MCP Deep Web Research Server
by qpd-v
A Model Context Protocol (MCP) server designed for advanced web research, enhancing Claude's capabilities. It brings real-time information into Claude with intelligent search queuing, enhanced content extraction, and deep research capabilities.
Last updated: N/A
What is MCP Deep Web Research Server?
The MCP Deep Web Research Server is a tool that extends the capabilities of Claude by providing advanced web research functionalities. It integrates with Claude through the Model Context Protocol (MCP) to enable intelligent search, content extraction, and deep web research.
How to use MCP Deep Web Research Server?
First, install the server globally or locally using npm, yarn, or pnpm. Then, configure Claude Desktop by adding the server details to the claude_desktop_config.json file. After installation, run npx playwright install chromium to install browser dependencies. Finally, start a chat with Claude and use prompts that benefit from web research, such as the provided agentic-research prompt.
Key features of MCP Deep Web Research Server
Intelligent Search Queue System with rate limiting and error recovery
Enhanced Content Extraction with TF-IDF relevance scoring and keyword proximity analysis
Google search integration
Webpage content extraction and Markdown conversion
Use cases of MCP Deep Web Research Server
Comprehensive research on complex topics
Gathering real-time information for Claude
Extracting structured data from web pages
Improving the accuracy and depth of Claude's responses
FAQ from MCP Deep Web Research Server
What is the purpose of the deep_research tool?
What is the purpose of the deep_research tool?
The deep_research tool performs comprehensive research with content analysis, allowing you to specify the topic, search depth, branching, timeout, and minimum relevance score.
How does the parallel_search tool work?
How does the parallel_search tool work?
The parallel_search tool performs multiple Google searches in parallel with intelligent queuing, limited to a maximum of 5 concurrent searches for reliable performance.
What does the visit_page tool do?
What does the visit_page tool do?
The visit_page tool visits a webpage and extracts its content, providing the URL, title, and Markdown-formatted content.
What is the agentic-research prompt?
What is the agentic-research prompt?
The agentic-research prompt is a guided research prompt that helps Claude conduct thorough web research, prioritizing high-quality sources and iteratively refining the research direction.
What should I do if I encounter a 'Too many requests' error?
What should I do if I encounter a 'Too many requests' error?
If you encounter a 'Too many requests' error, try increasing the SEARCH_DELAY_MS environment variable or decreasing the MAX_PARALLEL_SEARCHES variable.