Run & test

Tests that run on live phones and TVs.

Describe a test in a sentence for the AI (beta), or record it once on a live device. Run it on many phones and tablets, or on Apple TV, Android TV and Roku, at once, and see what every step did.

vibeview run-suites --tag smoke --output junit

Start free Read the docs

Show the prompt

Install the VibeView CLI (npm install -g vibeview) and sign in with `vibeview login`. List my test groups with `vibeview list-suites`, run the groups tagged smoke with `vibeview run-suites --tag smoke`, and tell me which tests failed and why, with a link to each run page.

A VibeView test run page: the steps on the left, each with what it did, and the device on the right showing the run's screen

Runs on

iPhone and iPad
Android
Apple TV
Android TV
Roku

Describe it, or record it

Write the test in a sentence and review the steps the AI (beta) drafts, or tap through the flow once on a live device. Maestro flows import too.

Replay first, AI only where needed

A Smart run replays recorded steps and hands only the steps whose screen changed to the AI (beta). A passing run saves the route it took, so the next run replays it without AI.

Many devices, one run

Pick iPhones, iPads and Android phones and tablets, or Apple TV, Android TV and Roku. Each runs in its own session at the same time, and the rest wait in the queue.

Every run shows its work

Each step shows what it did, next to the app log and the device's video when the run has one, and a screenshot check compares each step with its baseline.

Devices

Pick the devices. They run at once.

Choose iPhones, iPads and Android phones and tablets, or Apple TV, Android TV and Roku
Each device runs in its own session at the same time; the rest wait in the queue
Run from the dashboard, the CLI, CI, a coding agent or a schedule

The Run dialog for the Atlas sample app with iPhone 16e, iPhone 17 Pro and iPad Air 11-inch picked from the iOS device list

Results

One row per test, one column per device.

A group run shows every test on every device side by side
Each result links to its run, with how long it took and what it cost
Smart runs replay recorded steps and use the AI (beta) only where the screen changed

A run of all Atlas tests: three tests on iPhone 16e, iPhone 17 Pro and iPad Air 11-inch, 9 of 9 passed

Run page

Every step, with the video next to it.

Each step says whether it replayed or used the AI (beta), and how long it took
The run’s video plays beside the steps, and each step jumps to its moment
The app log of the run is one tab away

Run results docs

A passed Atlas run on iPhone 16e: two steps, one replayed and one by AI, with the run video beside them

Health

See what’s failing before anyone asks.

Pass rate, runs, failing tests and average duration over the last 30 days
Needs attention lists the tests where one of the last 3 runs failed
Filter runs by status, group or trigger, such as the CLI or a coding agent

The Runs overview for Atlas: 100% pass rate over 30 days, runs per day, and a list of runs started by a coding agent and the CLI

How it works

01

Create a test

Describe it in a sentence or record it by using your app on a live device; both end on Review steps, where you name the test and pick its group. Maestro flows import from the Tests page or the CLI.
02

Choose where each test starts

A group sets the start state: Fresh install, Clean app data, Restart the app, or Where the last test stopped. A test can also run other tests before it, such as signing in.
03

Run it on several devices

In the Run dialog, pick the devices and Smart, No AI or AI only. No AI replays exactly what you recorded and never calls the AI.
04
Run it in CI or on a schedule

Tag a group and run it on every pull request with GitHub Actions, or schedule it on chosen days at a set time or on every new build, on paid plans.
```
vibeview run-suites --tag smoke --output junit
```

Everything in it

Describe a test in a sentence

Write what the test should check in plain English. The AI (beta) drafts the steps, and you review them before anything runs.
Record a flow by using the app

Record a test by using your app on a live device, then replay it. When a step fails because the screen changed, a Smart run hands it to the AI (beta).
Import Maestro flows

Each Maestro command becomes a step. Scripts and conditions are flagged for you to look at, and the import uses no AI.
Tests that stop paying for AI

After a run passes, the route the AI took is saved and replayed next time, so those steps use no AI. Checks still use the AI.
Recorded on one, replayed on another

A flow recorded on iOS replays on Android, and one recorded on Apple TV replays on Android TV.
Exact element matching

Replay finds each recorded element by its id, or by its exact text ignoring case and surrounding spaces, never a near match.
Checks in plain English

Write a Verify step such as "Verify the cart shows 2 items". The AI (beta) reads the live screen and judges it.
Screenshot checks

On by default for new groups: each screenshot is compared with its baseline mid-run and marked ok, warn or fail. Change the warn and fail thresholds or switch it off per group or for one step, and accept a step’s screenshot as its new baseline from the run page.
Video and logs for every step

The run page shows the device's video when the run has one, with a jump to each step, and the app log with the step each line belongs to.
Step editor

Fix a recorded step without re-recording the flow: edit a step's element, button press or typed text inline.
Groups and setup

Put tests in groups, run other tests before one (such as signing in), and pin a run to a specific build to find which build broke it.
TV remotes

On Apple TV, Android TV and Roku the AI (beta) moves focus to the element it means instead of chaining arrow presses.
Explore (beta)

The AI uses your app for 5, 10 or 20 minutes and lists what looks broken. Save any finding as a test.
Scheduled runs

Run tests on chosen days at a set time or on every new build, on paid plans.
Results and flaky tests

Pass rate, runs and average duration per group over 30 days, which tests are flaky, and which failed recently and why.
CLI and CI

Run tests and groups from the command line. Results export as JSON and JUnit, and with GitHub connected each group posts a status on the pull request.

Use it for

Run tests on every pull request Build the app, run your tagged groups on that build and report on the pull request. Find bugs without writing tests Explore (beta) uses your app for a set time and lists what looks broken. Streaming apps on every TV Focus and remote checks, deep links and sign-in on Apple TV, Android TV and Roku.

Frequently asked questions

Do my tests need AI?

No. A recorded test run with No AI replays exactly what you did and never calls the AI; a step whose element is gone fails. Smart uses the AI (beta) only for steps with no recording or whose replay fails.

Which devices can a test run on?

iPhone and iPad simulators, Android phone and tablet emulators, Apple TV, Android TV and Roku, all live in the cloud. A phone test never runs on a TV, or the other way round.

Can I bring tests I already have?

Maestro flows import as tests, one step per command, with no AI. Scripts and conditions are flagged for you to look at.

Run your first test

Start free, record a flow on a live device and run it on several at once.

Start free Read the docs

Tests that run on live phones and TVs.

Describe it, or record it

Replay first, AI only where needed

Many devices, one run

Every run shows its work

Pick the devices. They run at once.

One row per test, one column per device.

Every step, with the video next to it.

See what’s failing before anyone asks.

How it works

Create a test

Choose where each test starts

Run it on several devices

Run it in CI or on a schedule

Everything in it

Describe a test in a sentence

Record a flow by using the app

Import Maestro flows

Tests that stop paying for AI

Recorded on one, replayed on another

Exact element matching

Checks in plain English

Screenshot checks

Video and logs for every step

Step editor

Groups and setup

TV remotes

Explore (beta)

Scheduled runs

Results and flaky tests

CLI and CI