What problem does it solve? Tests can be written first, go red, then green, and still be bad tests — coupling to implementation details, computing expected values the same way the code does, or asserting that artifacts merely exist. This Skill defines what a test should actually assert so tests survive refactoring and genuinely prove behavior. ## Core Features & Use Cases - Behavior-focused assertions: Assert observable behavior through public interfaces instead of internal call counts, private helpers, or file existence. - Independent expected values: Source expected values from spec scenarios, hand-computed examples, or known-good results rather than replicating implementation logic. - Harness-free RED→GREEN: Use a build plus a written-down manual check as the red-green signal in projects with no test runner, and produce evidence artifacts instead of unverifiable claims. - Use Case: When filling a plan's Test-first block or diagnosing a test that broke after a pure refactoring, consult this Skill to determine whether the test was asserting the wrong thing. ## Quick Start Ask the AI to review a test you are about to write against tdd-principles to confirm it asserts observable behavior with independently sourced expected values.
