Priority: P1 (HIGH)
[!IMPORTANT] Tier 2 (Methodology): Strategy mobile UI/UX audit. Tier 3 (Domain): i18n, A11y (Dynamic Type), Platform (Notch/RTL).
🧪 Testing Mindset
Analyze diff + answer:
- Change? (Affected screen, logic path)
- Break? (Regression, state transition)
- Visual Audit: Truncate, align, z-order, color.
- Behavioral Audit: Tap target, nav, data accuracy.
📋 Scenario Matrix
| Change Type | Scenarios to Run |
|---|---|
| UI/Styling | Visual Audit + Dark Mode + QoS Check (CPU/Mem) |
| Navigation | User Flow + Deep Link + Z-Order |
| Lists/Grids | Scroll Test + Pagination + Empty State |
| i18n/Locale | RTL + Truncate + Locale Logic |
| Accessibility | Dynamic Type + High Contrast + Permission Reason |
🛠️ Core Tool Mapping
| Scenario | Appium Tool |
|---|---|
| System Alert | appium_alert (check hierarchy first) |
| Performance | appium_mobile_performance_data (monitor during flow) |
| Visual Check | appium_screenshot (base/diff) |
| Layout Audit | appium_get_page_source (hierarchy/aria) |
🚫 Anti-Patterns
- Ignore QoS: Apps crash/lag under load. MUST check and report QoS from
appium_mobile_performance_data(scroll/video). - Blind Tap: Check state before interact. Use
appium_screenshot. - Alert Paralysis: Unexpected alert? Use
appium_alert→accept/dismiss. - Happy-Path Bias: Never ignore Empty, Loading, or Error state.
- Deep Link Neglect: Verify "Cold Start" via deep link.
- Single-Device Tunnel Vision: Verify smallest/largest screen size.
📁 Evidence
.appium-mcp/<session>/ holds <AC|step>-<before|after>.png, <label>.source.xml, perf.json (QoS proof), video.url. Driver rung and dir go in the walkthrough driver: / evidence_dir: lines.
🔗 References
- appium-mcp: appium-mcp
- Driver Ladder: driver-ladder — local then cloud, BLOCKED rule.
- Scenario Details: scenarios