THE EVALUATION HARNESS

PRACTICE LAB CONNECTION In the Resume Review Agent lab, which you will build in Chapter 68, you will construct an evaluation harness of your own. That lab agent reads a resume and a job posting and produces a match assessment. You will create a small eval set, run the agent over it, and record scores at the span, trace, session, and outcome levels, using the trace files the lab produces. The prompts below are designed to be tested against that eval set and those granularity levels, so keep your lab trace files and eval set nearby when you try them.