MCP Server for Deep Research is a tool for in-depth research on complex topics, designed to help users explore questions deeply, find relevant sources, and generate structured research reports. It is positioned as a personal research assistant that transforms research questions into comprehensive, fully cited reports.
Purpose: Used for comprehensive research on complex topics. Users can first download and install Claude Desktop, then run python setup.py in Terminal on macOS to complete installation and setup. To begin research, select the deep-research prompt template from MCP and provide a research question to start.
Main capabilities (complete research workflow):
- Question Elaboration: Expand and clarify the research question, identify key terms and concepts, and define scope and parameters.
- Subquestion Generation: Create subquestions focused on different aspects, ensure comprehensive coverage of the main topic, and provide structure for systematic research.
- Web Search Integration: Use Claude's built-in web search capabilities to perform targeted searches for each subquestion, identify relevant and authoritative sources, and gather diverse perspectives on the topic.
- Content Analysis: Evaluate information quality and relevance, synthesize findings from multiple sources, and provide appropriate citations for all sources.
- Report Generation: Generate well-structured, comprehensive reports as artifacts, correctly cite all sources used, present balanced views with evidence-based conclusions, and use appropriate formatting to ensure clarity and readability. This server provides the deep-research prompt, designed specifically for comprehensive research tasks using a structured approach.
How to use:
- Claude Desktop must be downloaded;
- On macOS, run python setup.py in Terminal to complete installation and setup;
- Select the deep-research prompt template from MCP, then provide a research question to begin research.
Modifying the server:
- Claude Desktop configuration file location: on macOS, ~/Library/Application\ Support/Claude/claude_desktop_config.json; on Windows, %APPDATA%/Claude/claude_desktop_config.json.
- Configuration methods for development (unpublished servers) and published servers are described separately in the repository.
Build and publish:
- Sync dependencies: uv sync
- Build distribution packages: uv build, which generates source and wheel distribution packages in the dist/ directory.
- Publish to PyPI: uv publish
Other information: Contributions are welcome, including fixing bugs, adding features, or improving documentation. The project uses the MIT License, see the LICENSE file for details. An official YouTube demo link is also provided:https://youtu.be/_a7sfo5yxoI。
Limitations: Information such as applicable platforms, system requirements, pricing, and additional usage restrictions is not yet provided.
