Frequently Asked Questions about Deep Research MCP Server

Q: What is the Deep Research MCP Server? A: It’s a Model Context Protocol (MCP) compliant server designed for comprehensive web research. It uses Tavily’s Search and Crawl APIs to gather detailed information and structures it for LLMs to create high-quality markdown documents.

Q: What is MCP? A: MCP is an open protocol that standardizes how applications provide context to LLMs.

Q: What are the prerequisites for using the Deep Research MCP Server? A: You need Node.js (version 18.x or later recommended) and npm (comes with Node.js) or Yarn. Also, you need a Tavily API key.

Q: How do I install the Deep Research MCP Server? A: You can install it via Smithery, using NPX, global installation with npm, or by cloning the repository for local project integration.

Q: How do I configure the Tavily API key? A: Set the TAVILY_API_KEY environment variable in a .env file, directly in the command line, or in your system environment variables.

Q: Can I customize the documentation prompt for the LLM? A: Yes, you can override the default prompt by setting the DOCUMENTATION_PROMPT environment variable or passing a documentation_prompt argument directly to the tool.

Q: How do I specify where the research documents and images should be saved? A: You can configure the output path using the output_path parameter, the RESEARCH_OUTPUT_PATH environment variable, or rely on the default path in the user’s Documents folder.

Q: What are the available search parameters? A: Search parameters include search_depth, topic, days, time_range, max_search_results, and more. These parameters allow you to fine-tune the Tavily Search API.

Q: What are the available crawl parameters? A: Crawl parameters include crawl_max_depth, crawl_max_breadth, crawl_limit, crawl_instructions, and more. These parameters allow you to control the Tavily Crawl API.

Q: How does the server work? A: The server receives a CallToolRequest from an LLM or AI agent, performs a Tavily Search, uses Tavily Crawl to extract content, aggregates the information, and returns a JSON string with structured data for the LLM to generate a markdown document.

Q: What is the output structure of the deep-research-tool? A: The tool returns a JSON string with fields like documentation_instructions, original_query, search_summary, research_data, and output_path.

Q: How do I troubleshoot API key errors? A: Ensure that the TAVILY_API_KEY is correctly set and valid.

Q: What should I do if I encounter SDK issues? A: Make sure that @modelcontextprotocol/sdk and @tavily/core are installed and up-to-date.

Q: What if I get no output or errors? A: Check the server console logs for any error messages. Increase verbosity if needed for debugging.

Q: Is the file writing feature secure? A: The file writing feature is disabled by default for security. When enabled, it includes directory restrictions and line limits to prevent abuse. Only enable it in trusted environments.

Q: How does the Deep Research MCP Server relate to UBOS? A: UBOS is a full-stack AI Agent Development Platform that allows you to orchestrate AI Agents, connect them with your enterprise data, build custom AI Agents with your LLM model, and create Multi-Agent Systems. The Deep Research MCP Server can be seamlessly integrated into UBOS workflows to enhance the research capabilities of AI agents.