Add a test which fails. Run the tests and see the new one fail. Add enough code to get it to pass. Run the tests and watch them pass. Refactor the code, ensuring tests continue to pass. Repeat.