Know it works before your users do.
Dev Scenario runs your Android, iOS and Flutter app through real user journeys on every build, and shows you exactly what broke when one fails.
The Dev Scenario app runs the Add to cart scenario on a Pixel 8.
The Network tab lists every call the run makes, mocked ones marked.
Recording in the Dev Scenario app: each tap on the phone becomes a step in the flow.
A regression suite runs on four devices at once, one lane each.
The Zephyr tab: a test cycle's cases, linked to scenarios, fill in with the results of a run.
Hosts
The usual pains, and what we did about them.
Sleeps and retries everywhere, and the suite is still flaky.
Steps wait for their own element. No fixed sleeps, no retry loops.
The backend is down, so every UI test is red.
Mock the calls a scenario needs, SSE and WebSockets too, and the run stops depending on it.
A red build that only says "element not found".
The failed step keeps its screenshot, view hierarchy, network calls and device log.
AI writes the tests now, and nobody knows what they catch.
Mutation testing breaks the UI on purpose and shows which tests noticed.
Test accounts pasted into test files.
Scenarios use keys like @email; the values live in profiles.
One device at a time, so the suite takes all afternoon.
Run on every connected phone, emulator and simulator at once.
Something else annoying you? Tell us on the roadmap.
Built from things that hurt.
Every feature started as a day something annoyed us enough to build the fix.
The fastest runner we could build
1.6–2.4× faster than Espresso and XCUITest on our public benchmark. Think you can beat it? Please try.
Tests you can actually trust
UI mutation testing breaks your UI on purpose and checks whether your tests notice.
- ✓Hidden: "Place order"step 5
- ✓Dead tap: "Pay"step 6
- ✕Text changed: pricesurvived
assertVisible("$129") to catch itOne test, every variant
Languages, users and device states multiply out from a single flow.
The same tests in CI
The same project runs in GitHub Actions. No seats, no run limits.
- ✓Build the app1m 04s
- ✓
devscenario run e2e/24 passed - ✓Upload the report3s
When it breaks, you see why.
Every failed step keeps what you need to fix it: the screenshot, the live view hierarchy, the network calls around it and the device log. No re-running with logging turned up.
- The step, the line and the reason it failed
- Screenshots as evidence after a check passes
- The same report from CI as from your desk
assertVisible("Order placed")not found after 10.0 sYour AI agent, in the loop.
Claude Code, Cursor and GitHub Copilot agents can drive Dev Scenario over MCP: read the screen of a real device, write a scenario, run it, read the report and fix what failed. Or ask the assistant built into the app.
How agents use itget_screenPixel 8 · 14 elementssave_flowcart/add-to-cartrun_scenariostep 3 failedget_run_reportscreen says "Added to bag"update_flow1 line changedrun_scenario4 / 4 passed
Also in the box.
The rest of what a mobile tester reaches for, in the same app as the runner.
Live screenshot and view hierarchy. Click an element, get a working selector.
Save links with variables and fire them at any Android or iOS device.
Labels, contrast and touch targets on the open screen. Android and iOS simulators.
Install the latest App Distribution build in one click (Android), and read Crashlytics and Remote Config.
Run a scenario or chain daily, on an interval or once, while the app is open.
Status, diffs, commits, branches and sync for a project in a repository.
App profiles, test data per profile, and the strings of every language a scenario runs in.
Link scenarios to test cases and send run results to a test cycle.
Made for teams with a security review.
Builds and test data stay on your machines and your devices. Enterprise plans add deployment through Intune or Jamf, a preset company license and settings IT can lock.
Pay for the app. CI is free.
Start with a 6-week trial of every feature, no card needed. Pro and Team are per seat; Enterprise adds managed deployment.
Frequently asked questions
Straight answers about what it does and doesn't do yet.
Does my app or test data leave my network?
Builds and test data stay on your machines. Tests run on your own emulators, simulators and phones. The app sends usage analytics and checks your license online, as described in the privacy policy.
Which platforms does it run on?
Tests run on Android and iOS: emulators, simulators and physical devices, native or Flutter apps. The desktop app is available for macOS today. Windows and Linux are coming.
Does it work with Flutter apps?
Yes. A Flutter app runs like any Android or iOS app, and the Dart packages add the in-app driver and network capture. See the libraries.
Can I keep my existing tests?
Yes. Your current suites keep running as they are; Dev Scenario runs next to them. Start with the journeys that break most often. If your manual cases live in Zephyr Scale, a case can become a draft scenario.
Can I mock WebSockets and SSE?
Yes. Monitoring and mocking live in the same tool as the runner, including SSE and WebSockets on mobile, and it works with apps that need a VPN.
How much faster is it than Espresso and XCUITest?
On our public benchmark it ran 1.6–2.4× faster than Espresso on Android and XCUITest on iOS. The setup and raw numbers are on the benchmark page. If you can beat it, tell us.
Is the CI runner really free?
Yes. The command line tool has no seats and no run limits. You pay for the desktop app, not for running tests.
What is UI mutation testing?
We break your UI on purpose and check whether your tests notice, so a passing run actually means something.
Run your first scenario before standup.
6-week free trial, no card needed. The CI runner is free.