cuozg/flutter-test

flutter-test

Detect test type, pick mode, write comprehensive tests. AAA pattern, mocktail mocking, 10+ tests per class.

Step 1 — Detect Mode

| Signal | Mode | |--------|------| | Provider, service, repository, model, utility, pure logic class | Unit | | Widget, screen, component, UI element, form validation display | Widget | | Full app flow, navigation, multi-screen journey, end-to-end | Integration |

State triage: "This is [mode] — [reason]."

Step 2 — Execute

Unit Mode

Load read_skill_file("flutter-test", "references/unit-mode.md") for patterns.

1. Read — read the target class; identify public API, dependencies, state mutations
2. List — enumerate 10+ test cases: happy paths, edge cases, error paths, boundary values, null/empty inputs
3. Mock — create mocks with mocktail for all dependencies; load read_skill_file("flutter-test", "references/mocking-patterns.md")
4. Write — implement tests using AAA pattern; group() by method, setUp/tearDown for isolation
5. Verify — lsp_diagnostics on test file; confirm all imports resolve

Widget Mode

Load read_skill_file("flutter-test", "references/widget-mode.md") for patterns.

1. Read — read the widget class; identify props, interactions, conditional rendering, provider dependencies
2. List — enumerate 10+ test cases: render states, tap actions, text input, error display, loading/empty states
3. Setup — create pump helper with ProviderScope overrides and MaterialApp wrapper
4. Write — implement tests with testWidgets, finders, pumpAndSettle, interaction gestures
5. Verify — lsp_diagnostics on test file

Integration Mode

Load read_skill_file("flutter-test", "references/integration-mode.md") for patterns.

1. Read — read the app entry point, router config, and screens involved in the flow
2. List — enumerate test scenarios for the complete user journey
3. Setup — configure IntegrationTestWidgetsFlutterBinding, mock external services
4. Write — implement end-to-end tests with navigation, state changes across screens, assertions at each step
5. Verify — lsp_diagnostics on test file

Rules

• AAA pattern in every test body — Arrange, Act, Assert with comments
• mocktail only — never use mockito; Mock, Fake, registerFallbackValue
• 10+ test cases per class under test — happy paths, edge cases, error paths, boundaries
• One assertion focus per test — multiple related asserts OK, but keep tests atomic
• ProviderContainer for provider tests — addTearDown(container.dispose) always
• Descriptive names — 'returns User on success' not 'test1'
• setUp/tearDown for isolation — never share mutable state across tests
• Mock at boundaries — repositories, API clients, external services — not internal classes
• Test file mirrors source — lib/features/auth/data/auth_repo.dart → test/features/auth/data/auth_repo_test.dart
• No print/debugPrint in tests — use expect assertions only

Escalation

| From | To | When | |------|----|------| | Unit | Widget | Test requires rendering UI to verify behavior | | Widget | Unit | Widget test is testing pure logic — extract to unit test | | Widget | Integration | Test needs navigation or multi-screen state | | Any | Unit+Widget | Feature needs both logic and UI coverage |

Carry forward context; tell user why mode changed.

Standards

Load on demand via read_skill_file("flutter-standards", "references/<path>"):

• testing-patterns.md — Foundation patterns for all test types
• dependency-injection.md — Riverpod DI patterns for testable architecture
• state-management-guide.md — Provider patterns relevant to test setup
• error-handling.md — Error types and exception testing patterns
• code-organization.md — Test file placement, feature-first test structure