G-Search MCP

What is G-Search MCP?

G-Search MCP is a Model Context Protocol server that enables parallel Google searches with multiple keywords simultaneously using Playwright. It is designed for users of AI assistants like Claude Desktop who need structured search results.

How to use G-Search MCP?

Install Playwright browser with npx playwright install chromium. Run directly with npx -y g-search-mcp. Configure as an MCP server in Claude Desktop by adding the command to the mcpServers configuration. Use the search tool with parameters queries, limit, timeout, locale, and debug to perform searches.

Key features of G-Search MCP

• Parallel searching with multiple keywords simultaneously
• Opens multiple tabs in a single browser instance
• Automatic CAPTCHA detection with visible browser mode
• Simulates real user browsing behavior to reduce detection
• Returns structured JSON search results
• Configurable parameters (limit, timeout, locale, debug)

Use cases of G-Search MCP

• Research multiple topics in parallel without manual switching
• Automate competitive analysis by searching several queries at once
• Gather structured data for data scraping or analysis projects
• Use with AI assistants to fetch current search results dynamically

FAQ from G-Search MCP

What is the main advantage over standard Google search?

G-Search MCP supports parallel searching with multiple keywords, opens multiple tabs in a single browser, and automatically handles CAPTCHA while simulating real user behavior.

What are the runtime dependencies?

Node.js 18 or higher and NPM or Yarn are required. Playwright must have the Chromium browser installed via npx playwright install chromium.

How do you configure the locale for search results?

Set the locale parameter in the search tool (e.g., "zh-CN"). The default locale is en-US.

How does the server handle CAPTCHA?

It intelligently detects CAPTCHA and enables visible browser mode for user verification when needed.

What is the default search result limit per query?

The default limit is 10 results per query. It can be changed via the limit parameter.