mcp -deep-research

The Deep Research MCP Server provides a guided workflow for conducting thorough, source-backed research. It supports a structured process that starts with elaborating your research question, then breaking it down into subquestions, performing web searches, analyzing sources, and finally generating a well-cited report. This server is designed to leverage Claude’s web search and analysis capabilities to gather diverse perspectives and present findings with proper citations and clear formatting. To get started, deploy the server using the published configuration, then access the Deep Research prompt template in MCP to begin a session. As you work, you’ll provide a research question and the server will help you expand the question, generate subquestions, fetch sources, critically evaluate them, and assemble a final report with citations and structured sections.

Prerequisites:

• Python 3.8+ installed on your system
• Access to Claude Desktop for integrated web search capabilities (if you’re using Claude features)
• MCP environment ready to run Python-based servers

Installation steps:

1. Install the published Deep Research MCP Server (Python-based) via your MCP setup tooling. This guide assumes you are using the uvx approach as shown in the README.
2. Ensure you have the required dependencies installed. If you’re using a virtual environment, activate it before proceeding.
3. Start the server using the published configuration:

Code:

• Ensure you have uv installed and available in your PATH. If you don’t, install via your Python package manager according to your environment.
• Start the server using the provided command: uvx mcp-server-deep-research

4. Verify the server is running by checking the MCP console and confirming the Deep Research prompt template is accessible.

Notes:

• If you encounter permission issues on macOS or Linux, ensure your user has rights to the project directory and Python executable.
• If Claude integration is used, ensure appropriate API access and any required authentication steps are completed as per Claude Desktop setup.

• The server relies on Claude's capabilities for web search and analysis when the Deep Research workflow requires external sources. Ensure Claude Desktop is installed and configured if you intend to leverage built-in web search features.
• The development configuration shows a local path and a directory-based run; for production, the published uvx-based command is recommended to simplify deployment.
• You can customize the Deep Research prompts under the MCP prompts section using the published/modeled prompt names, e.g., deep-research, to tailor expansion and analysis behavior.
• If you run into source citation issues, verify that your source collection step correctly captures bibliographic details and that the final report renderer formats citations consistently.