Graphiti MCP Server

Graphiti MCP server enabling AI agents to build and query temporally-aware knowledge graphs — storing entities, relationships, and episodic memories with time-tracking, enabling agents to maintain long-term memory that understands how facts change over time, built on Neo4j and developed by Zep.

Evaluated Mar 06, 2026 (0d ago) vcurrent
Homepage ↗ Repo ↗ Databases graphiti knowledge-graph temporal memory mcp-server neo4j zep
⚙ Agent Friendliness
72
/ 100
Can an agent use this?
🔒 Security
79
/ 100
Is it safe for agents?
⚡ Reliability
70
/ 100
Does it work consistently?

Score Breakdown

⚙ Agent Friendliness

MCP Quality
72
Documentation
75
Error Messages
72
Auth Simplicity
68
Rate Limits
72

🔒 Security

TLS Enforcement
88
Auth Strength
80
Scope Granularity
72
Dep. Hygiene
75
Secret Handling
78

Neo4j + OpenAI credentials. HTTPS. Commercial project. Store secrets securely.

⚡ Reliability

Uptime/SLA
72
Version Stability
68
Breaking Changes
68
Error Recovery
70
AF Security Reliability

Best When

An agent needs sophisticated long-term memory that tracks how facts change over time — particularly for agents operating over long periods where information evolves.

Avoid When

You need simple conversation memory — use lighter alternatives. Or if Neo4j infrastructure isn't available.

Use Cases

  • Building persistent agent memory with temporal awareness from long-running agents
  • Storing and querying entity relationships with time context from knowledge agents
  • Maintaining episodic memory of past conversations from conversational agents
  • Building knowledge graphs from structured and unstructured data from ingestion agents
  • Querying 'what was true at time T' from analytical agents
  • Agent memory systems that understand changing facts over time

Not For

  • Simple key-value or document storage (use simpler memory MCPs)
  • Teams without Neo4j infrastructure
  • Real-time streaming data (Graphiti is for episodic/semantic memory)

Interface

REST API
Yes
GraphQL
No
gRPC
No
MCP Server
Yes
SDK
Yes
Webhooks
No

Authentication

Methods: api_key
OAuth: No Scopes: No

Neo4j credentials required (URI, username, password). OpenAI API key required for embedding generation. Zep Cloud option available with Zep API key.

Pricing

Model: freemium
Free tier: Yes
Requires CC: No

Graphiti is open source. Requires Neo4j and OpenAI (or compatible LLM). Self-hosting is free minus infrastructure costs.

Agent Metadata

Pagination
none
Idempotent
Partial
Retry Guidance
Not documented

Known Gotchas

  • Requires Neo4j running — complex infrastructure dependency
  • OpenAI (or compatible) API key needed for embedding generation — adds cost
  • Temporal graph model has learning curve — agents need context to query effectively
  • From Zep (commercial company) — commercial cloud option available
  • Graphiti schema is opinionated — works best when used as designed
  • Knowledge graph quality depends on LLM extraction quality

Alternatives

Full Evaluation Report

Detailed scoring breakdown, competitive positioning, security analysis, and improvement recommendations for Graphiti MCP Server.

$99

Scores are editorial opinions as of 2026-03-06.

5220
Packages Evaluated
26151
Need Evaluation
173
Need Re-evaluation
Community Powered