ClickHouse RAG

ClickHouse RAG is a Python framework for finding and citing evidence in ClickHouse. Index documents with embeddings and optional relationship facts. Run hybrid search, follow relationships across documents, or execute a typed knowledge plan. The same namespace and metadata filters scope each operation.

Use the quickstart to run hybrid retrieval or the framework quickstart for GraphRAG, KAG and local models. If you have an existing 0.x installation, see the migration guide. The operations guide covers access, consistency and sizing.