Google News

📰 Google News MCP

Give any AI agent live news. This Model Context Protocol server lets Claude, Cursor, VS Code and other MCP clients search Google News, read the top stories in any country, and pull the text of an article.

It's built on google-news-scraper, so there's no API key, no account and no headless browser. Results come with the real publisher URL rather than a Google redirect link.

Tools

Each article comes back as { title, link, source, sourceUrl, datetime }.

Setup

Your MCP client starts the server itself, so there's nothing to keep running.

Claude Desktop

Download google-news.mcpb and double-click it, or drag it into Settings → Extensions. Claude Desktop installs it with its own Node.js, so you don't need Node installed.

Claude Code

/plugin install google-news --marketplace lewisdonovan/google-news-scraper

Or add the server directly with claude mcp add google-news -- npx -y google-news-scraper-mcp.

Cursor, Windsurf and most other clients

Requires Node.js 20.19 or later.

Add this to the client's MCP config (.cursor/mcp.json, claude_desktop_config.json, and so on):

{  "mcpServers": {    "google-news": {      "command": "npx",      "args": ["-y", "google-news-scraper-mcp"]    }  }}

VS Code

code --add-mcp '{"name":"google-news","command":"npx","args":["-y","google-news-scraper-mcp"]}'

Optional: a fallback when Google blocks you

Google rate-limits busy IPs, which is common on cloud servers. If you have a SerpApi key, the server can fetch results through SerpApi's Google News API instead. Set these environment variables in your client's MCP config:

{  "mcpServers": {    "google-news": {      "command": "npx",      "args": ["-y", "google-news-scraper-mcp"],      "env": { "GOOGLE_NEWS_THIRD_PARTY_APIKEY": "your-key", "GOOGLE_NEWS_THIRD_PARTY_PRIMARY": "false" }    }  }}

In Claude Desktop, enter the key in the extension's settings instead. Each SerpApi call uses one of your SerpApi searches.

Example prompts

• "What's the latest news on Nvidia from the past day?"
• "Summarise today's top stories in Germany, in English."
• "Find coverage of our product launch on techcrunch.com this month and read the most recent article."

Notes

• Google rate-limits bursts of requests from one IP. The scraper retries with backoff, but very heavy use can still get temporarily blocked. A third-party key keeps results coming.
• get_article_text reads the page's HTML directly, so paywalled pages and sites that only render with JavaScript may return an error.

Privacy

The server collects and stores nothing. It sends your queries to Google News, or to SerpApi if you set a key, and fetches the article pages you ask it to read. See the privacy policy.

License

MIT