MCP Memory Server - HTTP Streaming

MCP Memory Server - HTTP Streaming

Enables creation and management of knowledge graphs with entities, relationships, and observations through HTTP streaming. Supports persistent storage, search functionality, and CRUD operations for building and querying interconnected knowledge bases.

Category
Visit Server

README

MCP 内存服务器 - HTTP 流式传输

基于官方 MCP Memory Server 重构的 HTTP 流式传输版本,使用 Hono.js 框架。

项目来源

本项目基于 MCP Servers 仓库 中的官方 @modelcontextprotocol/server-memory 重构而来。原始内存服务器使用 stdio 传输,而此版本经过重构以支持 HTTP 流式传输,适用于独立部署场景。

与原版本的主要差异

  • 传输方式: HTTP 流式传输,而非 stdio
  • 框架选择: 使用 Hono.js,而非 Express.js(其他 HTTP 服务器使用)
  • 架构设计: 按单一职责原则重构为多文件架构
  • 部署方式: 可作为独立 HTTP 服务部署

架构设计

原始的单文件实现已重构为模块化结构:

src/memory-http/
├── types.ts           # 接口定义 (Entity, Relation, KnowledgeGraph)
├── storage.ts         # 文件存储操作
├── knowledge-graph.ts # KnowledgeGraphManager 类及所有图操作
├── tools.ts          # MCP 工具定义
├── server.ts         # MCP 服务器逻辑
└── http-server.ts    # Hono.js HTTP 服务器及流式传输支持

功能特性

完全保留原始内存服务器的所有功能:

  • 知识图谱存储: 存储和管理实体、关系和观察
  • 文件持久化: JSON 行格式存储(与原版兼容)
  • 搜索能力: 按名称、类型或观察内容查询实体
  • CRUD 操作: 对所有图元素的完整增删改查支持

可用工具

  1. create_entities - 在知识图谱中创建新实体
  2. create_relations - 在实体间创建关系
  3. add_observations - 向现有实体添加观察
  4. delete_entities - 删除实体及相关关系
  5. delete_observations - 从实体中移除特定观察
  6. delete_relations - 删除特定关系
  7. read_graph - 读取整个知识图谱
  8. search_nodes - 按查询字符串搜索实体
  9. open_nodes - 按名称检索特定实体

安装和使用

安装依赖

cd src/memory-http
pnpm install

构建项目

pnpm build

启动 HTTP 服务器

pnpm start

服务器将默认在 http://localhost:3002/mcp 启动。

环境变量

  • PORT - 服务器端口(默认:3002)
  • MEMORY_FILE_PATH - 内存存储文件路径(默认:./memory.json)

HTTP API 端点

服务器提供符合 MCP 协议的端点:

  • GET / - 服务器基本信息
  • GET /health - 健康检查端点(用于服务监控)
  • POST /mcp - 初始化会话并处理 MCP 请求
  • GET /mcp - 服务器发送事件(SSE)流式传输实时更新
  • DELETE /mcp - 会话终止

所有 MCP 请求必须包含 mcp-session-id 请求头进行会话管理。

健康检查示例

curl http://localhost:3002/health

响应示例:

{
  "status": "healthy",
  "service": "mcp-memory-http",
  "version": "0.6.3",
  "timestamp": "2024-01-15T10:30:00.000Z",
  "sessions": 2
}

存储格式

存储格式与原始内存服务器完全相同:

  • 每行包含一个带有 type 字段的 JSON 对象("entity" 或 "relation")
  • 实体包含 name、entityType 和 observations 数组
  • 关系包含 from、to 和 relationType 字段

基础用法和测试示例

1. 启动服务器

cd src/memory-http
pnpm start

服务器将在 http://localhost:3002/mcp 启动。

2. 使用 cURL 测试

初始化会话

curl -X POST http://localhost:3002/mcp \
  -H "Content-Type: application/json" \
  -d '{
    "jsonrpc": "2.0",
    "id": 1,
    "method": "initialize",
    "params": {
      "protocolVersion": "2024-11-05",
      "capabilities": {
        "tools": {}
      },
      "clientInfo": {
        "name": "test-client",
        "version": "1.0.0"
      }
    }
  }'

响应将包含 mcp-session-id 头,后续请求需要使用此 ID。

创建实体

curl -X POST http://localhost:3002/mcp \
  -H "Content-Type: application/json" \
  -H "mcp-session-id: YOUR_SESSION_ID" \
  -d '{
    "jsonrpc": "2.0",
    "id": 2,
    "method": "tools/call",
    "params": {
      "name": "create_entities",
      "arguments": {
        "entities": [
          {
            "name": "张三",
            "entityType": "人物",
            "observations": ["软件工程师", "喜欢编程", "在北京工作"]
          },
          {
            "name": "阿里巴巴",
            "entityType": "公司",
            "observations": ["中国电商巨头", "总部在杭州", "成立于1999年"]
          }
        ]
      }
    }
  }'

