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
✨ 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:
npm install
- Build the server:
npm run build
- Configure environment:
Modify your
claude_desktop_config.jsonwith the following content (adjust paths according to your system):
"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:
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.
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:
npx -y @smithery/cli install @chanmeng666/google-news-server --client claude
Installing via mcp-get
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 Processbefore this command.
Manual Installation
# 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 stringgl: Country code (e.g., 'us', 'uk')hl: Language code (e.g., 'en', 'es')topic_token: Token for specific news topicspublication_token: Token for specific publishersstory_token: Token for full coverage of a storysection_token: Token for specific sections
🔧 Development
# 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
{
"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
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.