g-search-mcp

G Search MCP Server

g-search-mcp runs Google queries in parallel and returns structured results through MCP. It serves Claude Desktop users and other MCP workflows that need search results grouped by query, with controls for result limits, locale, and browser visibility.

Server facts

g-search-mcp

Project

Author
jae-jae
Website
1team.top
License
MIT

Tech

Runtime / language
node
Transports
stdio
Remote available
No
Authentication
Not specified

Activity

GitHub stars
273
Last commit
Data as of

Tags

  • Google Search
  • g search
  • parallel searching

Overview

Google searches for several topics can run together through g-search-mcp’s search tool. It accepts an array of queries and uses Playwright to search in separate tabs within a shared browser instance. The JSON response groups results under their corresponding queries, with titles, links, and snippets for further processing. This gives MCP clients, including Claude Desktop, a way to request several searches together rather than submit each query separately.

Search behavior can be adjusted for the task: callers can cap results per query, change the page-loading timeout, select a locale, or choose not to save browser state. A debug setting makes the browser visible for an individual operation and takes precedence over the startup debug flag. When CAPTCHA verification is detected, the project exposes the browser so the user can complete it. The server runs on Node.js, communicates over stdio, and requires a Playwright Chromium installation. Its source is available under the MIT license.

g-search-mcp setup notes

  • g-search-mcp records stdio only; confirm a network transport upstream before planning remote access.

Related MCP servers