Knowledge check
Ten questions on the decisions you will actually face when writing and running Maestro flows. Pick an answer for each, then check them all at once.
1. A field already contains 0. Your flow taps the field, runs eraseText, types 10, and the field shows 100. What is the most likely cause?
eraseText sends backspace presses, which delete to the left of the cursor. Tapping the right-hand side of the field, for example with point: "90%,50%", puts the cursor after the existing text.
2. The Total field shows 150. Does assertVisible: "15" pass?
Text and id selectors match the full string as a regex. Use .*15.* if you really want a partial match, which here would have hidden the bug.
3. Your app's labels are translated into six languages. Which selector is the most robust for the "Submit" button?
Ids are invisible to users and do not change with the language. Text changes per locale, and screen coordinates change per device.
4. You run maestro test .maestro/, and your flows live in .maestro/auth/ and .maestro/checkout/ with no config.yaml. What runs?
Without a flows pattern in config.yaml, Maestro only runs top-level files. That same default is what keeps subflows/ from running on its own.
5. You run with --include-tags=smoke --exclude-tags=flaky. Which flows run?
Include is applied first, then exclude removes matches. With an include filter, untagged flows never run, which surprises people the first time.
6. A flow's header has env: { A: "7" }. You want maestro test -e A=20 flow.yaml to type 20, and 7 when no value is passed. What should the header say?
In our test run, a fixed constant in the header won over -e. A JavaScript default keeps a passed-in value and falls back to 7 otherwise.
7. A payment confirmation screen sometimes takes 20 seconds to appear and the flow fails with "not visible". Which fix is the right one?
Assertions only wait a few seconds. extendedWaitUntil waits up to an explicit timeout and moves on as soon as the element appears; a whole-flow retry repeats the payment and hides real problems.
8. On iOS, clearState resets your app, but the app still starts logged in with a token it stores in the Keychain. What do you add?
The Keychain is cleared by its own option. stopApp: false resumes a backgrounded app, and permissions are unrelated to stored credentials.
9. On TestingBot, you run 12 flows with no extra options and your plan allows 2 concurrent sessions. What happens?
By default every flow runs in a separate session, and sessions beyond your concurrency are queued. --shard-split 1 would put all flows in one session instead.
10. In CI, the TestingBot CLI exits with code 2. What does that tell you?
0 means all flows passed, 2 means flows failed, and 1 is a CLI or infrastructure error. Keeping them apart lets a pipeline retry infrastructure problems without hiding test failures.
From here, the TestingBot Maestro documentation covers the REST API, tunnels and every run option in more depth. If you also maintain WebDriver-based mobile tests, the Appium with WebdriverIO course is the natural next step.
You finished Maestro Mobile Testing
Put what you learned into practice. Run your tests on 6100+ real browsers and physical iOS and Android devices.