Applied AI·Evaluation and reliability
you keep a fixed set of inputs with the answers you already know are right, and every change gets scored against that same set.
Test case / golden set
Also calledgolden set
Draft summary, pending review
One case is an input plus its expected outcome; the golden set is your curated collection of them, weighted toward real and hard examples. Its quality bounds the meaning of every number the harness prints. Start at 20 to 50 cases and grow it from production failures.