tavily

The Tavily MCP Server adds AI-powered web search capabilities to your MCP client. It exposes a single tool called search that lets you perform Tavily-powered queries and fetch rich results, including titles, URLs, content snippets, and AI-generated summaries. Results are cached and stored persistently to improve response times on repeated queries. The server also exposes static and dynamic resources so you can retrieve recent or specific search results via resource URIs, enabling flexible data access beyond the tool-based workflow.

Prerequisites:

• Node.js v16 or higher
• npm (Node Package Manager)
• Tavily API key
• An MCP client (e.g., Cline, Claude Desktop, or your own integration)

Installation steps:

# 1. Clone the repository
git clone https://github.com/it-beard/tavily-server.git
cd tavily-mcp-server

# 2. Install dependencies
npm install

# 3. Build the project
npm run build

Run the server (after building):

# 4. Start the MCP server
npm run start

Note: Ensure you have your Tavily API key ready and configure the environment variable TAVILY_API_KEY when starting the server or via your MCP client configuration.

Tips:

• Environment variables: TAVILY_API_KEY is required for authenticating with the Tavily API. Ensure it is set in the environment or in your MCP client configuration file.
• Data persistence: The server stores search history and cached results in the data directory. This enables quick retrieval after restarts but ensure disk space is sufficient for growth.
• Resources vs Tools: Use the search tool to perform new queries, and use static/dynamic resources (e.g., tavily://last-search/result or tavily://search/{query}) to access cached or freshly computed results.
• Error handling: Common issues include invalid API keys, network errors, or rate limiting from Tavily. The server surfaces descriptive errors to help diagnose problems quickly.