1. Home
  2. Servers
  3. Search & Web Data
  4. Tavily MCP Server

Server profile

Tavily MCP Server

Tavily MCP Server connects AI clients to web search, page extraction, site mapping, and crawling. Run it locally with an API key or connect to the hosted service using API-key or OAuth authentication.

View repositoryInstall & configure

Server facts

Tavily MCP Server

Project

Author
tavily-ai
Repository
github.com/tavily-ai/tavily-mcp
License
MIT
Category
Search & Web Data

Tech

Runtime / language
node
Transports
stdio
Remote available
No
Authentication
Not specified

Activity

GitHub stars
2,433
Last commit
Data as of

Tags

  • Content Extraction
  • Research
  • Web Crawling
  • Web Search

Overview

Tavily MCP Server gives compatible AI applications a route to current web information and website content. The README describes separate capabilities for searching the web, extracting page content, mapping a site's URLs, and crawling through its pages. The supplied tool data also identifies a research operation that combines multiple sources into a detailed response. Together these functions cover both finding relevant material and collecting the pages needed to examine it, rather than limiting an assistant to search snippets alone.

Users can run the npm package locally or connect to Tavily's hosted MCP endpoint. Local execution requires Node.js version 20 or newer and a Tavily API key, with client examples for Claude Desktop and Cursor. Remote access supports an API key in the URL or Authorization header, and compatible clients can instead complete an OAuth flow. Default search behavior can be supplied through a local environment variable, while the hosted service accepts a defaults header. The documentation also describes an optional end-user identifier forwarded for per-user analytics, with server-side hashing before storage and a recommendation to avoid raw personal information. This is a practical choice for assistants that need a configurable web retrieval layer, provided the deployment accounts for authentication and the research tool's documented request limit.

Install & configure

Requirements

  • Node.js 20 or newer for local execution
  • A Tavily API key for the local server
  • An MCP-compatible client
Documented launch / install command
npx -y tavily-mcp@latest

The command launches the local server. The hosted endpoint is an alternative with API-key or OAuth authentication. DEFAULT_PARAMETERS can set local search defaults.

Config file · macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

Windows: %APPDATA%\Claude\claude_desktop_config.json

Use Settings → Developer → Edit Config when available. Do not assume a Linux configuration path; check the client documentation for your platform. Remote servers use the app’s connector flow when supported, not the local stdio file shown here. Merge this example with existing settings; do not overwrite unrelated servers.

Claude Desktop configuration
{
  "mcpServers": {
    "tavily-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "tavily-mcp@latest"
      ],
      "env": {
        "TAVILY_API_KEY": "<Tavily API key>"
      }
    }
  }
}

Read the Claude Desktop setup guide

Config file · Project: .cursor/mcp.json

User: ~/.cursor/mcp.json

The project path is relative to the workspace root. The tilde denotes your user home directory. Use Cursor’s MCP settings to confirm which configuration and scope your installed version is reading. Merge this example with existing settings; do not overwrite unrelated servers.

Cursor configuration
{
  "mcpServers": {
    "tavily-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "tavily-mcp@latest"
      ],
      "env": {
        "TAVILY_API_KEY": "<Tavily API key>"
      }
    }
  }
}

Read the Cursor setup guide

Config file · Workspace: .vscode/mcp.json

For user scope, use the “MCP: Open User Configuration” command if available. Do not guess a user configuration path across operating systems, profiles, containers, or remote workspaces; check the client documentation. Merge this example with existing settings; do not overwrite unrelated servers.

VS Code configuration
{
  "servers": {
    "tavily-mcp": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "tavily-mcp@latest"
      ],
      "env": {
        "TAVILY_API_KEY": "<Tavily API key>"
      }
    }
  }
}

Read the VS Code setup guide

Config file · macOS/Linux: ~/.codeium/windsurf/mcp_config.json

Windows: %USERPROFILE%\.codeium\windsurf\mcp_config.json

Open the MCP configuration through Windsurf’s settings when possible to confirm the active file. Paths can differ in managed or future releases; check the client documentation if this location is not present. Merge this example with existing settings; do not overwrite unrelated servers.

