NevaMind-AI/memU

English

memU is a memory framework built for 24/7 proactive agents. It is designed for long-running use and greatly reduces the LLM token cost of keeping agents always online, making always-on, evolving agents practical in production systems. memU continuously captures and understands user intent. Even without a command, the agent can tell what you are about to do and act on it by itself. memU Bot — Now open source. The enterprise-ready OpenClaw. Your proactive AI assistant that remembers everything.

memU treats memory like a file system—structured, hierarchical, and instantly accessible. File System memU Memory 🏷️ Categories (auto-organized topics) 🧠 Memory Items (extracted facts, preferences, skills) 🔄 Cross-references (related memories linked) 📥 Resources (conversations, documents, images) Why this matters:

memory/ ├── preferences/ │ ├── communication_style.md │ └── topic_interests.md ├── relationships/ │ ├── contacts/ │ └── interaction_history/ ├── knowledge/ │ ├── domain_expertise/ │ └── learned_skills/ └── context/ ├── recent_conversations/ └── pending_tasks/ Just as a file system turns raw bytes into organized data, memU transforms raw interactions into structured, searchable, proactive intelligence. If you find memU useful or interesting, a GitHub Star ⭐️ would be greatly appreciated. Capability Description 🤖 24/7 Proactive Agent Always-on memory agent that works continuously in the background—never sleeps, never forgets 🎯 User Intention Capture Understands and remembers user goals, preferences, and context across sessions automatically 💰 Cost Efficient Reduces long-running token costs by caching insights and avoiding redundant LLM calls

• Download-and-use and simple to get started (one-click install, Try now: memu.bot · Source: memUBot on GitHub
• Navigate memories like browsing directories—drill down from broad categories to specific facts
• Mount new knowledge instantly—conversations and documents become queryable memory
• Cross-link everything—memories reference each other, building a connected knowledge graph
• Persistent & portable—export, backup, and transfer memory like files

Architecture Overview

The memU system operates through a continuous sync loop between the main agent and the memory bot:

USER QUERY
↓     ↓
🤖 MAIN AGENT ◄───► 🧠 MEMU BOT
↑     ↑
PLAN & EXECUTE   MONITOR, MEMORIZE & PROACTIVE INTELLIGENCE

Key Features:

1. Proactive Content Monitoring

2. Communication Pattern Learning

3. Market Context Awareness

• Tracks user interests and research patterns
• Automatically surfaces relevant new content
• Learns topic preferences from browsing behavior
• Builds response templates for common scenarios
• Identifies priority contacts and urgent keywords
• Adapts communication style based on learned preferences
• Learns trading preferences from historical decisions
• Provides proactive alerts based on defined thresholds
• Correlates news events with portfolio impact

Proactive Behaviors

Content Recommendations

The system monitors user research patterns and automatically suggests relevant content:

User researching AI topics → New RAG optimization papers found
↓
Agent surfaces papers aligned with recent research
↓
Highlights relevant authors and their latest work

Email Management

MemU learns from email patterns over time:

• Auto-drafts context-aware responses
• Categorizes and prioritizes inbox
• Detects scheduling conflicts
• Summarizes long threads with key decisions

Investment Monitoring

• Tracks market movements against user preferences
• Provides proactive alerts based on thresholds
• Learns from executed vs ignored recommendations

Technical Implementation

Three-Layer Memory System

| Layer | Reactive Use | Proactive Use | |-------|-------------|--------------| | Resource | Direct access to original data | Background monitoring for new patterns | | Item | Targeted fact retrieval | Real-time extraction from ongoing interactions | | Category | Summary-level overview | Automatic context assembly for anticipation |

API Endpoints

• POST /api/v3/memory/memorize - Register continuous learning task
• GET /api/v3/memory/memorize/status/{task_id} - Check processing status
• POST /api/v3/memory/categories - List auto-generated categories
• POST /api/v3/memory/retrieve - Query memory with proactive context loading

Retrieval Methods

RAG (Fast Context)

LLM (Deep Reasoning)

• ⚡ Milliseconds response time
• 💰 Embedding-only cost
• Best for: Real-time suggestions
• 🐢 Seconds response time
• 💰💰 LLM inference cost
• Best for: Complex anticipation

Installation & Usage

pip install -e .
# Test with OpenAI API key
export OPENAI_API_KEY=your_api_key
python examples/example_1_conversation_memory.py

