MCP servers · by job

Memory & knowledge MCP servers

330 MCP servers that give an agent long-term memory, notes and a knowledge base. 69 are SAFE, 83 CAUTION and 178 BLOCK. Ranked by trust score; each card opens the full audit.

Ask an AI about this pageGChatGPTCClaudePPerplexityGeGeminiXGrok
More
Cite

Cite this page

ModelsAtlas. “Memory & knowledge MCP servers”. Retrieved 29 Oct 2026. https://modelsatlas.com/mcp/topic/memory/page/8

Verdicts · 330 servers

✓SAFE69!CAUTION83✕BLOCK178

Transports

A server can list more than one transport.

Tool risk

108

servers expose at least one destructive tool: delete, overwrite or run. The median server with a readable tool list exposes 10 tools.

Servers with no destructive tools →

330 memory & knowledge servers

169–192 of 330 · ranked by trust score, then stars

memorycrystal

nativist-rossini671/memorycrystal

✕BLOCK D

Store persistent memory for AI agents so every conversation, decision, and session stays informed

★ 13 destructivestdio

Claudex

Mcalesteryahoo411/Claudex

✕BLOCK D

Connect Claude Code interfaces to Codex and native Claude models using a local bridge.

★ 01 read

kinjot-mcp

zainzaidi37/kinjot-mcp

✕BLOCK D

jotnow - Save useful agent outputs

★ 02 writestdio

Memora

agentic-box/memora

✕BLOCK D

Give your AI agents persistent, collective memory — with deduplicating absorb, supersession lineage, semantic search, and a graph UI. Speaks MCP.

★ 7298 destructivestreamable-http

Compartment

maxfreedompollard/compartment

✕BLOCK D

Encrypted, fully offline agentic memory. One click install, GUI w/ memory map, all OS and agents. Superior memory creation, storage and retrieval.

★ 578Not listed in sourcestdio

AutoMem

verygoodplugins/mcp-automem

✕BLOCK D

MCP client for AutoMem — give Claude, Cursor, Codex, and other MCP tools durable graph+vector memory across conversations.

★ 652 destructivesse

relay

valtors/relay

✕BLOCK D

MCP server with 40 tools for AI agents. Memory, web fetch, search, file ops, screenshots, and multi-agent coordination in one Go binary.

★ 128 writestdio

StashBase

liliu-z/stashbase

!CAUTION D

An open-source, local-first writing workspace where you can write with Claude Code and Codex, drawing on your own sources and past work.

★ 9182 destructivestdio

Hyperweave

inneraura/hyperweave

✕BLOCK D

Visual CLI for agents. Deterministic SVG diagrams, tables, and charts with embedded structured metadata. Zero JavaScript. Renders anywhere markdown does.

★ 481 writestdio

SwarmClaw

swarmclawai/swarmclaw

✕BLOCK D

Open-source self-hosted AI agent runtime and multi-agent framework for autonomous agent swarms. Agent memory, MCP tools, schedules, delegation, and 23+ LLM providers (Claude, GPT, Gemini, OpenRouter, Ollama). A practical Claude Code and LangChain alternative.

★ 6852 destructivestdio

Brain

mordechaipotash/brain-mcp

✕BLOCK D

The recorder for your AI conversations — local-first MCP memory that keeps byte-exact copies of your Claude Code, Codex and Pi sessions (Claude Code deletes its own after 30 days) and answers with a citation or abstains. DuckDB.

★ 785 readstdio

Zettelforge

rolandpg/zettelforge

✕BLOCK D

Agentic memory for CTI in Python — STIX knowledge graphs, threat-actor alias resolution, offline-first RAG, MCP server for Claude Code and LangChain agents

★ 64Not listed in sourcestdio

ZettelForge

threatrecall/zettelforge

✕BLOCK D

Agentic memory for CTI in Python — STIX knowledge graphs, threat-actor alias resolution, offline-first RAG, MCP server for Claude Code and LangChain agents

★ 64Not listed in sourcestdio

vault-cortex

aliasunder/vault-cortex

✕BLOCK D

Standalone MCP server for Obsidian vaults — hybrid search, structured memory, task management, files, OAuth 2.1. Run locally, self-host, or one-click deploy for remote access.

★ 266 readstreamable-http

corpus

ryanjdillon/corpus

✕BLOCK D

Semantic search and structured query over your own content

★ 06 readstreamable-http

Eion

eiondb/eion

✕BLOCK D

Shared Memory Storage for Multi-Agent Systems

★ 1592 destructivestdio

agent-memory

Masikomoore/agent-memory

✕BLOCK D

One memory for every AI agent — self-hosted long-term memory over MCP/REST for Claude Code, Codex, Cursor, Gemini CLI, VS Code, OpenClaw, and more.

★ 33 destructivestreamable-http

MindForge

opok-ops/MindForge

✕BLOCK D

Production-grade lifelong memory for AI agents — 4-tier memory, knowledge graph, E2EE, federated P2P, MCP server.

★ 31 destructive

EMET

dreamforgestudiollc/EMET

✕BLOCK D

EMET is an MCP memory server built for ISO 9001 / ISO 13485 document control: persistent, shared memory for AI agents and bot teams. Nothing is deleted; records are superseded, retired and retained. Controlled templates, verified writes, a word-for-word transcript tier, stdio and remote HTTP with OA

★ 14 writestdio

samyama-graph

samyama-ai/samyama-graph

✕BLOCK D

Rust graph-vector database: OpenCypher (99.9% of evaluated TCK scenarios pass), vector search, graph algorithms, RESP + HTTP. LDBC SNB Interactive and BI complete; 1B edges on one machine.

★ 1762 write

RagRabbit

madarco/ragrabbit

✕BLOCK D

Open Source, Self-Hosted, AI Search and LLM.txt for your website

★ 1362 readstdio

NeuroStack

raphasouthall/neurostack

✕BLOCK D

Local RAG layer and optimizer for your Markdown knowledge base. CLI + MCP server: grounded answers for any AI client, stale-note detection, session harvesting into memories. Local-first.

★ 47Not listed in sourcestdio

noted

fabriziosalmi/noted

✕BLOCK D

Desktop app for taking notes in Markdown/HTML, with built-in AI, note links (wikilinks), global search, local history, and Git integration

★ 13 readsse

omneum

cole-h06/omneum

✕BLOCK D

An open-source context evaluator for AI agents.

★ 04 readstdio

Other jobs

By risk

Questions

How do you decide which servers are memory & knowledge servers?
By the categories each server's own repository declares, mapped onto this job by a hand-kept table. A server can belong to more than one job, and one whose repository declares no matching category is not listed here.
Which memory & knowledge MCP servers rank highest on trust?
Graphiti (getzep/graphiti), Memory Bank (alioshr/memory-bank-mcp) and Knowledge Graph Memory (shaneholloman/mcp-knowledge-graph), by trust score and then stars. Ranking never uses popularity to change a verdict.
How many memory & knowledge servers only read?
88 of the 330 declare tools that all only read; 108 expose at least one destructive tool.

Keep going in your AI

Opens with this page as the source. We add nothing else to the prompt.