Verified

Graphiti MCP Server

Temporally-aware knowledge graph queries for AI agents in dynamic environments.

Data last scanned last week · Reviewed 2 months ago

Overview

Graphiti MCP Server bridges the gap between AI assistants and temporally-aware knowledge graphs. It allows agents to manage entities, relationships, and episodes within a Graphiti-structured database, enabling context-aware data retrieval that persists over time through incremental updates rather than full recomputations. Designed for interactive applications, the server provides tools for semantic and hybrid searching of facts and nodes. Users can organize data into specific groups, manage nodes and relationships directly, and maintain the underlying indices to ensure efficient graph performance.

Our verdict

With a security score of 10/100, this server is high-risk and not recommended for critical production environments. The repository has not been updated in over 406 days and lacks a defined license, leaving users with unclear legal terms for usage. Furthermore, it is a community-maintained project that has not undergone editorial review, necessitating extreme caution for those deploying it in sensitive data contexts.

Supports incremental graph data updates
Enables both semantic and hybrid search capabilities
Zero direct code dependencies

No detected license for use
Last commit over 406 days ago
High-risk score of 10/100
Not official or vendor-maintained

Tools

Tool	Description	Risk
add_episode	Add an episode to the knowledge graph (supports text, JSON, and message formats)	medium
search_nodes	Search the knowledge graph for relevant node summaries	low
search_facts	Search the knowledge graph for relevant facts (edges between entities)	low
delete_entity_edge	Delete an entity edge from the knowledge graph	high
delete_episode	Delete an episode from the knowledge graph	high
get_entity_edge	Get an entity edge by its UUID	low
get_episodes	Get the most recent episodes for a specific group	low
clear_graph	Clear all data from the knowledge graph and rebuild indices	high
get_status	Get the status of the Graphiti MCP server and Neo4j connection	low

Compatibility

Client	Local	Docker	Remote	Read-only
Claude Desktop
Cursor
VS Code
Windsurf
ChatGPT

Frequently asked questions

›What are the core technical prerequisites for running this server?

You must have Python 3.10 or higher, an active Neo4j database (version 5.26 or later), and a valid OpenAI API key.

›How do I configure the server's database and LLM settings?

You can configure settings such as NEO4J_URI, NEO4J_PASSWORD, and OPENAI_API_KEY by creating a .env file in the project directory or by setting them as environment variables.

›Can I run the Graphiti MCP server using Docker?

Yes, the server supports deployment via Docker Compose, which includes a pre-configured Neo4j container.

›Does the server support different transport methods?

Yes, you can specify the transport method as either stdio or sse using the --transport flag when running the server.

›How can I override configuration values at runtime?

You can pass command-line arguments like --model, --small-model, and --temperature to override the corresponding environment variables.

›What is the purpose of the group-id flag?

The --group-id flag allows you to set a namespace for the knowledge graph, with the server defaulting to 'default' if no ID is provided.