AI-Assisted Mobile UI Test
Automation
using Claude

The mobile automation landscape, environment setup, and your first green Appium test — then a real login flow. Hands-on in both Node (WebdriverIO) and Python (pytest).

Session 1 (2h) — Introduction & Setup:

• Mobile testing landscape: iOS vs Android, native vs hybrid vs web, APK vs AAB, UiAutomator2 vs Espresso
• Why Appium? Architecture, driver model, and capabilities
• Environment verification and an Appium Inspector walkthrough
• Lab: Run your first green Appium test against the Settings app

Session 2 (2h) — Locators, Commands & Waits:

• Locator strategies: accessibility id (preferred), id, xpath, class name, UiAutomator selectors
• Core commands (tap, setValue, getText) and implicit vs explicit waits — never sleep
• Lab: Automate a login flow on the demo app using locators from Appium Inspector

Turn the login script into a clean, scalable suite — solid synchronization, gestures, data-driven tests, and parallel runs.

Session 1 (2h) — Synchronization & Gestures:

• Synchronization: implicit vs explicit vs fluent waits, and condition-based waiting
• Gestures: scroll-to-element, swipe, drag-and-drop, long-press, double-tap, pinch (mobile: commands vs W3C Actions)
• Labs: Three flavours of waits, and six real gestures with assertions

Session 2 (2h) — Data-Driven & Parallel:

• Data-driven testing: pytest parametrize / WebdriverIO data loops — one flow, many rows
• Parallel testing: maxInstances (Node) / pytest-xdist (Python) — one session per device
• Lab: Run the login suite across two emulators in parallel

Understand AI, let Claude author your tests, then scale the suite — Page Objects, reporting, CI, and cloud devices.

Session 1 (2h) — AI Foundations & Claude-Assisted Authoring:

• AI foundations in plain English: LLMs as prediction engines, tokens & context windows, prompting, and verifying everything
• A real Claude Code project: CLAUDE.md context, the Appium MCP server driving the device, custom slash commands, and a flaky-hunter sub-agent
• Hands-on: Claude authors specs, refactors flat tests into Page Objects, and hunts flakiness

Session 2 (2h) — Scaling the Suite:

• Page Object Model: Login / Catalog / Product, with one end-to-end flow (login → search → add to cart)
• Reporting (Allure / pytest-html with screenshot-on-failure) and CI on GitHub Actions with emulator auto-boot
• Lab: Run the same suite on the BrowserStack real-device grid

Meet taqwright — the Playwright test runner with a flat mobile locator API on Appium 3. Record, run, AI-generate, and debug in the cloud.

Session 1 (2h) — Getting Started with taqwright:

• What taqwright is: the Playwright runner, fixtures and expect, with a mobile fixture over Appium 3 (UiAutomator2 / XCUITest)
• Install & provision (npm init taqwright, doctor, devices); write tests with getById/getByLabel, click, fill and auto-waiting assertions
• Codegen: record a flow, rank locators, and emit a spec

Session 2 (2h) — Scale, AI-Generate & Cloud:

• taqwright test features: projects (Android + iOS), parallelism, parameterize, retries, timeouts, trace viewer, merged reports
• Lab: Run a Page Object spec on BrowserStack — cloud run & debug