Google News MCP Server

A Model Context Protocol (MCP) server implementation that provides Google News search capabilities via SerpAPI integration. Automatically categorizes news results and supports multiple languages and regions.

https://github.com/user-attachments/assets/1cc71c27-f840-4c94-9ab5-460d84ba4779

屏幕截图 2024-12-30 021446

屏幕截图 2024-12-30 021524

屏幕截图 2024-12-30 021914

屏幕截图 2024-12-30 021941

✨ Features

🔍 Flexible Search Options

Comprehensive search capabilities including query-based search, topic search, publication filtering and story coverage.

🌐 Global Coverage

Supports multiple languages and regions through configurable language and country codes.

📊 Smart Categorization

Automatically categorizes news results into topics like AI & Technology, Business, Science & Research, and Healthcare.

🔀 Multiple Result Types

Handles various news result types including headlines, stories, related topics and menu links.

🛠️ Robust Error Handling

Comprehensive error handling for API failures and invalid inputs, with helpful error messages.

🌍 Language Support

Automatic fallback to English for unsupported language codes with appropriate user notifications.

🔑 SerpApi Setup Guide

Before getting started, you'll need to obtain a SerpApi key. Here's how:

Visit SerpApi website and create an account
After registration, go to your Dashboard:
- Locate the "API Key" section
- Copy your API key
- New users get 250 free API calls
API Usage Details:
- Free tier: 250 searches per month
- Paid plans start at $75/month for 5000 searches
- Billing based on successful API calls
- Multiple payment methods: Credit Card, PayPal, etc.
Usage Limits:
- Request Rate: 2 requests/second
- IP Restrictions: None
- Concurrent Requests: 5
- Response Cache Time: 1 hour

👩‍🔧 Solution for MCP Servers Connection Issues with NVM/NPM

Click to view my configuration solution 👉 https://github.com/modelcontextprotocol/servers/issues/76

🚀 Quick Start

Install dependencies:

Terminal

npm install

Build the server:

Terminal

npm run build

Configure environment: Modify your claude_desktop_config.json with the following content (adjust paths according to your system):

json

    "google-news": {
      "command": "D:\\Program\\nvm\\node.exe",
      "args": [
        "D:\\github_repository\\path_to\\dist\\index.js"
      ],
      "env": {
        "SERP_API_KEY": "your-api-key"
      }
    }

Start the server:

Terminal

npm start

Troubleshooting

Invalid API Key

Verify API key configuration in claude_desktop_config.json
Confirm API key is active in SERP API dashboard

Request Failures

Check network connectivity
Verify API call quota hasn't been exceeded
Validate request parameter format

Running evals

The evals package loads an mcp client that then runs the index.ts file, so there is no need to rebuild between tests. You can load environment variables by prefixing the npx command. Full documentation can be found here.

bash

OPENAI_API_KEY=your-key  npx mcp-eval src/evals/evals.ts src/index.ts

📦 Installation

Installing via Smithery

To install Google News for Claude Desktop automatically via Smithery:

Terminal

npx -y @smithery/cli install @chanmeng666/google-news-server --client claude

Installing via mcp-get

Terminal

npx @michaellatman/mcp-get@latest install @chanmeng666/google-news-server

If you are using an old version of Windows PowerShell, you may need to run Set-ExecutionPolicy Bypass -Scope Process before this command.

Manual Installation

@chanmeng666/google-news-server

bash

# Using npm
npm i @chanmeng666/google-news-server
# or
npm install @chanmeng666/google-news-server

# Using yarn
yarn add @chanmeng666/google-news-server

# Using pnpm
pnpm add @chanmeng666/google-news-server

💻 Tech Stack

📖 API Documentation

The server implements the Model Context Protocol and exposes a Google News search tool with the following parameters:

q: Search query string
gl: Country code (e.g., 'us', 'uk')
hl: Language code (e.g., 'en', 'es')
topic_token: Token for specific news topics
publication_token: Token for specific publishers
story_token: Token for full coverage of a story
section_token: Token for specific sections

🔧 Development

bash

# Run in development mode with hot reload
npm run dev

# Run linting
npm run lint

# Run tests
npm run test

📝 License

This project is MIT licensed.