创建关系

curl -X POST http://localhost:3002/mcp \
  -H "Content-Type: application/json" \
  -H "mcp-session-id: YOUR_SESSION_ID" \
  -d '{
    "jsonrpc": "2.0",
    "id": 3,
    "method": "tools/call",
    "params": {
      "name": "create_relations",
      "arguments": {
        "relations": [
          {
            "from": "张三",
            "to": "阿里巴巴",
            "relationType": "在...工作"
          }
        ]
      }
    }
  }'

查询知识图谱

curl -X POST http://localhost:3002/mcp \
  -H "Content-Type: application/json" \
  -H "mcp-session-id: YOUR_SESSION_ID" \
  -d '{
    "jsonrpc": "2.0",
    "id": 4,
    "method": "tools/call",
    "params": {
      "name": "read_graph"
    }
  }'

搜索实体

curl -X POST http://localhost:3002/mcp \
  -H "Content-Type: application/json" \
  -H "mcp-session-id: YOUR_SESSION_ID" \
  -d '{
    "jsonrpc": "2.0",
    "id": 5,
    "method": "tools/call",
    "params": {
      "name": "search_nodes",
      "arguments": {
        "query": "软件工程师"
      }
    }
  }'

添加观察

curl -X POST http://localhost:3002/mcp \
  -H "Content-Type: application/json" \
  -H "mcp-session-id: YOUR_SESSION_ID" \
  -d '{
    "jsonrpc": "2.0",
    "id": 6,
    "method": "tools/call",
    "params": {
      "name": "add_observations",
      "arguments": {
        "observations": [
          {
            "entityName": "张三",
            "contents": ["精通 TypeScript", "有5年工作经验"]
          }
        ]
      }
    }
  }'

3. 查看存储数据

查看 memory.json 文件内容:

cat src/memory-http/memory.json

示例输出:

{"type":"entity","name":"张三","entityType":"人物","observations":["软件工程师","喜欢编程","在北京工作","精通 TypeScript","有5年工作经验"]}
{"type":"entity","name":"阿里巴巴","entityType":"公司","observations":["中国电商巨头","总部在杭州","成立于1999年"]}
{"type":"relation","from":"张三","to":"阿里巴巴","relationType":"在...工作"}

4. 使用 MCP 客户端连接

如果使用支持 MCP 协议的客户端(如 Claude Desktop),可以在配置中添加:

{
  "mcpServers": {
    "memory-http": {
      "command": "node",
      "args": ["/path/to/src/memory-http/dist/http-server.js"],
      "env": {
        "PORT": "3002"
      }
    }
  }
}

5. TypeScript 测试脚本

我们还提供了一个 TypeScript 测试脚本 test-memory-server.ts,展示如何以编程方式与 MCP 服务器交互:

# 使用 npm 脚本运行测试
pnpm test

# 或者直接运行
npx tsx test-memory-server.ts

TypeScript 测试脚本包含以下测试用例:

  • 会话初始化和管理
  • 创建实体和关系
  • 添加观察数据
  • 搜索和查询功能
  • 完整的知识图谱读取
  • 存储文件验证

6. Bash 测试脚本

也可以使用 Bash 脚本进行测试:

# 使用 npm 脚本运行
pnpm run test:bash

# 或者直接运行
./test-memory-server.sh

运行测试:

chmod +x test-memory-server.sh
./test-memory-server.sh

测试数据示例

测试完成后,memory.json 文件将包含类似以下内容:

{"type":"entity","name":"赵六","entityType":"人物","observations":["全栈开发者","TypeScript专家","在广州工作","熟悉Node.js","有4年工作经验","毕业于中山大学"]}
{"type":"entity","name":"WebApp项目","entityType":"项目","observations":["React应用","使用TypeScript","REST API"]}
{"type":"entity","name":"TechStart","entityType":"公司","observations":["初创公司","总部在广州","专注Web开发"]}
{"type":"relation","from":"赵六","to":"TechStart","relationType":"在...工作"}
{"type":"relation","from":"赵六","to":"WebApp项目","relationType":"开发"}

开发

监视模式

pnpm watch

文件结构

  • types.ts - TypeScript 接口和类型定义
  • storage.ts - 知识图谱持久化的文件 I/O 操作
  • knowledge-graph.ts - 图操作的核心业务逻辑
  • tools.ts - MCP 工具模式定义
  • server.ts - MCP 服务器设置和请求处理
  • http-server.ts - 基于 Hono.js 的 HTTP 服务器实现

兼容性

此服务器与以下内容完全兼容:

  • 原始 MCP Memory Server 数据格式
  • MCP 协议规范
  • stdio 版本的现有内存文件

许可证

MIT - 与原始 MCP Servers 项目相同

Recommended Servers

playwright-mcp

playwright-mcp

A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.

Official
Featured
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.

Official
Featured
Local
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
graphlit-mcp-server

graphlit-mcp-server

The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.

Official
Featured
TypeScript
Kagi MCP Server

Kagi MCP Server

An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.

Official
Featured
Python
E2B

E2B

Using MCP to run code via e2b.

Official
Featured
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Exa Search

Exa Search

A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.

Official
Featured
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured