Feature request: Add Hologres Vector Store node

Feature Requests
1

Node subcategory: Vector Store (AI / LangChain)

The idea is:

Add a new Hologres Vector Store node to @n8n/nodes-langchain that integrates with Alibaba Cloud Hologres — a cloud-native data warehouse with built-in high-performance vector search via its HGraph index (approximate nearest neighbor search).
The node would:

  • Connect to Hologres using dedicated credentials (host, port, database, user, password)
  • Support all standard vector store operation modes: insert, load, retrieve, and retrieve-as-tool
  • Automatically create and manage tables using Hologres-native float4[] vector columns with dimension CHECK constraints
  • Build and configure the HGraph vector index via ALTER TABLE SET (vectors = ...), with support for:
    • Distance methods: Cosine, InnerProduct, Euclidean
    • Base quantization type (e.g. rabitq) with optional high-precision reorder pass
    • Configurable max_degree and ef_construction graph construction parameters
  • Support metadata filtering during similarity search

My use case:

I am building AI Agent and RAG pipelines on Alibaba Cloud infrastructure. My data already lives in Hologres, and I want to use it as a vector store directly in n8n workflows without having to sync data to a separate vector database. Hologres is widely used in the Alibaba Cloud ecosystem, and many enterprise users in this ecosystem would benefit from a native n8n integration.

I think it would be beneficial to add this because:

  • No viable workaround exists today. Hologres is PostgreSQL-wire-compatible, but the existing PGVector node uses pgvector extension syntax (vector type, <-> distance operators) which is fundamentally incompatible with Hologres’s float4[] column type and HGraph index approach. PGVector simply does not work against a Hologres endpoint.

  • Growing user base. Hologres is a popular managed service for Alibaba Cloud users. Adding this node lowers the barrier for this community to adopt n8n for AI workflows.

  • Consistent experience. The implementation follows the exact same createVectorStoreNode factory pattern used by all other vector store nodes (Pinecone, Qdrant, Milvus, etc.), keeping the codebase consistent and the user experience familiar.

  • Hologres product overview: Introduction to Hologres - Alibaba Cloud Document Center

  • Hologres vector search (HGraph index): 404 阿里云错误页

  • Hologres PostgreSQL compatibility: 404 阿里云错误页

Are you willing to work on this?

Yes — I have already implemented this node and have a working PR ready to submit.