home / mcp / mcp deep web research server

MCP Deep Web Research Server

Enhanced MCP server for deep web research

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "qpd-v-mcp-deepwebresearch": {
      "command": "mcp-deepwebresearch",
      "args": [],
      "env": {
        "EXAMPLE_ENV": "PLACEHOLDER"
      }
    }
  }
}

The MCP Deep Web Research Server provides automated web research capabilities inside the Model Context Protocol (MCP). It combines intelligent search queuing, rich content extraction, and structured results to help you gather, analyze, and organize information from the web within Claude workflows.

How to use

You interact with this MCP server through an MCP client. Start a chat with your MCP-enabled assistant and request research tasks that can benefit from live web data. Use the built-in tools to perform broad explorations, refine topics, visit pages for content, and consolidate findings into structured outputs.

How to install

Prerequisites: Node.js 18 or newer and Claude Desktop installed on your machine.

Global installation (recommended)

# Install globally using npm
npm install -g mcp-deepwebresearch

# Or using yarn
yarn global add mcp-deepwebresearch

# Or using pnpm
pnpm add -g mcp-deepwebresearch

Claude Desktop integration

After installation, add the MCP server entry to Claude Desktop so it starts automatically when needed.

{
  "mcpServers": {
    "deepwebresearch": {
      "command": "mcp-deepwebresearch",
      "args": []
    }
  }
}
```
Location: Windows: %APPDATA%\Claude\claude_desktop_config.json
Location: macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

First-time setup

Install required browser dependencies to enable web navigation and page content extraction.

npx playwright install chromium

Usage notes

In Claude, start a new chat and prompt the server to perform a research task. You can also use the prebuilt prompt integration labeled agentic-research to guide the workflow. When you issue a query, the server will manage searches, visit pages as needed, and return structured findings with sources.

Tools

The MCP server exposes the following tools.

Troubleshooting and tips

If you encounter rate limiting or timeouts, adjust the concurrency settings and timeout thresholds in your environment to keep within MCP limits. Confirm that browser dependencies are installed and that the Claude Desktop MCP integration is enabled for automatic startup.

Development notes

Development workflows include dependency installation, build, and test cycles. Use the provided commands to run locally and verify features.

Configuration and environment

The server supports environment-level configuration to tune performance and reliability. Typical options include maximum parallel searches, delay between searches, retry counts, and request timeouts. Ensure these align with your MCP runtime constraints to maintain responsiveness.

Security and privacy considerations

When performing web research, be mindful of privacy and data handling. Limit the depth and breadth of scraping to avoid violating site terms of use. Use the server's timeout and error-handling features to prevent long-running or unresponsive sessions.

Requirements

Node.js >= 18 and Playwright dependencies are required for full functionality.

Available tools

deep_research

Performs comprehensive research with content analysis across multiple sources, returning main topics, insights, and sources.

parallel_search

Performs multiple Google searches in parallel with intelligent queuing and a limit on parallelism.

visit_page

Visits a webpage and extracts its content, returning URL, title, and Markdown-formatted content.