Repolode
Submit
MemTensor

MemTensor/MemOS

Memory operating system for LLMs and AI agents that stores, retrieves, and manages long-term memory as an editable graph.

Open-source alternative to:Mem0

View on GitHubWebsite
Stars
11,959
Forks
1,300
Open issues
149
Last commit
6 hours ago

What is it?

MemOS is a memory layer for LLM applications and AI agents. It provides one API to add, retrieve, edit, and delete memories, which are organized as a graph so they can be inspected and corrected rather than sitting in an opaque embedding store. It handles text, images, tool traces, and personas in the same memory system, and groups multiple knowledge bases into composable memory cubes that can be isolated or shared across users, projects, and agents.

The project offers several ways to run it: a hosted MemOS Cloud API, a self-hosted REST service started with Docker Compose (which pulls in Neo4j and Qdrant), and local plugins for OpenClaw, Hermes Agent, and DeepSeek Harness that store memory in on-device SQLite. The local plugins use hybrid retrieval combining FTS5 full-text and vector search, plus skill evolution across traces, policies, and world models.

Key features

  • 01Unified API to add, retrieve, edit, and delete memories
  • 02Graph-structured memory that can be inspected and corrected
  • 03Multi-modal memory covering text, images, tool traces, and personas
  • 04Multi-cube knowledge base management with isolation and sharing
  • 05Asynchronous memory ingestion through MemScheduler
  • 06Local plugins for OpenClaw, Hermes Agent, and DeepSeek Harness with on-device SQLite storage

Use cases

  • Building AI assistants that keep consistent, context-rich conversations across sessions
  • Customer support agents that recall past tickets and user history
  • Personalized agents that adapt to individual user preferences
  • Multi-agent systems that share or isolate memory between agents

Quick start

  1. 1

    Clone the MemOS repository

    git clone https://github.com/MemTensor/MemOS.git
  2. 2

    Copy the example env file, then fill in your API keys in .env

    cp docker/.env.example .env
  3. 3

    Go to the docker folder and start the MemOS API with Neo4j and Qdrant (API served at localhost:8000)

    docker compose up

Things to know

  • Self-hosting needs Neo4j and Qdrant running alongside the API.
  • The hosted Cloud API requires a paid-or-free MemOS API key from its dashboard and sends your data to MemOS Cloud.

Topics

#ai-agents#llm#long-term-memory#rag#memory-management#mcp#vector-search#agent#agentic-ai#ai

FAQ

What is MemOS?
Memory operating system for LLMs and AI agents that stores, retrieves, and manages long-term memory as an editable graph. MemOS is a memory layer for LLM applications and AI agents. It provides one API to add, retrieve, edit, and delete memories, which are organized as a graph so they can be inspected and corrected rather than sitting in an opaque embedding store. It handles text, images, tool traces, and personas in the same memory system, and groups multiple knowledge bases into composable memory cubes that can be isolated or shared across users, projects, and agents.
Is MemOS free to use? What license does it have?
MemOS is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is MemOS still maintained?
Actively maintained. The last commit to MemOS was on Oct 20, 2026.
What can I use MemOS for?
Building AI assistants that keep consistent, context-rich conversations across sessions Customer support agents that recall past tickets and user history Personalized agents that adapt to individual user preferences Multi-agent systems that share or isolate memory between agents
How do I install MemOS?
Clone the MemOS repository: git clone https://github.com/MemTensor/MemOS.git Copy the example env file, then fill in your API keys in .env: cp docker/.env.example .env Go to the docker folder and start the MemOS API with Neo4j and Qdrant (API served at localhost:8000): docker compose up
Where can I get MemOS?
The source code and installation instructions are on GitHub: https://github.com/MemTensor/MemOS

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects