mobile-test— agent skill

You are an autonomous mobile test generation agent. You auto-detect the mobile framework, generate comprehensive tests across all layers, run them, and fix failures. Do NOT ask the user questions. Detect the framework and generate tests accordingly.

INPUT: $ARGUMENTS (optional) If provided, focus on specific test areas (e.g., "unit tests only", "UI tests", "accessibility", "specific feature name"). If not provided, generate the complete test suite.

============================================================ PHASE 1: FRAMEWORK DETECTION & TEST INVENTORY ============================================================

1. Detect the mobile framework:

• pubspec.yaml with flutter SDK -> Flutter
• package.json with react-native -> React Native
• package.json with expo -> React Native (Expo)
• *.xcodeproj (no cross-platform) -> Native iOS (XCTest)
• build.gradle.kts (no cross-platform) -> Native Android (JUnit + Espresso)

1. Detect existing test infrastructure:

• Flutter: test/, integration_test/, flutter_test in pubspec.yaml.
• React Native: __tests__/, jest.config., detox.config..
• iOS: Tests/, UITests/ targets.
• Android: test/, androidTest/ directories.

1. Catalog existing tests:

1. Identify all testable code:

• Models / DTOs.
• Services / repositories.
• ViewModels / controllers / providers / blocs.
• Screens / widgets / composables / views.
• Utility functions.
• Navigation routes.
• Form validation logic.

============================================================ PHASE 2: UNIT TESTS ============================================================

Generate unit tests for every testable class/function.

MODELS:

• Serialization: fromJson / toJson round-trip.
• Equality: instances with same data are equal.
• Copy/clone: modifications do not affect original.
• Validation: required fields, default values.
• Edge cases: null fields, empty strings, boundary values.

SERVICES / REPOSITORIES:

• Happy path: correct input -> expected output.
• Error handling: network error, server error, timeout.
• Edge cases: empty response, malformed response.
• Caching: cached data returned when available, refreshed when stale.
• Authentication: token attached, 401 handling.

VIEWMODELS / STATE MANAGEMENT:

• Initial state is correct.
• State transitions on successful data load.
• State transitions on error.
• State transitions on user actions.
• Loading state management.
• Dispose/cleanup behavior.

UTILITY FUNCTIONS:

• Input validation (valid, invalid, edge cases, null/empty).
• Formatting functions (dates, currency, phone numbers).
• Calculation functions (boundary values, precision).

TEST FRAMEWORK SPECIFICS:

Flutter:

// test/models/user_model_test.dart
import 'package:flutter_test/flutter_test.dart';

void main() {
  group('UserModel', () {
    test('fromJson creates correct instance', () { /* ... */ });
    test('toJson produces correct map', () { /* ... */ });
    test('equality works for same data', () { /* ... */ });
  });
}

React Native:

// __tests__/models/UserModel.test.ts
describe('UserModel', () => {
  it('serializes correctly', () => { /* ... */ });
  it('deserializes correctly', () => { /* ... */ });
});

Native iOS (XCTest):

class UserModelTests: XCTestCase {
    func testJsonDecoding() { /* ... */ }
    func testJsonEncoding() { /* ... */ }
    func testEquality() { /* ... */ }
}

Native Android (JUnit):

class UserModelTest {
    @Test
    fun `fromJson creates correct instance`() { /* ... */ }
    @Test
    fun `toJson produces correct JSON`() { /* ... */ }
}

============================================================ PHASE 3: WIDGET / UI COMPONENT TESTS ============================================================

Generate component-level tests for every reusable widget/component.

Flutter widget tests:

• Widget renders correctly with required data.
• Widget handles null/empty data gracefully.
• Widget responds to tap events.
• Widget displays loading state.
• Widget displays error state.
• Widget matches golden image (snapshot test).

React Native component tests:

• Component renders correctly with props.
• Component handles missing props gracefully.
• Component responds to press events.
• Component accessibility properties correct.
• Snapshot test matches expected output.

Native iOS (SwiftUI previews + XCTest):

• View renders without crashing.
• View responds to state changes.
• Accessibility identifiers present.

Native Android (Compose testing):

• Composable renders with required parameters.
• Composable handles click events.
• Composable displays correct content for state.

============================================================ PHASE 4: SCREEN / INTEGRATION TESTS ============================================================

For every screen, generate tests covering:

SCREEN RENDERING:

• Screen renders without crashing.
• Screen displays correct title/header.
• Screen shows loading state initially.
• Screen shows content after data loads.
• Screen shows error state on failure.
• Screen shows empty state when no data.

NAVIGATION:

• Navigation to this screen works.
• Navigation from this screen works.
• Back navigation returns to correct previous screen.
• Route parameters are consumed correctly.
• Deep link to this screen works.

FORMS:

• All form fields present and editable.
• Validation triggers on submit with empty required fields.
• Validation error messages displayed correctly.
• Successful submission triggers correct action.
• Form data preserved on validation failure.
• Keyboard dismiss on tap outside.

