MCP Server - Model Context Protocol Server

A modular Node.js backend server that provides unified API endpoints for Weather, News, Stock Market, and AI services. Designed as a backend connector for AI models, enabling structured tool execution through natural language query parsing.

Overview

MCP Server (Model Context Protocol Server) acts as an intelligent middleware layer between external data services and AI systems. It integrates multiple third-party APIs and provides:

• Standardized REST endpoints
• Redis-based caching
• Natural language to structured tool-call parsing via OpenAI
• Clean, modular backend architecture

The system is built to handle real-world API interactions efficiently with caching strategies based on data volatility.

Tech Stack

Runtime & Framework

• Node.js (ES Modules)
• Express.js

Libraries

• axios – HTTP client for external API calls
• dotenv – Environment variable management
• ioredis – Redis client for caching
• openai – OpenAI API integration
• nodemon – Development auto-restart

Architecture

The project follows a modular backend structure:

mcp-server/
├── routes/        → API route definitions
├── controllers/   → Request handling logic
├── services/      → External API integrations & business logic
├── utils/         → Shared utilities
├── server.js      → Application entry point
├── package.json   → Dependencies and scripts
└── .env           # Environment variables

Design Principles

• Separation of concerns
• Service-layer abstraction for external APIs
• Centralized error handling
• Volatility-based caching strategy
• AI-powered request parsing

Features

API Integrations

Weather Service

• OpenWeatherMap integration
• Redis caching (TTL: 2 minutes)

News Service

• NewsAPI integration
• Returns top 5 articles by topic
• Cached responses

Stock Service

• API Ninjas stock data
• Redis caching (TTL: 1 minute)

AI Service

• OpenAI integration
• Parses natural language into structured tool calls
• Acts as an intelligent query router

API Endpoints

Weather

GET /api/weather/:city

Returns weather data for a specified city.

News

GET /api/news/:topic

Returns top 5 news articles for a topic.

Stocks

GET /api/stocks/:symbol

Returns stock price data for a given symbol.

AI Parsing

POST /api/ai/parse

Parses natural language input into structured tool calls.

Caching Strategy

The server uses Redis with differentiated TTL values based on data volatility:

Service	TTL	Reason
Weather	2 minutes	Moderate volatility
Stocks	1 minute	High volatility
News	Short-lived caching	Frequently updated

This reduces external API load and improves response times.

Environment Variables

Create a .env file with the following variables:

PORT=3000
WEATHER_API_KEY=your_key
NEWS_API_KEY=your_key
STOCK_API_KEY=your_key
OPENAI_API_KEY=your_key
OPENAI_MODEL=gpt-4o-mini
REDIS_URL=redis://localhost:6379

Defaults:

• OPENAI_MODEL → gpt-4o-mini
• REDIS_URL → redis://localhost:6379

Installation & Setup

Clone the repository:

git clone https://github.com/Daniel-wesley-06/mcp-server.git
cd mcp-server

Install dependencies:

npm install

Start the server:

npm start

Development mode (auto-restart):

npm run dev

The server runs on the port defined in the PORT environment variable.

Key Learnings

• Implementing volatility-based caching strategies
• Integrating LLMs as routing layers for backend systems
• Structuring backend services for extensibility
• Managing secure external API integrations
• Designing an AI-ready backend connector pattern

Future Improvements

• Health check endpoint
• Request logging middleware
• Basic test coverage
• Docker support
• CI/CD integration