The scenario runs your helper through a list of questions with known correct answers and scores each one — so you can see where it gets things wrong.