Home / MCP / Tavily-Search MCP Server
Provides Tavily-powered search via an MCP server with text results including AI responses, URLs, and titles.
Configuration
View docs{
"mcpServers": {
"tavily_search": {
"command": "uv",
"args": [
"--directory",
"C:\\\\your_path\\\\mcp-server-tavily",
"run",
"tavily-search"
],
"env": {
"TAVILY_API_KEY": "YOUR_TAVILY_API_KEY",
"PYTHONIOENCODING": "utf-8"
}
}
}
}This Tavily-powered MCP server lets you perform searches via the Tavily API and receive results in text form, including AI responses, URIs, and titles. You can query it from an MCP client and get structured search outputs to drive your workflows.
You will use an MCP client to interact with the Tavily search server. When you send a search request, you receive a text response that lists relevant results and AI-generated summaries. You can specify a basic search or an advanced search depth if your client supports it.
Prerequisites: ensure you have a modern environment with access to run MCP server commands. You will use a local setup with a standard runtime interface to execute the Tavily search server.
npx -y @smithery/cli install tavily-search --client claudeInstall the Tavily Search MCP server locally by cloning the repository and configuring Claude Desktop to point at the local server.
git clone https://github.com/Tomatio13/mcp-server-tavily.gitOpen the Claude Desktop configuration file and prepare the MCP server entry. The following configuration runs the Tavily search server from a specified directory and provides the necessary environment variable for the Tavily API key.
"mcpServers": {
"tavily-search": {
"command": "uv",
"args": [
"--directory",
"C:\\your_path\\mcp-server-tavily",
"run",
"tavily-search"
],
"env": {
"TAVILY_API_KEY": "YOUR_TAVILY_API_KEY",
"PYTHONIOENCODING": "utf-8"
}
}
}After saving the configuration, restart Claude Desktop to apply the new MCP server entry.
Notes on usage and environment variables are listed below to help you run and troubleshoot the Tavily search MCP server.
Performs searches using the Tavily API. Requires a 'query' string and can optionally accept 'search_depth' to control depth (basic or advanced).