MCP Directory

Tavily

Official

Production-ready MCP server for real-time web search, content extraction, site mapping, and crawling.

Verified
stdio (local)
API key
TypeScript

Add to your client

Copy the config for your MCP client and paste it into its config file.

Install / run
npx -y tavily-mcp@latest

Paste into ~/Library/Application Support/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "tavily": {
      "command": "npx",
      "args": [
        "-y",
        "tavily-mcp@latest"
      ],
      "env": {
        "TAVILY_API_KEY": "<your-tavily-api-key>"
      }
    }
  }
}

Step-by-step guides: Add to Claude Desktop · Add to Cursor · Add to Windsurf

Before you start

  • A Tavily API key from https://app.tavily.com (free account available, no credit card)
  • For local use: Node.js v20 or later with npx
  • An MCP-compatible client such as Claude Desktop, Claude Code, or Cursor

About Tavily

The Tavily MCP Server is the official Model Context Protocol server from Tavily, a search engine built specifically for AI agents. It packages real-time web search, content extraction, site crawling, and site mapping into a single server so agents can both find and retrieve structured web content.

It exposes four core tools: tavily-search for real-time web search, tavily-extract for pulling clean structured content from specific URLs, tavily-crawl for systematically traversing a site, and tavily-map for building a structured map of a website's URL graph. Search supports basic (1 credit) and advanced (2 credit) depths.

Tavily can run two ways: as a hosted remote server at mcp.tavily.com (connect over HTTP with your API key in the URL, with optional OAuth), or locally via npx for stdio clients. Defaults can be tuned with the DEFAULT_PARAMETERS environment variable.

Tools & capabilities (4)

tavily-search

Real-time web search optimized for AI agents.

tavily-extract

Extract clean, structured content from specific URLs.

tavily-crawl

Systematically crawl and explore a website.

tavily-map

Build a structured map of a website's URL structure.

When to use it

  • Use it when you want an agent to perform real-time web research with results tuned for LLM consumption.
  • Use it when you need clean, structured content extracted from a known set of URLs.
  • Use it when you want to crawl an entire site or section to gather context.
  • Use it when you need a map of a website's URL structure before deciding what to fetch.
  • Use it when you want a hosted, zero-install search backend reachable over HTTP.

Quick setup

  1. 1Sign up at tavily.com and copy your API key from the dashboard.
  2. 2For remote use, add it via the HTTP endpoint, e.g. `claude mcp add --transport http tavily https://mcp.tavily.com/mcp/?tavilyApiKey=<key>`.
  3. 3For local use, configure your client to run `npx -y tavily-mcp@latest` with TAVILY_API_KEY set.
  4. 4Optionally set DEFAULT_PARAMETERS to tune default search behavior.
  5. 5Restart your MCP client and verify with a test tavily-search call.

Security notes

Needs a TAVILY_API_KEY which gates billable API usage; keep it secret and rotate if leaked. Crawl/extract tools fetch arbitrary remote pages, so validate any URLs you pass and treat returned content as untrusted.

Tavily FAQ

Is there a free tier?

Yes. Tavily's free plan includes 1,000 API credits per month with no credit card required. Basic search costs 1 credit and advanced search costs 2 credits; unused credits do not roll over.

Can I use it without installing anything locally?

Yes. Tavily hosts a remote MCP server at mcp.tavily.com that you connect to over HTTP with your API key, so no local install is needed. A local stdio option via npx is also available.

How is it authenticated?

With a Tavily API key, passed either as the TAVILY_API_KEY environment variable (local) or in the remote URL query string. The remote server also supports optional OAuth.

Which clients are supported?

Documented clients include Claude Desktop, Claude Code, and Cursor (one-click install), and it works with any standard MCP client over stdio or HTTP.

Alternatives to Tavily

Compare all alternatives →

Official MCP reference server that fetches a URL and returns its content as clean Markdown, with chunking.

Verified
stdio (local)
No auth
Python
2 tools
Updated 8 days agoRepo

Official Firecrawl MCP server — scrape, crawl, map, search, and structured extraction for any LLM client.

Verified
stdio (local)
API key
TypeScript
8 tools
Updated 6 days agoRepo

Exa's neural web search and crawling MCP server — runs locally via npx or as a hosted remote endpoint.

Verified
stdio (local)
API key
TypeScript
6 tools
Updated 7 days agoRepo