🙋‍♀ Author

Created and maintained by Chan Meng.

🤖 AI Agent Instructions & GEO Optimization

For AI Agents and LLM Crawlers

This MCP server is optimized for AI agent usage with comprehensive instructions and structured data.

Quick AI Usage Guide

config.json

{
  "tool": "google_news_search",
  "parameters": {
    "q": "artificial intelligence latest developments",
    "gl": "us",
    "hl": "en"
  }
}

AI-Friendly Features

Automatic Categorization: News results are automatically sorted into relevant categories
Structured Output: Returns formatted data optimized for AI processing
Multi-language Support: Supports various languages with automatic fallbacks
Error Handling: Comprehensive error messages and graceful degradation

Response Format

The server returns categorized news with the following structure:

Categories: AI & Technology, Business, Science & Research, Healthcare, General
Article Fields: title, source, link, date, snippet, authors
Metadata: timestamp, menu_links for related topics

Best Practices for AI Agents

Use specific, descriptive keywords for better results
Leverage automatic categorization for topic-based workflows
Implement proper error handling and retry logic
Respect API rate limits (2 requests/second)
Parse structured responses for better context understanding

Advanced Usage Patterns

News Monitoring: Use company names or stock symbols for business news
Research Exploration: Leverage topic tokens for field-specific searches
Breaking News: Use "breaking news" or "latest news" queries
Multi-language: Combine appropriate gl and hl parameters

Error Handling

Invalid API key: Check SERP_API_KEY environment variable
Unsupported language: Server auto-falls back to English
Rate limiting: Respect 2 requests/second limit
Invalid parameters: Validate parameter types before calling

📊 GEO Metrics & Monitoring

Generative Engine Optimization (GEO) Features

This server includes comprehensive monitoring and optimization for AI agent usage.

Key Metrics Tracked

Citation Success Rate: Percentage of successful tool calls
AI Traffic Conversion Rate: Percentage of AI agents successfully using the tool
Query Coverage Rate: Percentage of queries returning relevant results
Response Time: Average response time for tool calls
Categorization Accuracy: Accuracy of automatic news categorization
Link Carry Rate: Percentage of responses including source links

Monitoring Configuration

typescript

interface GEOMonitoringConfig {
  trackingEnabled: boolean;
  metricsEndpoint: string;
  reportingInterval: number; // minutes
  alertThresholds: {
    errorRate: number;
    responseTime: number;
    successRate: number;
    satisfactionScore: number;
  };
}

Performance Optimization

Real-time metrics collection
Automated alerting for performance issues
Query pattern analysis for optimization
Agent usage tracking and analytics

Data-Driven Improvements

Continuous monitoring of AI agent satisfaction
Query success rate analysis
Response time optimization
Categorization accuracy improvements

For technical implementation details, see src/geo-metrics.ts.

🔍 Structured Data & Metadata

AI-Optimized Metadata

This server provides comprehensive structured data for AI agents and search engines.

JSON-LD Structured Data

The server includes structured data following Schema.org standards:

Software application metadata
Feature descriptions and capabilities
Technical requirements and dependencies
Usage patterns and integration guidelines

MCP Protocol Compliance

Protocol Version: 1.0.0
Tool Name: google_news_search
Response Format: Structured text with categorized results
Rate Limits: 2 requests/second, 5 concurrent requests

AI Discovery Features

llms.txt: Comprehensive AI usage guide at root level
Structured Metadata: Enhanced manifest.json with AI-specific information
Error Handling: AI-friendly error messages and fallbacks
Documentation: Detailed usage examples and best practices

Integration Guidelines

Designed for seamless integration with other MCP servers
Optimized for AI agent workflows
Supports various AI platforms and frameworks
Provides clear error handling and debugging information

For complete structured data, see structured-data.json.

Hosted deployment

A hosted deployment is available on Fronteir AI.

Chan Meng
Need a custom app like this one? I build them — let's talk.