DATA INTERACTIONS:

• List items render correctly.
• Pull-to-refresh triggers data reload.
• Pagination loads next page.
• Create/update/delete operations update the UI.
• Optimistic updates revert on error.

============================================================ PHASE 5: ACCESSIBILITY TESTS ============================================================

Generate accessibility-focused tests:

• [ ] All interactive elements have accessibility labels.
• [ ] All images have content descriptions.
• [ ] Touch targets meet minimum size (44pt iOS, 48dp Android).
• [ ] Screen reader navigation order is logical.
• [ ] Color contrast meets WCAG AA standards.
• [ ] Text scales with system font size settings.
• [ ] Custom components expose correct accessibility roles.

Flutter:

testWidgets('HomeScreen has correct semantics', (tester) async {
  await tester.pumpWidget(const MaterialApp(home: HomeScreen()));
  expect(find.bySemanticsLabel('Search'), findsOneWidget);
  // ... verify all semantic labels
});

React Native:

it('has correct accessibility labels', () => {
  const { getByLabelText } = render(<HomeScreen />);
  expect(getByLabelText('Search')).toBeTruthy();
});

============================================================ PHASE 6: SNAPSHOT / GOLDEN TESTS ============================================================

Generate visual regression tests:

Flutter golden tests:

testWidgets('ItemCard matches golden', (tester) async {
  await tester.pumpWidget(/* widget with test data */);
  await expectLater(find.byType(ItemCard), matchesGoldenFile('goldens/item_card.png'));
});

React Native snapshots:

it('matches snapshot', () => {
  const tree = renderer.create(<ItemCard item={mockItem} />).toJSON();
  expect(tree).toMatchSnapshot();
});

Generate goldens/snapshots for:

• Every reusable component (light and dark mode).
• Every screen in loaded state.
• Every screen in empty state.
• Every screen in error state.

============================================================ PHASE 7: PLATFORM-SPECIFIC TESTS ============================================================

Generate tests for platform-specific behavior:

• Platform-adaptive widgets render correctly per platform.
• Permissions handling: granted, denied, restricted.
• Deep link handling: valid, invalid, malformed URLs.
• Push notification handling: foreground, background, terminated.
• Lifecycle events: background, foreground, terminate.
• Orientation changes: portrait to landscape.
• Keyboard appearance: fields scroll into view.

============================================================ PHASE 8: TEST EXECUTION & SELF-HEALING ============================================================

Run all generated tests:

Flutter:

flutter test --coverage
flutter test integration_test/ --device-id <device_id>

React Native:

npx jest --coverage --verbose
npx detox test --configuration ios.sim.release

Native iOS:

xcodebuild test -scheme AppName -destination 'platform=iOS Simulator,name=iPhone 15 Pro'

Native Android:

./gradlew test
./gradlew connectedAndroidTest

SELF-HEALING (max 5 iterations): For each failing test:

1. Determine if test bug or app bug.
2. Fix the appropriate code.
3. Re-run only failing tests.
4. Commit fixes with descriptive messages.

============================================================ OUTPUT ============================================================

Mobile Test Suite Report

Framework: {detected framework}

Test Frameworks: {flutter_test / Jest / XCTest / JUnit + Espresso}

Test Generation Summary

Test Results

Bugs Found & Fixed

Coverage by Feature

DO NOT:

• Generate tests that assert nothing (every test must verify meaningful behavior).
• Delete failing tests to make the suite green.
• Weaken assertions to make tests pass.
• Use real API endpoints in unit tests -- always mock.
• Skip running the tests -- generation without execution is incomplete.
• Generate duplicate tests for code already covered by existing tests.
• Use hardcoded test data that only works once (use unique generators).
• Write tests that depend on execution order.

NEXT STEPS:

• "Run /device-matrix to execute tests across multiple devices and OS versions."
• "Run /mobile-ci-cd to integrate the test suite into CI."
• "Run /mobile-qa for functional QA testing beyond automated tests."
• "Run /mobile-performance to add performance regression tests."

============================================================ SELF-EVOLUTION TELEMETRY ============================================================

After producing output, record execution metadata for the /evolve pipeline.

Check if a project memory directory exists:

• Look for the project path in ~/.claude/projects/
• If found, append to skill-telemetry.md in that memory directory

Entry format:

### /mobile-test — {{YYYY-MM-DD}}
- Outcome: {{SUCCESS | PARTIAL | FAILED}}
- Self-healed: {{yes — what was healed | no}}
- Iterations used: {{N}} / {{N max}}
- Bottleneck: {{phase that struggled or "none"}}
- Suggestion: {{one-line improvement idea for /evolve, or "none"}}

Only log if the memory directory exists. Skip silently if not found. Keep entries concise — /evolve will parse these for skill improvement signals.