MCP Multi-Server System
A dual-server MCP system with PostgreSQL integration that provides financial tools (stock prices, portfolio calculation, financial news) on one server and utility tools (weather, time, data processing, text analysis) on another server.
README
MCP Multi-Server System
A multi-server system for the Model Context Protocol (MCP), featuring a unified frontend and two specialized backend servers for different toolsets.
🏗️ Architecture
The system consists of three main components:
- Frontend API (Port 3000): A FastAPI application that serves the web interface for managing agents and tools, and acts as a gateway to the backend MCP servers.
- MCP Server A (Port 3001): A FastAPI application that provides finance-related tools.
- MCP Server B (Port 3002): A FastAPI application that provides general utility tools.
- PostgreSQL Database: A central database for storing information about servers, tools, and agents.
🚀 Getting Started
1. Prerequisites
- Python 3.8+
- PostgreSQL
2. Installation
- Clone the repository.
- Install the required Python packages:
pip install -r requirements.txt
3. Configuration
-
Create a
.envfile in the root directory of the project. -
Add the following environment variables to the
.envfile, adjusting the values as needed:# Database Configuration DATABASE_URL=postgresql://user:password@localhost/mcp_config # Server Ports PORT_A=3001 PORT_B=3002
4. Running the System
You can run the entire system with a single command:
python app/main.py
This will start the Frontend API on port 3000, MCP Server A on port 3001, and MCP Server B on port 3002.
You can access the web interface by navigating to http://localhost:3000 in your web browser.
Alternatively, you can run each server individually:
# Terminal 1: Start the Frontend API
uvicorn app.frontend_api:app --port 3000 --reload
# Terminal 2: Start MCP Server A
uvicorn app.server_a:app --port 3001 --reload
# Terminal 3: Start MCP Server B
uvicorn app.server_b:app --port 3002 --reload
🔗 API Endpoints
Frontend API (Port 3000)
/: The main web interface for managing agents and tools./servers: Get a list of all available MCP servers./agents: Manage agents (create, list, update, delete)./manage-tools/{server_name}: A web interface for managing the tools of a specific server.
MCP Servers (Ports 3001 & 3002)
/health: Check the health of the server./tools: List all available tools on the server./check-tools: A simplified endpoint to check the available tools./tools/call: Execute a tool on the server.
Example: Check the tools on Server A
curl http://localhost:3001/check-tools
Example: Execute the get_stock_price tool on Server A
curl -X POST http://localhost:3001/tools/call \
-H "Content-Type: application/json" \
-d
{
"name": "get_stock_price",
"arguments": {
"symbol": "AAPL"
}
}
🛠️ Development
Adding a New Tool
- Add the tool's logic to the appropriate server (
app/server_a.pyorapp/server_b.py). - Register the tool in the database using the web interface or by calling the
/toolsendpoint.
Database
The database schema is defined by the SQLAlchemy models in app/database.py. The database is initialized with some default data when the system starts up.
📝 License
This project is licensed under the MIT License.
推荐服务器
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。
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。
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。
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。
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。
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。
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。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
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。