Continuous Process-Level Evaluation for Evolving Enterprise AI Agent Skills.

arXiv:2610.01833v1 Announce Type: new Abstract: Enterprise AI agent skills evolve as tool APIs, models, and specifications change, yet final-output evaluation can miss process-level behavioral drift. We present a continuous evaluation framework combining outco