An all-in-one multimodal RAG framework that parses documents, builds knowledge graphs, and enables hybrid retrieval over text and images.