Google News Data — MCP Server

AI agents and AI-native applications get programmatic access to Google News search results, full article content, and trending top headlines through a public RSS feed. No API key required — this MCP server handles authentication, parsing, and rate limiting so you can focus on building intelligence pipelines, not wrestling with feed formats.

Overview

The Google News MCP server delivers three core capabilities: news search by keyword with configurable location and result count, full article content extraction from any Google News URL, and current top headlines segmented by category (business, technology, science, health, sports, entertainment). Built on the public Google News RSS index, it requires zero authentication while respecting rate limits to ensure reliable operation. Each tool is priced as a pay-per-event (PPE) call on the Apify platform, making it cost-predictable for both low-volume monitoring and high-throughput aggregation systems.

Features

• Real-time news search — Query Google News by keyword with geographic filtering (US, GB, DE, FR, etc.) and configurable result counts from 1 to 100 articles per call
• Full article extraction — Fetch complete article content including title, author, publication date, and body text from any Google News article URL
• Category-based headlines — Retrieve current top headlines across seven categories: general, business, technology, science, health, sports, and entertainment
• No authentication required — Public RSS feed access means zero API keys, zero OAuth, zero configuration overhead
• AI-native protocol — Standard JSON-RPC 2.0 over HTTP POST to /mcp endpoint, compatible with Claude, GPT-4, and any MCP-compatible agent
• Structured output — All tools return well-formed JSON with consistent field naming for easy parsing and downstream processing
• Geographic localization — Filter news by country to capture regional perspectives and local market intelligence
• Configurable result volume — Control how many results you receive per query to balance comprehensiveness against processing overhead

Quick Start

# Install dependencies
npm install

# Run a news search via command line
echo '{"tool": "search_news", "params": {"query": "artificial intelligence", "location": "US", "num_results": 10}}' | INPUT_STORE=1 apify run

# Run via MCP protocol
curl -X POST http://localhost:9000/mcp \
  -H "Content-Type: application/json" \
  -d '{"method": "tools/call", "params": {"name": "search_news", "arguments": {"query": "AI startups", "num_results": 5}}}'

Input Schema

Parameter	Type	Required	Default	Description
tool	string	Yes	—	Tool name: search_news, get_article, or get_top_headlines
params.query	string	No	—	Search query for search_news tool
params.location	string	No	"US"	Geographic location code (US, GB, DE, FR, etc.)
params.num_results	integer	No	20	Number of results, 1-100
params.url	string	No	—	Article URL for get_article tool
params.category	string	No	"general"	Headline category for get_top_headlines

Tools

search_news

Search Google News and return article metadata including title, URL, source name, published date, and text snippet.

• Parameters: query (required), location (US/GB/DE/FR/etc.), num_results (1-100, default 20)
• PPE cost: $0.03 per call
• Returns: Array of article objects with title, url, source, published, snippet

{
  "tool": "search_news",
  "params": {
    "query": "machine learning",
    "location": "US",
    "num_results": 5
  }
}

get_article

Fetch full article content from a Google News URL. Returns complete article metadata and body text suitable for RAG pipelines, content analysis, or archival storage.

• Parameters: url (required, must be a Google News article URL)
• PPE cost: $0.05 per call
• Returns: Article object with title, content, source, author, published, url

{
  "tool": "get_article",
  "params": {
    "url": "https://news.google.com/articles/..."
  }
}

get_top_headlines

Get current top headlines for a given category. Returns trending news stories segmented by topic, useful for daily briefings, sentiment tracking, and topic monitoring.

• Parameters: category (general/business/technology/science/health/sports/entertainment), location
• PPE cost: $0.03 per call
• Returns: Array of headline objects with title, url, source, published

{
  "tool": "get_top_headlines",
  "params": {
    "category": "technology",
    "location": "US"
  }
}

Output Structure

All tools return standardized JSON with consistent field naming:

{
  "status": "success",
  "data": [
    {
      "title": "Article Title Here",
      "url": "https://news.google.com/articles/...",
      "source": "Publication Name",
      "published": "2026-05-19T10:30:00Z",
      "snippet": "First 200 characters of article description..."
    }
  ],
  "meta": {
    "query": "search term",
    "location": "US",
    "count": 5,
    "cost": 0.03
  }
}

Pricing

All tools are priced as pay-per-event (PPE) calls on the Apify platform. Each tool call counts as one event regardless of the number of results returned.

Tool	PPE Cost	Notes
search_news	$0.03/call	Up to 100 results per call
get_article	$0.05/call	Full article content extraction
get_top_headlines	$0.03/call	Category-filtered headline results

Cost estimation: A typical monitoring pipeline processing 100 queries per day with 10 articles each would cost approximately $0.60/day ($18/month).

Troubleshooting

Articles returning empty content: Some paywalled or premium publisher articles may return truncated content. Use the snippet field from search_news as a fallback content source.

Rate limiting: The public RSS feed has implicit rate limits. If you encounter 429 errors, implement exponential backoff and reduce query frequency. For high-volume use cases, consider caching results.

Location filtering not returning localized results: Google News location filtering uses ISO country codes. Verify you're passing valid codes (US, GB, DE, FR, JP, etc.). Invalid codes fall back to US results.

Timeouts on large result sets: Requesting num_results above 50 increases processing time. For bulk operations, batch requests with lower result counts and implement parallel processing.

MCP protocol errors: Ensure you're sending JSON-RPC 2.0 formatted requests with correct Content-Type: application/json headers. The endpoint path is /mcp (not /api/mcp or other variants).

Data Source

All data is sourced from the public Google News RSS index at news.google.com. This is a freely accessible feed with no authentication requirements. The MCP server handles feed parsing, HTML content extraction, and response formatting.

Use Cases

• Market intelligence monitoring and competitive analysis
• Brand mention tracking across thousands of publishers
• Content opportunity research for editorial teams
• News aggregation pipelines for media products
• AI training data collection for NLP models
• Real-time sentiment analysis on current events

Build and Deploy

npm install
apify build
apify push

Configuration

• Memory: 1024 MB
• Timeout: 3600s
• Template: ts-standby MCP actor
• Port: containerPort (not standbyPort)
• Protocol: JSON-RPC 2.0 over HTTP POST

License

MIT