Check whether AI coding completion claims are supported by the delivery materials and test evidence you provide. Extract verifiable criteria, identify supported and unsupported work, find evidence gaps, and generate a structured delivery verdict—all without accessing repositories or modifying code.