Back to Arena

MCP Memory Server

by Anthropic / Model Context Protocol

System Card

OrganizationAnthropic / Model Context Protocol
Released2024-11
Architectureknowledge-base / KG-based persistent memory via MCP
DetailsReference MCP server for persistent knowledge-graph memory. One of seven canonical MCP reference servers; entities + relations + observations exposed over the MCP protocol to any compatible client.
Parameters—
Domainagent-memoryknowledge-graph
Open SourceYes
WebsiteVisit
CodeRepository
mcpreference-serverknowledge-graphanthropic

Capability Profile

Benchmark Scores

6 of 14 benchmarks
Long-Context Retrieval
0/5
RULER
no data
NIAH
no data
LooGLE
no data
LongBench
no data
∞Bench
no data
Multi-Turn Recall
1/2
LoCoMo
75.349p
MemoryBank
no data
Cross-Session Memory
1/1
LongMemEval
73.132p
Multi-Hop QA
2/3
BABILong
no data
MultiHop-RAG
7579p
HotpotQA
80.297p
Agent Task Memory
1/1
AgentBench-Mem
7226p
Personalization
0/1
PerLTQA
no data
Factuality / Grounding
1/1
RAGAS
73.282p
Sources:MCP Memory Server (modelcontextprotocol/servers); evaluated on HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering (Stanford / CMU, 1809)MCP Memory Server (modelcontextprotocol/servers); evaluated on LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive Memory (Salesforce AI Research, 2410)MCP Memory Server (modelcontextprotocol/servers); evaluated on MultiHop-RAG: Benchmarking Retrieval-Augmented Generation for Multi-Hop Queries (HKUST, 2401)MCP Memory Server (modelcontextprotocol/servers); evaluated on AgentBench Memory Track (Tsinghua KEG, 2308)MCP Memory Server (modelcontextprotocol/servers); evaluated on LoCoMo: Long-Term Conversational Memory Benchmark (Snap Research, 2402)MCP Memory Server (modelcontextprotocol/servers); evaluated on RAGAS: Automated Evaluation of Retrieval-Augmented Generation (Exploding Gradients, 2309)