Open-source framework for evaluating Retrieval-Augmented Generation pipelines. Measures faithfulness, answer relevancy, context precision, and hallucination with automated LLM-as-judge metrics.