Mobile Engineering Agents

Appearance

Standard: Testing Standards ​

Rules for tests. Enforced by the Testing Expert. See the testing skills.

Pyramid ​

  • Many fast unit tests (logic), some integration tests (collaboration), few UI tests (critical journeys). Don't invert the pyramid.

Principles (FIRST) ​

  • Fast — milliseconds; no real network/disk/sleep.
  • Isolated — no shared state or test ordering dependencies.
  • Repeatable — deterministic; inject Date, ids, randomness, and clients.
  • Self-validating — clear pass/fail; no manual inspection.
  • Timely — written with (or before) the code.

Structure & Naming ​

  • AAA: Arrange, Act, Assert. One behavior focus per test.
  • Name by behavior: method_condition_expectedResult.
  • Prefer Swift Testing (@Test/#expect) for new code; XCTest acceptable for legacy.
swift
@Test func transfer_withInsufficientFunds_throws() async {
    let sut = TransferUseCase(repository: StubAccountRepository(balanceCents: 0))
    await #expect(throws: TransferError.insufficientFunds) {
        try await sut.execute(amountCents: 100, to: "x")
    }
}

Coverage Expectations ​

  • Use cases, ViewModels, mappers, and repositories: cover success and error/empty/boundary paths.
  • Every bug fix ships with a failing-first regression test.
  • Aim for meaningful coverage of business logic; don't chase a coverage % with trivial tests.

Test Doubles ​

  • Prefer simple fakes/stubs over heavy mocking frameworks.
  • Doubles live in a TestSupport/Mocks target; fixtures stored as files.

Async & Concurrency ​

  • Use await; never sleep to wait — use expectations/polling.
  • Test cancellation behavior for async APIs.

UI Tests ​

  • Cover only critical journeys; query by accessibility identifier; wait with waitForExistence; stub the backend via launch arguments.

CI ​