# BAI-LAB/MemoryOS

[EMNLP 2025 Oral] MemoryOS is designed to provide a memory operating system for personalized AI agents.

Repository: https://github.com/BAI-LAB/MemoryOS
Canonical: https://ross.abutalabs.com/products/memoryos
Homepage: https://baijia.online/memoryos/
Language: Python
License: Apache-2.0
License Family: permissive
Topics: agent, language-model, llm, long-term-memory, operating-system, personalization, rag, retrieval-augmented-generation
Last push: 2026-07-07T12:32:18+00:00

## Health v2 (maintenance only)
Score: 61/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 91, release rhythm 40, longevity 32
- inputs: {"age_days": 460, "days_push": 57, "days_rel": 411, "gap_med": 2.5, "n_releases_24m": 3}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1561, forks 160 (observed 2026-08-28T04:05:03.848542+00:00)

## What it is
MemoryOS is a Python library that provides an operating-system-inspired memory layer for personalized AI agents, using a hierarchical short/mid/long-term storage architecture with storage, updating, retrieval, and generation modules. It ships as a plug-and-play SDK plus an MCP server so any LLM-backed application can gain persistent long-term memory.

## Use cases
- add long-term memory to my ai agent
- make chatbot remember user preferences across sessions
- store user profile and knowledge base for personalized llm responses
- plug memory into claude desktop or cursor via mcp
- improve multi-turn conversation coherence with llm
- keep chat history local without sending data to cloud
- benchmark agent memory on locomo

## When to choose
- you need persistent, personalized memory for LLM agents with SOTA benchmark results
- you want an MCP server to inject memory into existing AI clients like Claude Desktop or Cursor
- you need pluggable storage, update, and retrieval strategies and local data storage for privacy

## When to avoid
- you only need simple short-context conversation without cross-session memory
- you need a fully managed cloud memory service rather than a self-hosted Python library
- your stack is not Python or LLM-agnostic tooling is required

## Facets
- artifact type: library
- maturity: active
- function: rag, agent-framework, llm-inference, mcp, chatbot, search-engine
- domain: large-language-models, developer-tools
- platform: python, cross-platform, self-hosted
- tags: long-term-memory, memory-management, personalization, mcp-server, hierarchical-storage, emnlp-2025, ai-agents, retrieval-augmented-generation, natural-language-processing

## Member repositories
- BAI-LAB/MemoryOS (main) score 61

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:03.848542+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:29:42.864678+00:00, confidence not recorded.
  - readme: https://github.com/BAI-LAB/MemoryOS (fetched 2026-08-28T04:05:03.848542+00:00, sha 29c5e335b010)
  - homepage: https://baijia.online/memoryos/ (fetched 2026-08-29T11:29:27.501198+00:00, sha 6a842464109d)
- Data as of 2026-08-30T08:39:29.467469+00:00.