“llm-as-a-judge”

Building an Eval Test Suite for ADK Agents

Move from vibes-based testing to structured evaluation for ADK agents, tracking tool call accuracy, trajectory efficiency, and LLM-as-a-judge quality scoring.