Adding Tests

Core Principle

One test is usually enough. Only add multiple tests if the bug manifests differently under multiple conditions.

Test behavior, not implementation. Validate outcomes, not code paths.

A test's value is measured by:

1. Does it fail when user-facing behavior breaks?
2. Does it stay green when internals are refactored?
3. Does it catch the real issue it was written for?

Routing

Identify your context, then read the corresponding file: