RAGLight is a Python framework for building RAG pipelines that combine document retrieval with LLM inference. It provides modular components allowing users to swap between different LLM providers (OpenAI, Mistral, Ollama, LM Studio, HuggingFace), embedding models, and vector stores. The framework includes a CLI for chatting with documents, a REST API server with a chat UI, and MCP integration for connecting external tools and data sources to the retrieval pipeline.