AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
MCP verified MIT Self-run

Yusi

mcp-aseubel-yusi · by Aseubel

面向长期陪伴与自我理解的隐私优先 AI 记忆平台,融合 RAG、人生图谱、多模型路由与 MCP 能力网关。

No reviews yet
0 installs
9 views
0.0% view→install

Install

$ agentstack add mcp-aseubel-yusi

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access Used
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/mcp-aseubel-yusi)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
10d ago

Declared compatibility

Claude CodeClaude DesktopCursorWindsurf

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.

How agent discovery & health will work →
Are you the author of Yusi? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

Yusi

一个隐私优先、以长期记忆为核心的开源 AI companion。

English · Issues · 安装指南

> 当前项目处于 active development / developer preview 阶段。接口、配置和数据模型仍可能演进,不建议直接用于生产关键数据。

Yusi 是什么

Yusi(「予思」)把日记、对话和人生经历组织成可持续积累的个人记忆。它不是给传统应用附加一个聊天窗口,而是围绕 AI 的感知、记忆、推理与行动来构建一个能长期理解用户的 companion。

核心理念是:认识自我,遇见同频。 记录具体时刻和真实选择,理解会比标签更接近一个人的本来面貌。

核心能力

| 能力 | 说明 | | --- | --- | | Memory Journal | 记录重要时刻、选择和感受,支持加密存储与富文本内容 | | Layered Memory | 结合短期对话、中期记忆、长期摘要和向量检索,形成可持续的上下文 | | RAG Chat | 从个人记忆中检索相关内容,再生成有依据的对话回复 | | Life Graph | 抽取人物、地点、事件和情绪,构建可探索的人生关系图谱 | | Situation Room | 在具体情境中记录选择,生成行为与情绪分析 | | Soul Matching | 基于行为和叙事的深层理解探索精神共鸣,而非简单标签匹配 | | Model Control Plane | 按语言、场景和故障状态统一管理模型路由、权重与 failover | | MCP Gateway | 通过标准 MCP 对外提供记忆工具;Go 网关经 gRPC 调用 Java 内部能力,并复用后端鉴权与数据边界 |

架构概览

External MCP clients
        |
        v
Go MCP Gateway (HTTP / Streamable HTTP / SSE)
        | gRPC
        v
Java internal capabilities (memory, diary, life graph)
        |
Spring Boot API / WebSocket
        |
Domain services + AI capability layer
        |
MySQL · Redis · Milvus/Zilliz · Object Storage

MCP 层是面向外部模型和客户端的能力适配边界,不是一个独立的记忆后端。记忆查询、解密和 scope 鉴权仍由 Java 后端负责,Go 服务只负责协议适配、工具注册和请求转发。

技术栈

  • Backend: Java 21, Spring Boot 3.4.5, Spring Data JPA, MySQL, Redis, Milvus/Zilliz
  • AI: LangChain4j 1.18.0, OpenAI-compatible APIs, DashScope, RAG, embeddings
  • Integration: gRPC, Protocol Buffers, MCP (Model Context Protocol), WebSocket
  • Frontend: React 19, TypeScript, Vite, Tailwind CSS, Radix UI, Zustand, Tiptap
  • MCP gateway: Go, MCP Go SDK, Gin, gRPC
  • Security: JWT, AES/GCM encrypted diary content, scoped MCP authorization

项目结构

yusi/
├── src/main/java/com/aseubel/yusi/
│   ├── controller/       # HTTP / WebSocket 接口
│   ├── service/          # 按领域组织的业务能力
│   │   ├── ai/           # chat, embedding, prompt, rag, model, asr 等
│   │   ├── memory/       # 记忆检索、摘要与上下文
│   │   └── cognition/    # 情绪与认知分析
│   ├── repository/       # 持久化访问
│   ├── pojo/             # Entity、DTO 与领域数据结构
│   ├── config/           # Spring、AI、数据和安全配置
│   └── grpc/             # 对外部 MCP 网关开放的内部能力边界
├── src/main/resources/   # application 配置与模板
├── frontend/             # React web client
├── mcp/                  # Go MCP gateway 与 protobuf
└── docs/                 # PRD、设计、指南、计划与工程记录

快速开始

环境要求

  • Java 21+
  • Maven 3.9+(或仓库内 Maven Wrapper)
  • Node.js 18+
  • pnpm 11.9+
  • Go 1.25+
  • MySQL 8+
  • Redis 7+
  • Milvus/Zilliz(向量检索功能需要,可按配置选择)

1. 获取代码与基础设施

git clone https://github.com/Aseubel/yusi.git
cd yusi

创建数据库:

CREATE DATABASE yusi CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci;

2. 配置后端

复制并按本地环境调整 src/main/resources/application-dev.yml,至少配置 MySQL、Redis、模型 API 和日记加密密钥。API key 只应通过本地未提交配置或环境变量注入,禁止写入 Git。

export YUSI_ENCRYPTION_KEY=""
export CHAT_MODEL_APIKEY=""
export EMBEDDING_MODEL_APIKEY=""

3. 启动后端

./mvnw spring-boot:run

4. 启动前端

cd frontend
pnpm install
pnpm run dev

5. 启动 MCP 网关(可选)

先确保 Java 后端的 gRPC 地址和鉴权配置已就绪:

cd mcp
go mod download
go run ./cmd/server

默认 MCP 端口为 11611。支持 Streamable HTTP(推荐)和传统 SSE;具体客户端配置、scope 与 proto 生成方式见 [mcp/README.md](mcp/README.md)。

开发验证

# 后端编译(不执行测试)
./mvnw -DskipTests compile

# 前端类型检查与生产构建
cd frontend
pnpm run build

# MCP 编译检查
cd ../mcp
go build ./...

文档导航

  • [安装与本地开发](docs/guides/installation.md)
  • [产品理念与开发哲学](docs/design/philosophy.md)
  • [后端设计](docs/design/backend-design.md)
  • [模型管理与路由框架](docs/design/model-management-framework.md)
  • [LangChain4j 1.18 架构演进记录](docs/record/langchain4j-1.18-architecture-evolution.md)
  • [后端目录结构审计记录](docs/record/backend-structure-review-2026-08-02.md)
  • [PRD v4](docs/prd/prd_v4.md)

许可证

本项目基于 [MIT License](LICENSE) 开源。

Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.