MCP server, by Tavily
Tavily MCP Server
Production ready MCP server with real-time search, extract, map, and crawl tools powered by Tavily.
Quick facts
What it does
The Tavily MCP server provides search, extract, map, and crawl tools, with real-time web search capabilities through the tavily-search tool, intelligent data extraction from web pages via the tavily-extract tool, a web mapping tool that creates a structured map of a website, and a web crawler that systematically explores websites, per the README.
The README documents a remote MCP server at https://mcp.tavily.com/mcp/ using a Tavily API key in the URL query parameter or an Authorization Bearer header, with optional OAuth authentication, and a local MCP run with npx -y tavily-mcp@latest requiring Node.js v20 or higher and a Tavily API key. Default parameters for the tavily-search tool can be set with the DEFAULT_PARAMETERS environment variable.
Configuration
Remote: use https://mcp.tavily.com/mcp/?tavilyApiKey=<your-api-key> or OAuth via claude mcp add --transport http tavily https://mcp.tavily.com/mcp. Local: npx -y tavily-mcp@latest with TAVILY_API_KEY env in the client config, per README examples.
{"mcpServers":{"tavily-mcp":{"command":"npx","args":["-y","tavily-mcp@latest"],"env":{"TAVILY_API_KEY":"your-api-key-here"}}}}Source: https://github.com/tavily-ai/tavily-mcp. Placeholders only. Replace them in your own environment, and never commit real keys to a repository.
| Environment variable | Required | Purpose |
|---|---|---|
TAVILY_API_KEY | Yes | Tavily API key, required for local NPX configuration per README examples and for remote use via URL parameter or Authorization header. |
DEFAULT_PARAMETERS | Optional | JSON object of default parameter values for the tavily-search tool, per README Default Parameters Configuration. |
TAVILY_HUMAN_ID | Optional | Optional end-user identifier forwarded as the X-Human-Id header on every API call for per-user analytics, per README. |
Verified tools and features
tavily-search Config checked
Real-time web search capabilities, named as the tavily-search tool in the README.
tavily-extract Config checked
Intelligent data extraction from web pages, named as the tavily-extract tool in the README.
Web mapping tool Config checked
Creates a structured map of a website, listed among the Tavily MCP server tools in the README. Exact tool identifier was not stated in the fetched README text.
Web crawler Config checked
Systematically explores websites, listed among the Tavily MCP server tools in the README. Exact tool identifier was not stated in the fetched README text.
Security notes
- Remote API key can be passed in the URL query parameter tavilyApiKey or via Authorization Bearer header if the client supports it, per README. Prefer header or OAuth where supported to avoid exposing keys in URLs.
- Privacy note in README: Tavily hashes human_id server-side with SHA-256 before storage so the raw value is never persisted. Prefer opaque identifiers over raw PII like emails for TAVILY_HUMAN_ID.
An MCP server runs with the permissions of the process you start it from. Scope credentials to the narrowest access the job needs, and review what the client will send before connecting anything sensitive.
Sources and freshness
- Official source: https://github.com/tavily-ai/tavily-mcp
- Repository: https://github.com/tavily-ai/tavily-mcp
Last checked 2026-10-07. Verification label: config checked, which means the configuration and feature list trace to the official source above. It does not mean AgentsUse ran the server. Spotted an error? Send a correction. Back to the MCP server index.