What is clickproof?
clickproof runs end-to-end tests on your app without you writing or maintaining any test scripts. You describe what to check in plain English, and an AI vision agent works through your product the way a person would, then returns a video recording and a pass/fail verdict with its reasoning. It tests two surfaces, and the tests are written the same way for both:Websites
Give it a URL. The agent drives a real hosted Chromium browser against your
deployed site or preview environment. Nothing to upload.
Mobile apps
Upload an
.apk or a simulator .app.zip. The agent taps through it on a
real cloud Android device or an Apple Simulator.How it works
1
Create a project
A website project takes a URL. A mobile project takes builds. Either way the
project owns that product’s tests and its run history.
2
Describe a test
Write the goal in plain English - “sign up, then reach the home screen.” No
selectors, no scripts. clickproof turns it into a plan automatically.
3
Run it
The agent drives a real browser or a real device, looks at each screen, and
decides the next action. Everything is recorded.
4
Get a verdict
A pass or fail result, the reasoning behind it, the step-by-step trace, and a
shareable video report.
Built for CI
clickproof is designed to run in your pipeline. The GitHub App tests every pull request and fails the check when a test fails, so a broken flow blocks the merge the way a failing unit test would. It takes the work your CI already does rather than repeating it: for a mobile project, the build artifact your workflow uploads; for a website project, the preview deployment your host has just finished. There is no clickproof workflow file to write and no API key to store.Quickstart
Run your first test in about five minutes, on either surface.
Pull request testing
Gate every pull request on your app’s real behavior.
Reports
Watch the recording and read the agent’s step-by-step trace.
API reference
Drive builds, runs, and results from your own tooling.