MCP Memory Server

MCP Memory server providing persistent memory and knowledge graph capabilities for AI agents — storing and retrieving facts, entities, and relationships across conversations, enabling agents to remember context between sessions, build up knowledge over time, and maintain long-term state without relying solely on in-context window memory.

Evaluated Mar 06, 2026 (0d ago) vcurrent
Homepage ↗ Repo ↗ AI & Machine Learning memory mcp-server knowledge-graph persistence agent-memory context
⚙ Agent Friendliness
77
/ 100
Can an agent use this?
🔒 Security
84
/ 100
Is it safe for agents?
⚡ Reliability
68
/ 100
Does it work consistently?

Score Breakdown

⚙ Agent Friendliness

MCP Quality
68
Documentation
70
Error Messages
68
Auth Simplicity
98
Rate Limits
95

🔒 Security

TLS Enforcement
85
Auth Strength
85
Scope Granularity
78
Dep. Hygiene
75
Secret Handling
92

Local disk. No encryption. No network. Avoid sensitive data. Single-user. Community MCP.

⚡ Reliability

Uptime/SLA
72
Version Stability
65
Breaking Changes
65
Error Recovery
68
AF Security Reliability

Best When

An agent needs persistent memory across sessions — storing entities, relationships, and facts that accumulate over time for long-running assistant or research workflows.

Avoid When

You need high-performance vector similarity search, relational data, or shared memory across many concurrent agents — use specialized database MCPs instead.

Use Cases

  • Persisting agent memory across conversation sessions from long-running agents
  • Building knowledge graphs of entities and relationships from research agents
  • Remembering user preferences and past interactions from personalization agents
  • Storing and retrieving facts learned during agent operations from learning agents
  • Enabling agents to build cumulative knowledge over many sessions from assistant agents
  • Cross-session context preservation for complex multi-step workflows from workflow agents

Not For

  • High-throughput vector search (use dedicated vector DBs like Qdrant or Pinecone)
  • Structured relational data (use database MCPs for SQL data)
  • Real-time shared memory across multiple concurrent agents

Interface

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

Authentication

Methods: none
OAuth: No Scopes: No

No authentication — local memory store persisted to disk. No external service required. Memory files stored locally.

Pricing

Model: free
Free tier: Yes
Requires CC: No

Free, open source local memory MCP.

Agent Metadata

Pagination
none
Idempotent
Full
Retry Guidance
Not documented

Known Gotchas

  • Memory grows unbounded over time — implement periodic pruning for long-lived deployments
  • No built-in encryption for stored memories — sensitive data written to disk in plaintext
  • Concurrent access from multiple agents not designed for — single-agent use intended
  • Memory format may differ from Anthropic's official MCP memory server — check compatibility
  • Search quality depends on implementation — exact vs semantic search varies by version
  • Community MCP — less battle-tested than the official Anthropic memory server

Alternatives

Full Evaluation Report

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

$99

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

5175
Packages Evaluated
26151
Need Evaluation
173
Need Re-evaluation
Community Powered