age.Install the extension

Turn a rough request into a prompt an agent can follow.

Describe your loose idea in one box. Get a complete prompt with the missing context, boundaries, output format, and checks added for you.

You do not need to fill out a prompt-engineering worksheet. The Prompt Refiner interprets the request, labels the assumptions it had to make, and returns a ready-to-use prompt. It improves the prompt. It does not execute the task inside it.

Your submission is processed through OpenRouter and the selected model host. Do not submit passwords, private keys, or sensitive personal data.

A vague review request becomes testable.

Review this landing page and make it better.

01Status: Needs context

02Review the homepage copy for a first-time SaaS buyer. Identify the five issues most likely to weaken comprehension or conversion.

03For each issue, quote the relevant sentence, explain the problem in plain language, and provide one replacement.

04Preserve factual claims. Do not invent customer proof or product capabilities. Finish with a revised headline, supporting paragraph, and primary CTA.

How it works.

01

Diagnose the gaps

Check whether the draft defines its goal, context, constraints, output, and verification standard.

02

Rewrite without changing intent

Preserve what you meant while making success observable and separating instructions from source material.

03

Test the result

Use the assumptions, missing context, and three quick test cases before saving the approved prompt.

Refine with live browser context and images, place the result back into Moxby Chat for review, and save the approved version as a Smart Shortcut.

Use prompts inside Moxby

Other free Moxby tools.

Questions, answered.

These tools give value before installation and connect the finished plan to the Moxby browser capability built to run it.

Does the tool run my prompt?

No. It improves the prompt and gives you a way to evaluate the result.

Will it work with different models?

Yes. It focuses on durable prompt structure instead of model-specific tricks.

Should I paste private information?

No. Use placeholders for credentials, customer records, private code, and sensitive company data.

Why include test cases?

A reliable prompt should work across realistic inputs instead of being tuned to only one example.

Four connected ways to work in your browser.

Chat on any page, build Mods, run Missions, and add ready-made products from the Marketplace.