Last released Oct 19, 2026
Agent-as-judge grading framework for evaluating AI agent outputs against rubric criteria