Supported Features

• Continuous Learning: Processes inputs in real-time with immediate memory updates
• Proactive Filtering: User-specific and multi-agent context awareness
• Custom LLM Support: Supports OpenRouter, custom embeddings, and vision models
• Multi-modal Support: Handles conversations, documents, images, videos, and audio

Example Implementations

Conversation Memory

export OPENAI_API_KEY=your_api_key
python examples/example_1_conversation_memory.py

Proactive Behaviors:

• Automatically extracts preferences from casual mentions
• Builds relationship models from interaction patterns
• Surfaces relevant context in future conversations
• Adapts communication style based on learned preferences

Skill Extraction

export OPENAI_API_KEY=your_api_key
python examples/example_2_skill_extraction.py

Proactive Behaviors:

• Learns from execution logs and interaction history
• Suggests optimizations based on patterns
• Extracts reusable skills from successful task completions
• Continuously improves performance metrics

中文

memU 是为 24/7 主动代理构建的记忆框架。 它专为长时间运行而设计，大大降低了保持代理始终在线的 LLM token 成本，使始终保持在线、不断发展的代理在实际生产系统中变得实用。

memU 持续捕获并理解用户意图。即使没有命令，代理也能告诉您将要做什么并自主行动。

memU 将记忆视为文件系统——结构化的、分层的、即时可访问的。

文件系统 memU 记忆

📁 文件夹 🏷️ 类别（自动组织的主题）

📄 文件 🧠 记忆项（提取的事实、偏好、技能）

🔗 符号链接 🔄 交叉引用（相关记忆链接）

📂 挂载点 📥 资源（对话、文档、图像）

为什么这很重要：

• 像浏览目录一样导航记忆——从广泛类别到具体事实逐层深入
• 即时挂载新知识——对话和文档变成可查询的记忆
• 交叉链接一切——记忆相互引用，构建连接的知识图谱
• 持久且可移植——像文件一样导出、备份和传输记忆

memU 的三层系统既支持响应式查询，也支持主动上下文加载：

| 层级 | 响应式使用 | 主动使用 | |------|------------|----------| | 资源 | 直接访问原始数据 | 背景监控新模式 | | 项目 | 目标化事实检索 | 从持续交互中实时提取 | | 类别 | 概述级别概览 | 为预期自动组装上下文 |

memU 的主要功能：

🤖 24/7 主动代理：永远在线的记忆代理，在后台持续工作——永不睡眠，永不忘记 🎯 用户意图捕获：自动跨会话理解和记住用户目标、偏好和上下文 💰 成本高效：通过缓存洞察和避免冗余 LLM 调用来降低长时间运行成本

架构概览

memU 系统通过主代理和记忆机器人之间的连续同步循环运行：

用户查询
     ↓     ↓
🤖 主代理 ◄───► 🧠 memU 机器人
     ↑     ↑
计划与执行   监控、记忆与主动智能

主要功能：

1. 主动内容监控

• 跟踪用户兴趣和研究模式
• 自动推送相关新内容
• 从浏览行为中学习主题偏好

2. 通信模式学习

• 为常见场景构建响应模板
• 识别优先联系人和紧急关键词
• 根据学习的偏好调整通信风格

3. 市场上下文感知

• 从历史决策中学习交易偏好
• 根据定义的阈值提供主动警报
• 将新闻事件与投资组合影响关联

API 端点：

• POST /api/v3/memory/memorize - 注册持续学习任务
• GET /api/v3/memory/memorize/status/{task_id} - 检查处理状态
• POST /api/v3/memory/categories - 列出自动生成的类别
• POST /api/v3/memory/retrieve - 查询记忆（支持主动上下文加载）

检索方法

RAG（快速上下文）

• ⚡ 毫秒级响应时间
• 💰 仅嵌入成本
• 适用于：实时建议

LLM（深度推理）

• 🐢 秒级响应时间
• 💰💰 LLM 推理成本
• 适用于：复杂预期

安装与使用

pip install -e .

# 使用 OpenAI API key 测试
export OPENAI_API_KEY=your_api_key
python examples/example_1_conversation_memory.py

示例实现

对话记忆 从对话中自动提取偏好并建立关系模型，在未来的对话中推送相关上下文。

技能提取 从执行日志和交互历史中学习，根据模式建议优化，从成功的任务完成中提取可重用技能。