thermiteau/mav-bp-integration-testing

Integration Testing Standards

Ensure integration tests verify that components work together correctly across real boundaries — APIs, databases, file systems, and third-party services.

Principles

1. Test real interactions — integration tests exercise actual wiring between components, not mocked substitutes
2. Isolate test runs — each test controls its own data and state; parallel runs never collide
3. Repeatable environments — tests produce the same result regardless of when or where they run
4. Acceptable speed — integration tests are slower than unit tests but must not be wastefully slow; keep each test under a few seconds where possible
5. Complement, don't duplicate — integration tests cover cross-boundary behaviour that unit tests cannot; don't re-test pure logic already covered by unit tests

Test Scope

What to Test

• API contracts — HTTP endpoints return correct status codes, headers, and response shapes for valid and invalid inputs
• Database interactions — queries, transactions, migrations, and constraints behave correctly against a real (or containerised) database
• Service-to-service calls — upstream/downstream services communicate correctly over real protocols (HTTP, gRPC, message queues)
• File system operations — reads, writes, and path handling work in realistic directory structures
• Authentication and authorisation flows — tokens, sessions, and permission checks work end-to-end
• Configuration loading — the application starts and behaves correctly with real configuration files and environment variables

What NOT to Test

• Pure business logic — already covered by unit tests
• Third-party library internals — trust the library; test your usage of it
• UI rendering — that belongs in end-to-end or visual regression tests
• Performance benchmarks — use dedicated performance tests, not integration tests

External Dependencies

Strategies (prefer in order)

| Strategy | When to use | Example | | -------- | ----------- | ------- | | Containerised real service | Database, cache, message broker | Testcontainers, Docker Compose | | Local server | Lightweight service your code controls | In-process HTTP server, embedded DB | | Sandbox / test account | Third-party API with sandbox mode | Stripe test mode, AWS LocalStack | | Contract test + stub | External API you cannot control | Pact, WireMock with recorded responses |

Rules

• Never hit production services — integration tests use sandboxes, containers, or stubs
• Pin dependency versions — lock the image tag or service version to avoid surprise breakages
• Start dependencies before tests — use setup scripts, Docker Compose, or framework hooks to guarantee services are ready
• Health-check before running — wait for dependencies to be healthy, don't rely on sleep timers

Test Data & Isolation

Data Rules

• Each test owns its data — create what you need in setup, clean up in teardown
• Use unique identifiers — prefix or randomise names/IDs so parallel tests never collide
• Never depend on seed data — tests must work against a clean environment
• Transaction rollback where possible — wrap each test in a transaction that rolls back, avoiding manual cleanup

Environment Isolation

• Separate test database — never share the development or production database
• Isolated network — containerised tests run in their own network namespace
• Clean state between suites — reset shared resources (queues, caches) between test suites if rollback is not possible

Test Structure

Setup and Teardown

1. Suite setup — start containers, run migrations, seed minimal reference data
2. Test setup — create test-specific data
3. Act — execute the integration scenario
4. Assert — verify outcomes across all affected systems (response + side effects)
5. Test teardown — remove test-specific data
6. Suite teardown — stop containers, drop test database

Naming Convention

• Test names describe the integration scenario: should persist order and publish event when checkout completes
• Group by integration boundary using describe blocks: describe('POST /orders')

Timeouts

• Set explicit timeouts per test — integration tests may wait for external systems
• Fail fast on dependency unavailability rather than hanging until a global timeout

Test Organisation

• Keep integration tests in a dedicated directory (e.g., tests/integration/, __tests__/integration/)
• Separate from unit tests so they can run independently with different commands
• Shared fixtures and helpers in a dedicated test utilities directory
• Configuration for test environments in a dedicated config file or environment variables

Project Implementation Lookup

Before applying these standards, load the project-specific testing implementation:

digraph lookup {
    "docs/maverick/skills/integration-testing/SKILL.md exists?" [shape=diamond];
    "Read and use alongside these standards" [shape=box];
    "Invoke upskill" [shape=box];
    "Read generated skill" [shape=box];

    "docs/maverick/skills/integration-testing/SKILL.md exists?" -> "Read and use alongside these standards" [label="yes"];
    "docs/maverick/skills/integration-testing/SKILL.md exists?" -> "Invoke upskill" [label="no"];
    "Invoke upskill" -> "Read generated skill";
    "Read generated skill" -> "Read and use alongside these standards";
}

1. Check for docs/maverick/skills/integration-testing/SKILL.md
2. If missing, invoke the upskill skill with:
   • topic: integration-testing
   • scan hints:
     • dependencies: supertest, testcontainers, docker-compose, pytest, httpx, rest-assured, wiremock, pact, localstack
     • grep: describe\(|it\(|test\(|expect\(|assert|@SpringBootTest|@IntegrationTest|TestContainers
     • files: **/integration/**, **/e2e/**, **/*.integration.*, **/docker-compose*.test*
3. Read the project skill and apply these best practices in the context of the project's specific technology

Detecting Integration Testing Issues in Code Review

| Pattern | Issue | Fix | | ------- | ----- | --- | | Test hits production service | Risk of side effects and flakiness | Use container, sandbox, or stub | | No cleanup after test | Data leaks between tests | Add teardown or use transaction rollback | | Sleep-based waits for dependencies | Flaky timing | Use health-check polling with timeout | | Test duplicates unit test coverage | Wasted execution time | Delete and rely on unit test | | Shared mutable test data across tests | Order-dependent failures | Each test creates its own data | | No timeout set | Hanging test blocks CI | Set explicit per-test timeout | | Hard-coded connection strings | Breaks in different environments | Use environment variables or config | | Test requires manual environment setup | Not reproducible | Automate with containers or scripts |