# SuanmoSuanyangTechnology/MemoryBear

MemoryBear Equip AI with human-like memory capability

Repository: https://github.com/SuanmoSuanyangTechnology/MemoryBear
Canonical: https://ross.abutalabs.com/products/memorybear
Homepage: https://www.memorybear.ai/
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2026-08-26T12:52:25+00:00

## Health v2 (maintenance only)
Score: 84/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 99, release rhythm 98, longevity 24
- inputs: {"age_days": 348, "days_push": 7, "days_rel": 12, "gap_med": 8, "n_releases_24m": 30}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 5088, forks 396 (observed 2026-08-28T04:09:09.446196+00:00)

## What it is
MemoryBear is an AI memory management system that gives LLM agents and chatbots human-like long-term memory across a full lifecycle of perception, extraction, association, and biologically-inspired forgetting. Built in Python with FastAPI and Neo4j, it stores memories as dynamic knowledge graphs, compresses chat history to reduce token usage, and exposes an API that integrates with any agent or workflow.

## Use cases
- give my chatbot long-term memory across conversations
- stop my LLM agent forgetting user preferences
- store and retrieve personalized user context for AI assistants
- compress conversation history to save tokens
- share memory between multiple agents and workflows
- build a knowledge graph of user interactions
- add memory with forgetting to an AI companion app

## When to choose
- you need persistent, evolving memory for LLM agents beyond the context window
- you want graph-based associative memory with consolidation and forgetting rather than plain vector lookup
- you need cross-agent memory sharing and token-efficient history compression
- you prefer a self-hostable Python service with an HTTP API

## When to avoid
- you only need simple key-value session storage or a plain vector store
- your stack cannot run Python 3.12+ and Neo4j
- you need a battle-tested incumbent like Mem0 or Zep with a large ecosystem
- you require fully offline operation without LLM calls for memory extraction

## Facets
- artifact type: service
- maturity: active
- function: vector-database, rag, agent-framework, llm-inference, search-engine, api-framework
- domain: artificial-intelligence, large-language-models, chatbots, databases
- platform: python, self-hosted, cross-platform
- tags: ai-memory, long-term-memory, memory-graph, knowledge-graph, neo4j, fastapi, forgetting-mechanism, personalization, agent-memory, ai-agents, retrieval-augmented-generation, natural-language-processing, docker

## Member repositories
- SuanmoSuanyangTechnology/MemoryBear (main) score 84

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:09.446196+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 2026-08-29T18:02:32.471183+00:00, confidence not recorded.
  - readme: https://github.com/SuanmoSuanyangTechnology/MemoryBear (fetched 2026-08-28T04:09:09.446196+00:00, sha e73c4c37c854)
  - homepage: https://www.memorybear.ai/ (fetched 2026-08-29T08:57:19.514308+00:00, sha e63ca6f275ac)
  - site_page: https://www.memorybear.ai/about (fetched 2026-08-29T08:57:19.518916+00:00, sha 89e47b294bb7)
  - site_page: https://www.memorybear.ai/pricing (fetched 2026-08-29T08:57:19.516978+00:00, sha ddbe38bedb80)
- Data as of 2026-08-30T08:39:29.467469+00:00.