This Week in Neo4j: NODES, MCP, AI Agents, Access Control and more

FEATURED COMMUNITY MEMBER: Annabelle Blangero

NODES: Agenda now live

NODES is back on November 6 – our global, 24-hour graph dev conference spotlighting real-world apps, intelligent systems and all things Neo4j. The agenda is now published, so you can check out what we have in store for you this year!

MCP: Implementing Neo4j GraphRAG Retrievers as MCP Server

Tomaz Bratanic shows how to wrap Neo4j’s VectorCypherRetriever into an MCP (Model Context Protocol) server. This lets LLM agents invoke semantic + graph-based search via a single tool interface without handling raw embeddings or Cypher syntax themselves.

AI Agents: Creating a Neo4j Agentic Memory Multi-User Dungeon

In this post, Christoffer Bergman builds a Multi-User Dungeon (MUD) where Neo4j serves as the game master’s memory -persisting rooms, NPCs, player state, and relationships so the world doesn’t “forget” when you backtrack or re-encounter characters.

Access Control: Multi Tenancy and Fine Grained Access Control For SAAS Applications

In this article, Aniket Roy explores how multi-tenant SaaS apps can move beyond coarse role-based access and adopt fine-grained, relationship-aware permissions – ensuring data isolation, per-tenant resource ownership, and least-privilege enforcement. He explains how modelling users, tenants, roles, and resources as graphs can simplify complex access control logic and enforce it dynamically using policies rather than hardcoded checks.

POST OF THE WEEK: Tuana Çelik

Please share it if you like it!