QA.pro
quality assurance · in plain language · on your hardware

QA.pro eats bugs so your users don't.

Quality assurance is someone using your site the way a customer does, before the customer has to. QA.pro is that someone, every night: on a desktop and a folded phone, judging every screen and keeping the screenshot.

Run the tests See the live demo

Open-source engine, Apache 2.0. The dinosaur eats bugs and never went extinct.

It's as easy as…

One command. No account. It puts the open-source panel, the browser engine and a vision model on this machine, on the GPU if it has one, on the CPU otherwise. Then open the panel, paste a shop URL, press run.

curl -fsSL https://qa.pro/local.sh | sh

Needs docker, python 3 and git. Windows: Docker Desktop, or run the Linux line inside WSL2. Ten to thirty minutes, depending on your line: the vision model is 6 GB and the browser image builds for a few minutes. Rather not run anything? Create a team on Cloud: paste a URL, verify it is yours, and our runners do the walking.

Built for the people who get the 11 pm message

Same engine, same evidence. A different first sentence.

Agencies

Thirty client stores. Which one broke last night?

Every site walked nightly on a desktop and a phone, red at the top, a branded share page per run.

03:19Add to cart failed on one store, folded phone. Screenshot attached.
See an agency dashboard → Online stores

Uptime is green. Is your checkout?

Search, filters, cart, coupons and checkout in test mode, every night, on the phones your customers use.

03:12Apply filter fails on iPhone: the drawer covers the grid.
Plans for stores → Engineering teams

The twenty flows that matter, on every preview URL.

A run from CI per deployment, screenshots back in the pull request, runners inside your own network.

14:09Invite red on the preview: the accept button is off-screen at 393px.
Read the CI docs → Solo builders

Your users shouldn't be your QA.

One command, no account, no card. Five tests in plain language and a ping when something goes red.

02:04Login red: the new auth library rejects the test account.
Get the one command →

What you get

Four screens you will actually use. See them with real data: open the demo team.

One dashboard. Every site, its last result, who is online, when it runs next.
Evidence per test, per device. Pass or fail with a reason, the screenshot, the log.
Tests in one line. Type what a visitor must be able to do; the steps are drafted for you (a model refines them where one is reachable); you approve.
A page for your client. Your name on it, no login, revoked in one click. The alert lands in Slack or mail first.

How it works

Three steps, then it runs on its own every night.

  1. 1

    Paste your sites

    QA.pro looks at each one and finds which of the pack's 19 tests it can use. A shop without a wishlist is never "failing" its wishlist test. Write your own test in one line.

    shop.example15 of 19 tests apply
    blog.example9 of 19 tests apply
    no wishlist, no couponsleft out
  2. 2

    Prove it is yours, or bring a runner

    Our runners only test sites you have verified with a file, a meta tag or a DNS record. Or connect the open-source engine on your own machine: no limits, and test accounts stay at home.

    ✓ filemeta tagDNS
    office-gpuyour runner · online
  3. 3

    Read the evidence

    Pass or fail per test and device, a screenshot for anything red, a one-line reason. Share it with a link. Get pinged when it turns red, and again when it is green.

    ✕ Apply filterthe drawer covers the grid
    → #ops, mailscreenshot attached

Pricing

Two doors: Cloud, where our runners do the walking and you can bring your own for unlimited runs, or Self-hosted, everything on your infrastructure. Free for one person on your own machine. Solo $29, Team $149, Agency $349 a month. Per site, not per test; bring your own runner and runs are unlimited.

See all plans Open the live demo