Summary

This issue tracks the discussion of a new integration library that enables seamless use of Microsoft.Extensions.VectorData with the ArangoDB .NET Standard client. The goal is to provide idiomatic, efficient, and extensible vector search capabilities backed by ArangoDB’s native support for graph and document modeling.

Features

• ✅ Plug-and-play IVectorStore, IVectorSearchable and IKeywordHybridSearchable implementation for ArangoDB
• 🧠 Supports vector similarity search via ArangoSearch and custom AQL
• 📁 Supports collection manipulation via VectorStoreCollection
• 🔍 Enables hybrid search (Vector Search + Full Text Search)

Motivation

ArangoDB’s multi-model architecture makes it a compelling backend for agentic workflows, especially when combining vector search with graph traversal. This library bridges the gap between ArangoDB and the .NET VectorData ecosystem, enabling:

• Native vector search with ArangoSearch
• Full-text + vector hybrid queries
• Future integration with agent memory systems

Roadmap

• 🔀 Hybrid Search: Combine vector similarity with graph traversal for contextual recall
• 🧠 Microsoft.KernelMemory integration: Implement IMemoryDb for ArangoDB to give agents persistent, graph-native memory

I have been working on this library for last couple of days and almost done with my works.
Do you have a plan to integrate the Microsoft.Extensions.VectorData for ArangoDB library?