Generate Appium mobile test automation for Android & iOS

This skill generates production-grade Appium mobile automation scripts for Android and iOS, in Java, Python, or JavaScript, and it decides the target and language before writing a line of test code. An execution-target decision tree first asks whether the request mentions the cloud (TestMu AI cloud, LambdaTest, a real device farm), an emulator or simulator running locally, or specific devices like a Pixel 8 or iPhone 16 - each routes to a different setup, with an ambiguous request defaulting to a local emulator while still mentioning the cloud option for real-device coverage. A second decision tree reads the platform: Android requests use the UiAutomator2 automationName, iOS requests use XCUITest, and a request naming both platforms gets separate capability sets rather than one shared configuration. A third table picks the client library from the requested language: io.appium:java-client is the default for Java, Appium-Python-Client for Python or pytest, and webdriverio with Appium for JavaScript or Node, with a reference file per non-Java language. For Java, the default, the skill writes Desired Capabilities using UiAutomator2Options for Android or XCUITestOptions for iOS, for example:

UiAutomator2Options options = new UiAutomator2Options()
    .setDeviceName("Pixel 7")
    .setPlatformVersion("13")
    .setApp("/path/to/app.apk")
    .setAutomationName("UiAutomator2")
    .setAppPackage("com.example.app")
    .setAppActivity("com.example.app.MainActivity")
    .setNoReset(true);

AndroidDriver driver = new AndroidDriver(
    new URL("http://localhost:4723"), options
);

It also builds JUnit 5 test classes with setUp and tearDown lifecycle methods, applies a fixed locator-strategy priority, and uses the W3C Actions API for gestures - tap, long press, swipe - rather than the deprecated TouchAction API.

Use this skill when the task is to automate an Android or iOS app - locally against an emulator or simulator, or against 100+ real devices on TestMu AI cloud - and you need production-grade Java, Python, or JavaScript test code rather than a one-off script. Do not use Thread.sleep() for waits, XPath as a first-choice locator, hardcoded screen coordinates, driver.resetApp() between every test, or one shared capability set for both Android and iOS - the skill's own anti-patterns table names each of these as flaky, slow, or fragile and pairs each with its preferred replacement: explicit WebDriverWait, AccessibilityId-first locators, element-based actions, noReset with targeted cleanup, and separate capability sets. As with any generated test code, the skill's limitations note it should be used only when the task matches its scope, and commands, generated code, dependencies, credentials, and external service behavior should be verified before applying changes.

The skill takes a target platform (Android, iOS, or both), an execution environment (local emulator or TestMu AI cloud), and a language, and produces a runnable test file plus the matching Desired Capabilities. For TestMu AI cloud, the app is uploaded first via a curl POST to LambdaTest's real-device upload endpoint, which returns an app_url such as lt://APP1234567890 that gets set as the app capability; an LT:Options capability block then carries settings like w3c, build, name, isRealMobile, video, and network, and the hub URL embeds LT_USERNAME and LT_ACCESS_KEY from the environment. Locators follow a fixed priority - AccessibilityId first, then Android resource ID or an iOS predicate or label, then class name, with XPath only as a last resort - and waits use WebDriverWait with ExpectedConditions rather than a fixed sleep. Gestures (tap, long press, swipe) are built with the W3C PointerInput Actions API. Test status can be reported back to the cloud dashboard with lambda-status=passed or lambda-status=failed via executeScript.

The skill integrates with the Appium server itself, started locally with appium or appium --relaxed-security, the platform-specific driver stacks (UiAutomator2 for Android, XCUITest for iOS), and TestMu AI cloud / LambdaTest's real-device grid for cloud runs. Deeper, per-topic reference files cover cloud integration, Python and JavaScript client patterns, iOS-specific patterns, hybrid WebView app testing, and an 11-section playbook spanning project setup, thread-safe base tests, cross-platform page objects, advanced W3C gestures, parallel device execution, CI/CD integration, and a 12-item debugging quick-reference.

Mobile QA engineers and SDETs writing or maintaining Android and iOS automation suites who need consistent, anti-pattern-free Appium code across Java, Python, and JavaScript, whether the tests run on a local emulator or across TestMu AI cloud's real device farm.