Windsurf configuration
{
  "mcpServers": {
    "tavily-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "tavily-mcp@latest"
      ],
      "env": {
        "TAVILY_API_KEY": "<Tavily API key>"
      }
    }
  }
}

Read the Windsurf setup guide

Config file · Project scope: .mcp.json at the repository root

The project file uses mcpServers. For user scope, use claude mcp add --scope user rather than assuming .mcp.json in your home directory is the user configuration. Check the client documentation for managed settings and storage details. Merge this example with existing settings; do not overwrite unrelated servers.

Claude Code configuration
{
  "mcpServers": {
    "tavily-mcp": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "tavily-mcp@latest"
      ],
      "env": {
        "TAVILY_API_KEY": "<Tavily API key>"
      }
    }
  }
}

Read the Claude Code setup guide

Examples are translated from the published configuration. Supply your own credentials and verify requirements in the upstream README. Remote connectors are shown only where a supported configuration format is known.

Tools

Source: repository tool metadata. Tool descriptions are a documented snapshot, not a live capability check.

ToolWhat it does
tavily_searchFind current web information and return source links with excerpts.
tavily_extractRetrieve page content from URLs as Markdown or text.
tavily_crawlExplore site pages with adjustable crawl depth and breadth.
tavily_mapDiscover URLs and outline the structure of a website.
tavily_researchCombine multiple sources into a detailed research response; limited to 20 requests per minute.

Names, permissions, and defaults can change. Check the current upstream documentation before granting access.

Use cases

  • Find current web sources for an assistant response.
  • Extract page content as Markdown or plain text.
  • Discover a website URL structure before collecting pages.
  • Crawl websites or assemble research from multiple sources.

Strengths

  • Combines discovery, extraction, mapping, and crawling.
  • Supports hosted access as well as a local package.
  • Offers OAuth alongside API-key authentication for the hosted service.

Limitations

  • Local execution requires Node.js 20 or newer and a Tavily API key.
  • The research tool is limited to 20 requests per minute in the supplied data.

Frequently asked questions

Yes. The README provides a hosted MCP endpoint and remote configuration examples.

It accepts an API key through a URL parameter or Authorization header, and also supports OAuth with compatible clients.

The README specifies Node.js version 20 or higher.

Yes. DEFAULT_PARAMETERS configures local defaults, and a header of that name can carry defaults to the remote service.

Independent listing, not an affiliation or security review. Grant the least access needed and verify the upstream instructions before running code.

Related MCP servers

Firecrawl MCP Server

Search & Web Data

Firecrawl MCP connects AI agents to web search, scraping, crawling, and live browser interaction. It offers hosted access or a local server for collecting structured web context and running research tasks.

nodestdio7,615 GitHub stars

AgentQL MCP Server

Search & Web Data

AgentQL MCP Server extracts structured JSON from a web page using a natural-language description of the desired fields. It provides one focused extraction tool for MCP assistants.

nodestdio181 GitHub stars

Exa MCP Server

Search & Web Data

Exa MCP Server gives AI assistants web search and readable page retrieval, with optional advanced search and agent research tools. It supports hosted connections and a local npm package for research and coding context.

nodestdio5,138 GitHub stars

Perplexity API Platform MCP Server

Search & Web Data

The Perplexity API Platform MCP Server gives assistants web search, conversational answers, research, and reasoning tools. It connects to Perplexity’s Search API and Sonar models through local stdio or an HTTP deployment.

nodestdio2,573 GitHub stars

ArXiv MCP Server

Search & Web Data

Search arXiv, download papers, and read their contents through an AI assistant. Local storage and research prompts support repeatable literature review and paper analysis.

pythonstdio3,221 GitHub stars

Bright Data Web MCP

Search & Web Data

Give AI assistants live web search, page extraction, and optional browser automation through Bright Data. Tool groups also provide structured datasets, package information, and AI brand-visibility queries.

nodestdio2,678 GitHub stars

DuckDuckGo Search MCP Server

Search & Web Data

Give MCP assistants DuckDuckGo search and webpage text retrieval. It suits research workflows that need regional results, readable page content, and configurable filtering.

pythonstdio1,538 GitHub stars

Yargı MCP

Search & Web Data

Search Turkish court and regulatory decisions from MCP-compatible assistants. Yargı MCP retrieves decision texts as Markdown and offers both local execution and a hosted streamable HTTP endpoint.

pythonstdio1,007 GitHub stars