mcp-webresearch

Overview

The MCP Web Research Server provides real-time web research capabilities for Claude by integrating Google search, webpage content extraction, and screenshot functionality. It maintains research sessions with visited pages, search queries, and timestamps. While marked as pre-alpha code, it offers a functional interface through Claude Desktop with tools for searching Google, visiting pages, and taking screenshots. The server uses Playwright for web automation and requires Node.js 18+ to run.

Try asking AI

After installing, here are 6 things you can ask your AI assistant:

• search_google{ query: string }

  Performs Google searches and extracts results

  Try:Search for latest AI researchTry:Find news about climate change
• visit_page{ url: string, takeScreenshot?: boolean }

  Visits a webpage and extracts its content

  Try:Visit https://example.com and summarize the contentTry:Extract information from this article
• take_screenshot

  Takes a screenshot of the current page

  Try:Capture a screenshot of this pageTry:Take a screenshot for reference

Comparable tools

Installation

{
  "mcpServers": {
    "webresearch": {
      "command": "npx",
      "args": ["-y", "@mzxrai/mcp-webresearch@latest"]
    }
  }
}

FAQ

What platforms are supported?

Currently verified for macOS. Linux support is marked as work in progress.

Is this ready for production use?

No, the README explicitly states this is pre-alpha code, so expect bugs and limited functionality.

On Hacker News

Recent discussion from the developer community.

• Show HN: Research the web from Claude Desktop using MCP▲ 2
  Story by mbm · 2024-12-06

Compare mcp-webresearch with