standards-testing_skill

Overview

This skill helps prevent common testing anti-patterns that weaken test confidence and pollute production code. It enforces testing real behavior over asserting mock presence, avoids adding test-only methods to application code, and promotes thoughtful, minimal mocking strategies. Use it to keep tests meaningful, maintainable, and aligned with TDD principles.

How this skill works

The skill inspects test files and mocking patterns to detect assertions that verify mocks instead of actual behavior, identifies methods only referenced by tests, and flags incomplete or overly complex mock setups. It guides decisions on when to use real dependencies versus mocks, suggests extracting test utilities for cleanup, and recommends completing mock data structures to match real responses. It also provides a checklist and decision steps to choose integration tests when mocks become a liability.

When to use it

• When writing or modifying any test file (.test.ts/.spec.ts, _test.py, test_*.py, *_test.go, *_spec.rb, etc.)
• When adding mocks, stubs, spies, or test doubles to a test suite
• When considering adding a method or property to production code that would only be used by tests
• When test setup is long or complex compared to assertions
• When tests fail and you are tempted to change mocks to make them pass
• During code reviews focused on test implementation and mocking strategies

Best practices

• Assert on real behavior, not the presence or calls of mocks
• Keep production code free of test-only methods; create test utilities instead
• Before mocking, list side effects and run tests once with real implementations
• Mock only external/slow operations at the lowest effective level
• Use complete mock data structures that match real API responses
• Prefer TDD: write a failing test first, then implement minimally

Example use cases

• Refactoring a brittle test that asserts mock implementation details instead of outcomes
• Deciding whether to mock a network call or run an integration test against a sandbox
• Removing a cleanup method from a production class and moving it to a test utility
• Fixing a failing duplicate-detection test by preserving side effects in mocks
• Creating realistic mock API responses with all documented fields to avoid silent failures