Tavily MCP Server

Overview

Tavily MCP Server is the official Model Context Protocol (MCP) server developed by Tavily. It bridges AI coding assistants and agents (Claude, Cursor, Windsurf, etc.) directly to Tavily’s AI-optimized web search and extraction APIs.

Agents can now perform real-time research, fetch up-to-date information, extract structured data from pages, map entire websites, and crawl domains — all using natural language without manual API integration.

Key Features

• tavily-search: Advanced real-time web search with AI-generated summaries, relevance scoring, and topic filtering
• tavily-extract: Intelligent structured data extraction from any webpage
• tavily-map: Generate site maps and understand website structure
• tavily-crawl: Crawl entire domains or specific paths with depth control
• Remote Hosted Option: Use Tavily’s official remote MCP endpoint (no local setup needed)
• Local Deployment: Run via npx or Docker with your own Tavily API key
• Production Ready: Optimized for low latency, high reliability, and agent reasoning

How It Works

1. Obtain a free Tavily API key at tavily.com
2. Connect your AI agent to the MCP server (remote or local)
3. Prompt naturally, e.g., “Search for the latest AI news and extract key statistics”, “Map the structure of example.com”, or “Crawl and summarize the pricing pages of top SaaS tools”
4. The agent receives structured, high-quality results via MCP tools

Installation & Setup

Remote (Recommended) Use Tavily’s hosted MCP server with your API key:

https://mcp.tavily.com/mcp/?tavilyApiKey=tvly-YOUR_API_KEY

Local (npx)

TAVILY_API_KEY=tvly-YOUR_API_KEY npx -y tavily-mcp@latest

Docker

docker run -e TAVILY_API_KEY=tvly-... mcp/tavily

Supported AI Agents

• Claude Desktop / Claude Code
• Cursor
• Windsurf
• Roo Code and other MCP-compatible clients

Use Cases

• Real-time research and fact-checking for AI agents
• RAG (Retrieval-Augmented Generation) workflows with fresh web data
• Automated competitive analysis and market research
• Content extraction and summarization pipelines
• Website mapping and crawling for SEO or data collection
• Building autonomous research agents

Official Links

• GitHub Repository: https://github.com/tavily-ai/tavily-mcp
• Documentation: https://docs.tavily.com/documentation/mcp
• Tavily Website: https://tavily.com
• npm Package: tavily-mcp