Open-source AI agent with brain-inspired long-term memory. Three-path fusion retrieval + hierarchical bubble compaction + surprise detection. Self-hosted, mobile-first PWA.
<div align="center"> <img src="web/public/bubble-192.png" alt="Bubble Agent OS" width="120" /> # Bubble Agent OS **An AI assistant that actually *remembers*.** Every conversation, every imported spreadsheet, every piece of knowledge becomes a "bubble" that links, compresses, and evolves — so the AI understands you better over time. **一个真正有记忆的 AI 助手。** 每次对话、每份导入的数据都会变成一个个"泡泡",通过关联、压缩、演化,让 AI 越来越懂你。 [](LICENSE) [](https://nodejs.org) [](https://www.typescriptlang.org) [](https://github.com/luckincoco/bubble-agent-os/stargazers) [](https://github.com/luckincoco/bubble-agent-os/pulls) </div> <div align="center"> <img src="docs/screenshots/chat-view.png" alt="Chat Interface" width="720" /> </div> --- ## Why Bubble? / 为什么是泡泡? Most AI chatbots are goldfish — they forget everything after each conversation. RAG helps, but it's just basic vector search. **Bubble Agent OS** takes a fundamentally different approach inspired by how human memory actually works. 大多数 AI 聊天机器人都是"金鱼记忆"——每次对话后什么都忘了。RAG 有所帮助,但本质上只是向量检索。**泡泡 Agent OS** 从人类记忆的工作方式出发,采用了截然不同的方法。 > **Bubble Theory / 泡泡理论**: Every piece of information is a small bubble. Related bubbles naturally cluster together into larger bubbles. Over time, atomic facts compress into concepts, and concepts compress into user portraits — progressively approximating a true understanding of you. > > 每一条信息都是一个小泡泡。相关的泡泡自然聚合成更大的泡泡。随着时间推移,原子事实被压缩为概念,概念被压缩为用户画像——逐步逼近对你的真正理解。 ### Bubble Agent vs Traditional RAG | | Traditional RAG | Bubble Agent OS | |---|---|---| | **Retrieval** | Vector similarity only | Keyword + Vector + Graph + Recency (4-path fusion) | | **Weights** | Fixed |
HAL 分层混合模型工作流 — 强模型(Claude)负责理解/拆解/验收,低成本模型(DeepSeek)负责检索/提取/清洗。Hermes Agent skill。
An LLM agent fine-tuned on DeepSeek for spaced repetition, dynamically integrating knowledge points based on the Ebbinghaus forgetting curve.
基于 STM32F103 构建的端到端 AI 智能手表生态。自研“零重定位”原生机器码动态加载引擎与页面栈式 UI 框架;集成生产级 OTA 回滚保护机制与高带宽(921600 baud)串口协议栈。通过 Node.js 中继实现 DeepSeek AI 语义控制及 ASRPRO 语音全双工交互,是一个集成了分布式计算、现代存储管理与 AI Agent 的嵌入式全栈工程。
A Meta-Agent-Driven Self-Evolving Multi-Agent System for UAV Detection and Tracking
One command to run Hermes AI Agent with a browser UI. Zero prerequisites. 一行命令,AI 就位。
网页应用Agent,接入DeepSeek、Mimo等模型