Tavily MCP

What is Tavily MCP?

Tavily MCP is a server that connects advanced web search and data extraction tools with AI assistants via the standardized Model Context Protocol (MCP). It allows applications and AI models to perform real-time, domain-specific web searches, extract structured data from web pages, map website structures, and conduct automated web crawling—all through secure and configurable endpoints. Tavily MCP is developed by Tavily and is designed for seamless integration with popular AI tools and editors like Claude Desktop, Cursor, and VS Code.

How to Configure Tavily MCP

To set up Tavily MCP, follow these steps:

1. Obtain a Tavily API Key

   • Register on Tavily and generate your free API key.

2. Install Node.js

   • Ensure Node.js v20 or higher is installed.

3. Install Tavily MCP Server

   • Run directly with npx:

     npx -y [email protected]
     

   • Or install via Smithery:

     npx -y @smithery/cli install @tavily-ai/tavily-mcp --client claude
     

4. Configure Your Client Application

   • Claude Desktop/VS Code/Cursor:
     Add the appropriate server configuration JSON with your API key.
     Example for Claude Desktop (macOS):

     {
       "mcpServers": {
         "tavily-mcp": {
           "command": "npx",
           "args": ["-y", "[email protected]"],
           "env": {
             "TAVILY_API_KEY": "your-api-key-here"
           }
         }
       }
     }
     

   • For VS Code, you can modify your User Settings or add a .vscode/mcp.json in your workspace.

5. Start or Refresh Your Client

   • Restart the client application (e.g., Claude Desktop, Cursor, VS Code) after configuration. The Tavily MCP server will appear under available MCP tools.

How to Use Tavily MCP

1. Verify Server Installation

   • After installing and configuring the Tavily MCP server, you should see a hammer icon or a tools menu in your AI app interface.
   • Click the tools icon to see a list of available Tavily MCP functions.

2. Invoke Tools via Chat or Commands

   • In Claude Desktop, Cursor, or VS Code, simply enter a request (e.g., "Search for recent developments in quantum computing") and the assistant will leverage the appropriate Tavily tool.
   • You can explicitly reference tools such as tavily-search and tavily-extract in your prompts for precise control.

3. Combine Tools for Advanced Workflows

   • You may chain search and extract operations, like:
     "Search for news articles about AI startups from the last 7 days and extract the main content from each article to generate a detailed report."

4. Explore Structured Website Mapping and Crawling

   • Use the tavily-map and tavily-crawl tools to build structured overviews or traverse websites automatically.

5. Review and Use Results

   • Results will be returned in your chat interface, ready for further analysis or follow-up queries.

Key Features

• Real-time, domain-specific web search from within AI chat or coding environments.
• Automated extraction of clean, main article or web page content.
• Website mapping for generating a structured view of sites and their link relationships.
• Web crawling for deep, programmatic exploration and data collection.
• Secure integration—API keys and configuration remain within user infrastructure.
• Seamless installation and configuration for Claude Desktop, VS Code, Cursor, and more.
• Supports best practices in data security, provider flexibility, and workflow design.
• Easily extensible for combining Tavily with other MCP-compatible servers such as Neo4j.

Use Cases

• Knowledge Assistant Augmentation:
  Power your AI chat, coding, or automation agents with live web search and document extraction for up-to-the-minute information retrieval and analysis.

• Automated Market and News Analysis:
  Query the latest industry trends or company news and extract article bodies for summarization or sentiment analysis.

• Academic and Scientific Research:
  Search and extract from research portals and scientific journals, combine with domain-specific filters or site restrictions.

• Intelligent Data Pipeline Construction:
  Automate information gathering by crawling and mapping websites, feeding structured datasets into graphs or databases.

• Code Assistant Web Integration:
  Allow code assistants (e.g., in VS Code) to fetch, extract, and analyze content from the web to support user tasks.

FAQ

1. I'm not seeing the Tavily MCP tools in my app interface. What should I do?

• Double-check your configuration file for typos or formatting errors.
• Ensure Node.js and npx are properly installed and accessible from your system PATH.
• Restart your app after configuration changes.

2. How do I fix API key issues?

• Make sure your Tavily API key is correctly pasted, with no extra spaces or quotes.
• Verify that your API key has not expired; generate a new one from the Tavily dashboard if necessary.

3. The Tavily MCP server fails to start with npx errors. What can I do?

• Try specifying the full path to your npx executable in the command field.
• Run which npx (macOS/Linux) or where npx (Windows) to get the correct path.

4. Can I use Tavily MCP with other MCP servers?

• Yes! You can configure your client to connect to multiple MCP servers (e.g., combine Tavily with Neo4j MCP for knowledge graph workflows).

5. Are there costs for using Tavily MCP?

• Tavily offers a free plan with API usage limits. Check Tavily pricing for details and upgrade options.