Deep Research

GitHub
Claude Deep Research

An MCP (Model Context Protocol) server that enables comprehensive research capabilities for Claude and other MCP-compatible AI assistants. This server integrates web and academic search functionality, allowing AI models to access current information from multiple sources, follow relevant links, and provide well-structured research results.

Overview

Claude Deep Research is a powerful research tool that extends the capabilities of LLMs by providing:

  1. Web search integration through DuckDuckGo
  2. Academic research access through Semantic Scholar
  3. Content extraction from web pages
  4. Comprehensive analysis with structured formatting
  5. Visualization guidance for data representation

The server follows MCP design principles to provide a seamless integration with Claude and other AI assistants.

Features

  • Unified Research Tool: Single interface for web and academic information
  • Multi-Source Integration: Combines information from various sources into cohesive research
  • Content Extraction: Pulls relevant information from web pages
  • Academic Source Discovery: Finds scholarly articles related to your topic
  • Smart Formatting: Properly formats research with citations
  • Visual Framework: Provides guidance for creating effective data visualizations
  • Structured Analysis: Organizes research using academic methodologies

Research Workflow

Installation

Prerequisites

  • Python 3.8 or higher
  • pip or uv package manager

Quick Install

# Using pip
pip install mcp httpx beautifulsoup4

# Clone the repository
git clone https://github.com/yourusername/claude-deep-research.git

Configuration

The server works out of the box with default settings, but you can modify the following parameters in deep_research.py for customization:

# Configuration
USER_AGENT = "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36"
MAX_CONTENT_SIZE = 8000  # Maximum characters in the final response
MAX_RESULTS = 3         # Maximum number of results to process

Usage

Running the Server

Modify your Claude desktop config and restart Claude. On a Mac this is at ~/Library/Application Support/Claude

  "search-scholar": {
      "command": "<Path to Python>/python",
      "args": [
        "<Path to deep research>/deep_research.py"
      ]
    }

Using with Claude Desktop

Once installed, you can access the server in Claude Desktop:

  1. Tool Access: Use the deep_research tool directly in conversation

Research Tool

The main deep_research tool accepts the following parameters:

  • query (required): The research question or topic
  • sources (optional): Which sources to use: "web", "academic", or "both" (default)
  • num_results (optional): Number of sources to examine (default 2, max 3)

Example prompts:

Can you research the latest developments in quantum computing?

I need comprehensive information about climate change mitigation strategies. Use the deep_research tool to help me.

Research the history and cultural significance of origami using academic sources.

Research Prompt

The server includes a structured research prompt that guides Claude through a comprehensive research process:

  1. Initial Exploration: Gathers information from multiple sources
  2. Preliminary Synthesis: Organizes findings with visualization
  3. Follow-up Research: Identifies and explores knowledge gaps
  4. Comprehensive Analysis: Integrates all information with visual elements
  5. Proper Citations: Formats references using APA style

Troubleshooting

Common Issues

  • Server Connection Failures: Ensure you're using the correct path to the server file.
  • Search Errors: Some searches may time out or return limited results. Try a more specific query.
  • Web Access Issues: The server requires internet access to function properly.
  • Content Formatting: Very large responses may be truncated to fit within size limits.

Logs

The server outputs logs to stderr that can help diagnose issues:

# View logs when running directly
python deep_research.py 2> server.log

# View logs from Claude Desktop (macOS/Linux)
tail -f ~/Library/Logs/Claude/mcp-server-deepresearch.log

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

Acknowledgments

Made with ❤️ for extending AI capabilities through MCP