Dynamic MCP Server

Dynamic MCP Server

A flexible, extensible framework for building MCP servers with API key authentication, user management, and dynamic tool sharing.

Category
访问服务器

README

Dynamic MCP Server Framework

A flexible, extensible framework for building Model Context Protocol (MCP) servers with modern API key authentication, user management, and dynamic tool sharing.


🚀 Overview

Dynamic MCP Server enables secure, user-aware, and extensible AI tool servers. It supports:

  • API key-based authentication (simple, compatible with all MCP clients)
  • User management and authorization (MongoDB-backed)
  • Session-based, per-user tool loading
  • Tool sharing and fine-grained access control
  • Extensible HTTP and database layers for downstream projects

🌟 Key Features

  • Dynamic Tool Management: Create, delete, and authorize tools at runtime—tools are not limited to static definitions at startup or in code. This enables true runtime extensibility and is a primary differentiator from most other MCP servers.
  • User Management: Add, update, delete, and list users; admin bootstrapping; role-based access.
  • Tool Sharing: Share tools with other users, manage access levels, and receive real-time updates.
  • Modern Auth: Simple API key authentication, MongoDB for authorization.
  • Extensibility: Add custom HTTP routes and MongoDB collections in downstream projects.
  • Session-based Tool Loading: Tools are loaded per user session, not globally.

🔑 Authentication (API Key)

  • Each user is assigned a unique apiKey (generated automatically on user creation).
  • To authenticate, clients must provide the apiKey via query parameter OR header:
    • Query Parameter: ?apiKey=your-key or ?apikey=your-key
    • Header: x-apikey: your-key or apikey: your-key
  • Streamable HTTP Transport (Modern): /mcp?apiKey=... or /mcp with header
  • The server authenticates the user by looking up the apiKey in the database.

🚀 Transport Protocols

Dynamic MCP Server supports both legacy and modern MCP transport protocols:

Streamable HTTP Transport

  • Protocol Version: 2025-03-26
  • Endpoint: /mcp
  • Features: Modern, efficient, single-endpoint design
  • Authentication: Query parameter ?apiKey=your-key OR header x-apikey: your-key
  • Usage: Recommended for new integrations

� Documentation

Tool Visibility Model: All tools are visible to users by default. Users can hide tools from their view using the hide-tool and unhide-tool actions. See User Management and Tool Management & Sharing for details.


🛠️ Quick Start

See Getting Started for installation and basic usage.


📝 Contributing

Contributions are welcome! See the documentation for more details.

📄 License

MIT


🧩 Handler & Tool Registration (New Pattern)

Tools are now registered via handler packages. Each handler package groups related tools and provides a handler function for tool execution. This is the recommended pattern for extensibility and maintainability.

Example:

const myHandlerPackage = {
  name: "my-domain",
  tools: [
    {
      name: "my-tool",
      description: "A custom tool",
      inputSchema: {
        /* ... */
      },
      handler: { type: "my-domain", config: {} },
      rolesPermitted: ["user", "admin"],
    },
    // ...more tools...
  ],
  handler: async (args, context, config, toolName) => {
    // Dispatch logic based on toolName, if needed
  },
};

const server = new DynamicMcpServer({ name: "my-mcp", version: "1.0.0" });
await server.start();
await server.registerHandler(myHandlerPackage);

See Getting Started and Examples for more details.


🧩 Tool Argument Mapping (config.args)

When defining a tool, you can use the config.args field to map user-supplied tool options (from inputSchema) to the arguments your action handler expects. This mapping supports:

  • Literal values (e.g., "country": "US")
  • Template variables (e.g., "city": "{{location}}")
  • Nested objects and arrays (all templates are resolved recursively)

Example:

handler: {
  type: "weather-tools",
  config: {
    url: "https://api.openweathermap.org/data/2.5/weather",
    args: {
      queryParams: {
        appid: "{{OPENWEATHER_API_KEY}}",
        q: "{{location}}",
        units: "{{units}}"
      }
    }
  }
}

When the tool is called, the system automatically resolves all {{...}} templates in config.args using the tool input and environment variables. The handler receives the merged and resolved arguments—no manual mapping is needed.

See Tool Authoring for full details and examples.



推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选