mcp-webresearch

Description:
mcp-webresearch is an open-source MCP (Model Context Protocol) server designed for conducting web research and integrating real-time information—such as Google search results and webpage content—into the Claude desktop app.

Source: https://github.com/mzxrai/mcp-webresearch

Category: data-access-integration-mcp-servers

Tags: mcp, google, web-search, research

Features

• Google Search Integration: Perform Google searches and extract results programmatically.
• Webpage Content Extraction: Retrieve and extract content from visited web pages.
• Research Session Tracking: Maintains a session log recording search queries, visited pages, extracted content, screenshots, and timestamps.
• Screenshot Capture: Take and save screenshots of visited web pages as MCP resources.
• Agentic-Research Prompt: A guided prompt for Claude that assists in conducting thorough, iterative, and source-cited research.
• Resource Management: Exposes screenshots and research sessions as MCP resources for easy access within Claude Desktop.
• Interactive Research: Allows users to guide and refine research direction interactively, with Claude citing sources.
• Tools:
  • search_google: Performs Google search and extracts results.
  • visit_page: Visits a webpage and extracts content (optionally takes a screenshot).
  • take_screenshot: Captures a screenshot of the current page.
• Platform Support: Verified on macOS and Linux.
• Pre-alpha Status: Early-stage software; bugs may be present.
• Open Source: Licensed under MIT.

Requirements

• Node.js >= 18
• Claude Desktop app
• Playwright (installed as a dependency)

Pricing

• Open Source (MIT License)
• No pricing plans; free to use and modify.

Note: The repository is currently archived and in pre-alpha state. Users may encounter bugs and are encouraged to check logs for troubleshooting.