home / mcp / tavily search mcp server

Tavily Search MCP Server

An MCP server implementation that integrates the Tavily Search API, providing optimized search capabilities for LLMs.

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "apappascs-tavily-search-mcp-server": {
      "command": "node",
      "args": [
        "/Users/<username>/<FULL_PATH...>/tavily-search-mcp-server/dist/index.js"
      ],
      "env": {
        "TAVILY_API_KEY": "your_api_key_here"
      }
    }
  }
}

You can deploy the Tavily Search MCP Server to empower your LLM workflows with optimized web search, content extraction, and configurable options. It integrates Tavily’s search capabilities to deliver targeted results, including optional images, descriptions, and raw content when needed.

How to use

Run the Tavily Search MCP Server locally or in a container and connect it to your MCP client. You can choose between running the server as a local process or via Docker. Use the stdio transport for a direct Node.js run, or the SSE transport if your client requires server-sent events.

When you connect your MCP client, you’ll be able to issue search queries like generic web queries, with optional depth, topic, time range, and domain filtering. The server can include image results and short LLM-generated answers if you enable those options.

How to install

Prerequisites you need to prepare before installing include a Tavily API key and a runtime for Node.js or Docker, depending on your chosen deployment method.

Step-by-step installation and setup:

# Clone the MCP server repository
git clone https://github.com/apappascs/tavily-search-mcp-server.git

# Change into the project directory
cd tavily-search-mcp-server

# Install dependencies
npm install

# Build the project
npm run build

Additional notes

Environment variables: you need a Tavily API key. Store it securely and pass it to the server via environment variables as shown in the configuration examples.

Configuration and running options

You have multiple MCP startup options shown for connecting Tavily Search to Claude Desktop. Use the one that matches how you want to run the server.

Run with Node.js (stdio transport) after building the project:

node dist/index.js

Run with Node.js (SSE transport) after building the project:

node dist/sse.js

Run with Docker (stdio transport). Ensure you have built the image and replace the API key appropriately in your environment:

docker build -t tavily-search-mcp-server:latest .

docker run -it --rm -e TAVILY_API_KEY="your_api_key_here" tavily-search-mcp-server:latest

Run with Docker (SSE transport) to expose port 3001 and pass the API key and transport mode:

docker build -t tavily-search-mcp-server:latest .

docker run -it --rm -p 3001:3001 -e TAVILY_API_KEY="your_api_key_here" -e TRANSPORT="sse" tavily-search-mcp-server:latest

Available tools

tavily_search

Executes web searches using the Tavily Search API with configurable depth, topic, days, time_range, and filters. Returns results with optional images, descriptions, and raw content.