Neo4j Memory System FAQs

Question 1

How is the Neo4j Memory System integrated and used with Claude?

Accepted Answer

It integrates via an MCP (Model Context Protocol) connector. Users add specific system instructions to Claude (either for all chats or per project) that guide Claude to call dedicated memory tools like `memory_get_context` at the start of conversations and `memory_store_message` and enrichment tools during interactions.

Question 2

How does this system enhance Claude's capabilities?

Accepted Answer

It allows Claude to remember past conversations, specific entities (people, projects, tools), user preferences, and learned facts, eliminating the need for manual note-taking. This results in an AI assistant that provides more relevant and continuous interactions by recalling prior context in real-time.

Question 3

What types of data does the Neo4j Memory System store?

Accepted Answer

It stores various types of data modeled as a property graph, including conversation sessions, individual messages, named entities (PERSON, OBJECT, LOCATION, EVENT, ORGANIZATION), user preferences, factual triples (subject-predicate-object), and detailed reasoning traces for complex tasks.

Question 4

What is the Neo4j Memory System?

Accepted Answer

The Neo4j Memory System (neo4j-mcp-memory) is a connector that provides Claude AI with a durable, graph-backed memory store. It uses Neo4j to persist conversational context, entities, user preferences, facts, and reasoning traces across sessions.

Question 5

Can Claude update its memory in real-time during a conversation?

Accepted Answer

Yes, during a conversation, Claude can enrich the memory graph in real-time. It uses tools like `memory_add_entity`, `memory_create_relationship`, `memory_add_fact`, and `memory_add_preference` to store new information as it's learned, ensuring its memory is always current and relevant.

Neo4j Memory System

Neo4j Memory System

主な機能

ユースケース

主な機能

ユースケース