Evaluation & quality

Measuring whether an agent actually works — and stays working.

← All concepts

• ADK Eval Starter

  Public

  A starter kit for evaluating ADK agents — metrics, evalsets, and trajectory scoring.

  • Evaluation & quality
  • Python
  • ADK
  Code → ★ 7

• Customer Service Eval

  Public

  Evaluating a customer-service agent against realistic conversations.

  • Evaluation & quality
  • Python
  • ADK
  Code →