Home / MCP / MCP Web Research Server

MCP Web Research Server

Provides real-time web research via Google search, page extraction, session tracking, and screenshots for Claude integration.

javascript

17stars

View on GitHub

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

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

The MCP Web Research Server lets you bring real-time information into Claude to support deep, topic-focused research. It combines web search, content extraction from pages, session tracking, and screenshots so you can research efficiently inside your chat.

How to use

You interact with the server through an MCP client. Start a chat with Claude and ask questions that benefit from live web research. If you want a guided workflow, use the agentic-research prompt to drive thorough exploration, citations, and iterative refinement. The server tracks your research session, including search queries, visited pages, extracted content, and screenshots, so you can review your research history and results.

How to install

Prerequisites are Node.js version 18 or newer (which includes npm and npx) and the Claude Desktop application.

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

Additional setup and usage notes

To use the web research features inside Claude Desktop, ensure Claude Desktop is installed and configured to launch the MCP server automatically when needed. You can start a session by sending a prompt in Claude that benefits from live web data.

Prompts specialize the research flow. The agentic-research prompt provides a guided, citation-rich approach. Access it in Claude Desktop by clicking the Paperclip icon in the chat input, choosing an integration, selecting webresearch, and picking agentic-research.

Tools

search_google — Performs Google searches and extracts results. Arguments: { query: string }.

visit_page — Visits a webpage and extracts its content. Arguments: { url: string, takeScreenshot?: boolean }.

take_screenshot — Takes a screenshot of the current page. No arguments required.

Prompts

agentic-research is a guided research prompt that helps Claude conduct thorough web research. It instructs Claude to start with broad searches, prioritize high-quality sources, iteratively refine the direction, keep you informed, and always cite sources with URLs.

Resources and session data

Screenshots are stored as MCP resources and can be accessed via Claude Desktop using the Paperclip icon. The server maintains a research session that includes search queries, visited pages, extracted content, screenshots, and timestamps.

Suggestions

If you skip the agentic-research prompt, consider suggesting high-quality sources for Claude to use. For example, prompt with news today from Reuters or AP to guide general-topic research.

Problems

This is early-stage software and may contain bugs. If you encounter issues, check Claude Desktop MCP logs for clues. A common diagnostic command is provided below.

Troubleshooting

tail -n 20 -f ~/Library/Logs/Claude/mcp*.log

Development

You may work on the server locally by installing dependencies, building, and running in development mode as needed.

Requirements

Node.js version 18 or newer is required. Playwright is automatically installed as a dependency.

Verified platforms

macOS is supported.

License

MIT

Author

mzxrai

Available tools

search_google

Performs Google searches and extracts results.

visit_page

Visits a webpage and extracts its content.

take_screenshot

Takes a screenshot of the current page.