## RAGVUE: A New Tool for Evaluating RAG Systems Evaluating Retrieval-Augmented Generation (RAG) systems remains a challenging task. Existing metrics often provide aggregated scores that do not offer a clear view of the causes of errors. To address this challenge, RAGVUE, a diagnostic framework for automated and explainable evaluation of RAG pipelines, has been developed. ## Key Features of RAGVUE RAGVUE decomposes the behavior of RAG systems into several key components: * Retrieval quality * Relevance and completeness of answers * Accuracy of claims * Model calibration Each metric includes a structured explanation, making the evaluation process more transparent. The framework supports both manual metric selection and fully automated agentic evaluation. RAGVUE provides a Python API, a command-line interface (CLI), and a local Streamlit interface for interactive usage. ## Integration and Availability Experimental results demonstrate that RAGVUE identifies failures that other tools often overlook. The source code and detailed instructions for use are available on GitHub, facilitating the integration of RAGVUE into research projects and the practical development of RAG systems. RAG systems are increasingly popular in various sectors, as they allow combining the power of large language models (LLMs) with external information retrieved in real time. This approach overcomes the limitations of pre-trained models, providing more accurate and contextualized answers. The ability to accurately evaluate these systems is therefore crucial to ensure their reliability and effectiveness.