Ship it when you can prove it works.
clickproof runs your end-to-end tests on real Android and iOS devices and in a real browser. Describe the test in plain English; get back a video, a step-by-step report, and a verdict on every pull request.
“Sign up, add an item to the cart, and check out.”
- Tapped “Create account”
- Filled email and password
- Added “Cotton tee” to the cart
- Opening checkout
- Confirm the order total
“Sign up, add an item to the cart, and check out.”
- Clicked “Create account”
- Filled email and password
- Added “Cotton tee” to the cart
- Opened checkout
- Order total was $24.00
One sentence. Two surfaces.
The same plain-English description drives a real phone or a real browser. Nothing about how you write a test changes when the surface does - which means one testing habit covers your app and your marketing site and your dashboard.
Upload the build your CI already produces.
Drop in an .apk or .ipa and the agent drives it on real cloud Android and iOS hardware - not a simulator. Real timing, real keyboards, real system dialogs, real permission prompts.
- Android and iOS, on real devices
- Flutter, React Native, native and hybrid
- No SDK, no instrumentation, no code changes
- Deep links and saved test accounts
Or just point us at a URL.
Web tests run in a real hosted Chromium at full resolution - the browser your users have, not a headless approximation. The agent reads the rendered page and clicks where a person would.
- Any framework, any host, any stack
- Staging behind basic auth or a cookie
- Console and network evidence per step
- No device minutes - web runs are cheaper
From build to answer in three steps.
The whole loop takes about a minute, and none of it involves writing a script.
- 01
Point us at the thing
Upload a build straight from CI, or give us a URL. There is no SDK to install, nothing to instrument, and no test script to write.
- 02
Describe the test in English
“Sign up, add an item to the cart, and check out.” That sentence is the whole spec - the same sentence works on the phone and in the browser.
- 03
Get a video and a verdict
An agent taps or clicks through your app like a user and tells you what broke, with a screenshot and a reason for every step.
Catch it in the pull request, not in production.
Install the GitHub App once and every pull request gets a real end-to-end run. Results come back as a Check Run and a comment, with a linked recording for each test, so a reviewer can watch the failure instead of guessing at it.
- Runs automatically on every push to a PR
- Comment on demand with “@clickproof-ai run”
- Web checks need no build at all - just a preview URL
- Or call the API from any pipeline with the GitHub Action
- Sign up and check outreport
- Apply a discount codereport
- Guest checkoutPay button never enabledreport
Posted automatically on every push. Each report links to the recording of that run.
Everything you need to trust “it works”.
Built for the tests that matter most and break the hardest: the ones that go all the way through your product.
Real devices, real browsers
Cloud Android and iOS hardware for apps; a full-resolution hosted Chromium for the web. Nothing under test is emulated.
Plain-English tests
No Appium, no Playwright selectors, no brittle scripts. Write what a user would do and let the agent work out the taps.
Video and step reports
Every run returns a scrubbable recording plus a screenshot and a stated reason for each step it took.
Runs on every PR
Install the GitHub App and each pull request gets a real end-to-end run - on a device, in a browser, or both - before it merges.
Survives redesigns
Because the agent reads the screen instead of matching selectors, moving a button rarely breaks a test.
Teams and billing built in
Organizations, seats, run allowances, and metered device-minutes out of the box.
Simple, usage-based pricing.
Start free. Upgrade when you ship more. Every plan includes every feature, on both surfaces.
Try it end to end
- 10 runs / month
- 1 concurrent runs
- 30 device-minutes
- 1 seat
- Live browser view + cursor
- Multi-tab replay
- Console + network logs
- GitHub App + Action
For solo builders
- 100 runs / month
- 2 concurrent runs
- 200 device-minutes
- 3 seats
- 10 projects
- Live browser view + cursor
- Multi-tab replay + debugging logs
- Auto-run on pull requests
For product teams
- 600 runs / month
- 5 concurrent runs
- 1,200 device-minutes
- Unlimited members
- Unlimited projects
- Unlimited team members
- Live browser observability
- Multi-tab replay + console/network
- Scheduled runs
- Priority email support
For heavy CI usage
- 1,200 runs / month
- 10 concurrent runs
- 2,400 device-minutes
- Unlimited members
- Everything in Team
- Live browser observability
- Multi-tab replay + console/network
- Priority run queue
- Onboarding session
- Support SLA
Custom run volume, dedicated devices, SSO/SAML, security review, and invoicing.
A run is one test executed on a real device or in a real browser, capped at 10 minutes. Web and Android runs count as one; iOS runs count as two, since an iOS device costs twice as much to run. Runs pause once you reach your monthly allowance and resume at the start of the next cycle.
Questions, answered.
Can it test web apps as well as mobile apps?
Yes, and from the same test description. A mobile project takes an .apk or .ipa and runs it on real cloud Android or iOS hardware. A web project takes a URL and runs it in a real hosted Chromium at full resolution. A project is one or the other, because the two UIs are genuinely different, but nothing about how you write a test changes between them.
Do I need to change my app or add an SDK?
No. On mobile you upload a build exactly as your CI produces it - there is no SDK, no instrumentation, and no code change. On the web there is nothing to install at all; you give us a URL. The agent drives the product through the device or the browser, the same way a person would.
How is this different from Appium, Maestro or Playwright?
Those need you to write and maintain scripts against selectors, which break whenever the UI moves. Here the test is a sentence in plain English, and a vision agent decides what to tap or click. When your layout changes, the test usually still passes.
What happens when a test fails?
You get the reason in plain language, a screenshot of every step, and a video of the whole run. Web runs additionally carry console and network evidence for the step that failed. On a pull request, all of that lands as a Check Run and a comment with a link to each report.
Does it run on real devices and real browsers?
Yes. Real cloud Android and iOS devices, not simulators, which means real timing, real keyboards and real system dialogs. Web runs use a real hosted Chromium at a full desktop viewport, not a headless approximation.
Can it reach staging, or anything behind a login?
Yes. Web projects support sites behind basic auth or a session cookie, and both surfaces let you store test-account credentials against the project so the agent can sign in. Stored credentials are encrypted and only ever decrypted inside the run container.
Can I run it in CI?
Two ways. Install the GitHub App and tests run automatically on every pull request, or call the API from any pipeline with the GitHub Action. Web checks are the easiest to adopt, because they need only a preview URL rather than a build artifact.
What counts as a run?
One test executed once. Free includes 10 runs to try the whole flow end to end; paid plans renew their allowance every billing period. A web run and an Android run each count as one. An iOS run counts as two, because iOS device time costs us twice as much.
Test your app or your site in the next five minutes.
Upload a build or paste a URL, write one sentence, and watch it run. 10 free runs, no card required.