testing-patterns - SKILL.md Agent Skill

name: testing-patterns description: TDD, mocking, and test structure patterns. Use when writing tests, implementing TDD, creating mocks, or discussing test coverage and testing strategies. allowed-tools: Read, Bash

Testing Patterns

Universal testing patterns that apply across languages and frameworks.

Test-Driven Development (TDD)

The TDD Cycle

Red → Green → Refactor

Red: Write a failing test
Green: Write minimal code to pass
Refactor: Improve code, keep tests passing

TDD Example (Pseudocode)

# Step 1: Red - Write failing test
test "add returns sum of two numbers":
    assert add(2, 3) == 5  # Fails - function doesn't exist

# Step 2: Green - Minimal implementation
function add(a, b):
    return 5  # Passes, but not generalized

# Step 3: Refactor - Proper implementation
function add(a, b):
    return a + b  # Correct implementation

Test Structure

Arrange-Act-Assert (AAA)

test "user can log in with valid credentials":
    # Arrange - Set up test conditions
    user = create_test_user(email="test@example.com", password="secret")

    # Act - Perform the action
    result = login(email="test@example.com", password="secret")

    # Assert - Verify the outcome
    assert result.success == true
    assert result.user.email == "test@example.com"

Given-When-Then (BDD Style)

test "authenticated user can access dashboard":
    # Given a logged-in user
    user = login_as_test_user()

    # When they access the dashboard
    response = get_dashboard(user)

    # Then they see their data
    assert response.status == 200
    assert response.contains(user.name)

Mocking Patterns

When to Mock

Mock when:

External services (APIs, databases)
Slow operations
Non-deterministic behavior (time, random)
Side effects (emails, payments)

Don't mock:

The thing you're testing
Simple value objects
When a real implementation is easy

Mock Types

Stub

Returns predetermined values:

stub_api.get_user = returns({ id: 1, name: "Test" })

Spy

Records calls for verification:

spy_logger = spy(logger)
do_something()
assert spy_logger.called_with("Something happened")

Fake

Working implementation with shortcuts:

fake_db = InMemoryDatabase()  # Real behavior, no persistence

Test Categories

Unit Tests

Test one function/method in isolation
Fast (milliseconds)
Mock dependencies

Integration Tests

Test components working together
Use real dependencies where practical
Slower (seconds)

End-to-End Tests

Test full user flows
Real browser/client
Slowest (minutes)

Testing Pyramid

         /\
        /  \  E2E (few)
       /────\
      /      \  Integration (some)
     /────────\
    /          \  Unit (many)
   /────────────\

Quick Checklist

Before completing a feature:

Happy path tested
Error cases tested
Edge cases tested
All tests pass
No flaky tests introduced