Practical overview
When it fits
Use it when a .NET application needs the documented PostgreSQL/pgvector or OpenSearch vector-store implementation, or one of its listed chat-history backends.
Who it is for
- .NET developers adding vector search over PostgreSQL with pgvector or OpenSearch k-NN.
- Developers who need in-memory, file-based, Redis or MongoDB chat-message history storage.
Documented uses
- Define keyed records with text and embedding fields for a vector store.
- Upsert embedded records and search them with a query embedding.
- Store chat-message history using one of the provided in-memory, file-based, Redis or MongoDB packages.
Requirements & limits
- Vector-store implementations are provided for PostgreSQL with pgvector and OpenSearch k-NN.
- Chat-history packages cover in-memory and file-based storage, Redis, and MongoDB.
- The README recommends using official Semantic Kernel connector packages directly for backends with those connectors, including SQLite, Chroma and Qdrant.
- The library can be used independently of the main LangChain.NET project.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Set up a vector collection and upsert an embedded record
Create a collection through a documented vector-store backend and add a record containing an embedding.
Before you start
- A .NET/C# application using Microsoft.Extensions.VectorData.
- A PostgreSQL/pgvector or OpenSearch backend.
- A record type with vector-store key, data and vector fields.
- A connection string for PostgreSQL, or an OpenSearch connection URI.
- An embedding value to store.
Steps
Define a record type with VectorStoreKey, VectorStoreData and VectorStoreVector metadata.
Create a PostgresVectorStore with a connection string, or an OpenSearchVectorStore with a ConnectionUri.
Get a collection for the record type and call EnsureCollectionExistsAsync.
Call UpsertAsync with a record containing its text and embedding.
Expected outcome
A collection is ensured and an embedded record is submitted to it through the selected backend.
Guide limitations
- The documented steps do not cover generating embeddings.
- The guide does not include backend installation or configuration steps.
Search a vector collection with a query embedding
Retrieve records from a configured collection using a query vector.
Before you start
- A collection created through a documented backend.
- Records with embeddings already upserted.
- A query embedding represented as ReadOnlyMemory.
Steps
Call SearchAsync on the collection with the query embedding.
Set the top argument to the desired result count, as in the README example.
Enumerate the returned results and read each record and score.
Expected outcome
The search returns records with scores for the supplied query embedding.
Guide limitations
- The documented example does not explain embedding generation or score interpretation.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 13 October 2026
- Original website title
- tryAGI/LangChain.Databases · Official homepage title · retained in its source language