Flexible Graphrag

Flexible GraphRAG is an open source Python platform for building retrieval over your own content, in either a knowledge graph mode or a conventional vector mode. Built on LlamaIndex and released under the Apache 2.0 license, it is designed around breadth of choice: rather than committing you to one database vendor, it lets you point the same pipeline at whichever storage layer you already run. The supported set covers eight graph databases including Neo4j, ArcadeDB, FalkorDB, NebulaGraph, Memgraph and Amazon Neptune in both its forms, ten vector databases including Qdrant, Milvus, Weaviate, Chroma and the vector modes of Elasticsearch, OpenSearch and Neo4j, three search engines, and thirteen data sources. Features include automatic knowledge graph building with schemas, document processing through Docling, hybrid search, and an AI query and chat interface. The codebase ships React, Vue and Angular frontends against a FastAPI backend, all running on Windows, Mac and Linux either standalone or in containers, plus a server that exposes the system over the Model Context Protocol. A docker compose setup brings up any chosen combination of graph, vector and search databases together with their dashboards and an Alfresco content repository, selected by uncommenting the includes you want. There is no pricing. The software is free to download and self host from GitHub under Apache 2.0, with no commercial edition, hosted service or paid support plan published. Infrastructure, database and model costs are yours to carry. One thing to weigh before adopting it: the project is published on a personal site by its individual maintainer rather than by a company, so evaluate it as a community effort with no commercial support channel behind it.