tavily-streamable-http

Tavily Streamable HTTP MCP Server implements the Model Context Protocol (MCP) on Cloudflare Workers to provide fast network search and content extraction. It exposes MCP-compliant endpoints that let clients request tools such as get_current_time, tavily_search, and tavily_extract via MCP calls on /mcp. The server is designed to work with MCP clients like LobeChat, Claude Desktop, and other MCP-enabled integrations, delivering streamable HTTP responses with search results, page content, and time information. To use it, deploy the Cloudflare Worker (via the Cloudflare dashboard or Wrangler CLI), configure your Tavily API key as an environment variable, and then send MCP-formatted requests to the /mcp endpoint.

What you can do with this MCP server:

• get_current_time: fetch current date/time in various time zones and formats.
• tavily_search: execute Tavily-powered web searches with options for depth, result count, and domain filtering.
• tavily_extract: extract content from one or more URLs. These capabilities enable seamless integration with AI chat clients that rely on MCP for dynamic information gathering and content retrieval. The endpoints are exposed via the MCP protocol, so you can embed them into other MCP clients or automation pipelines. Use the health endpoint for basic availability checks and the root endpoint for a simple service information page.

Prerequisites:\n- Node.js (LTS) and npm installed on your development machine.\n- Cloudflare account with access to Workers or Wrangler toolchain.\n- Access to a Tavily API key (required by the server).\n\nInstallation steps:\n1) Install Wrangler globally (optional if you deploy via Cloudflare dashboard):\nbash\nnpm install -g wrangler\n\n\n2) Clone the repository (or download the source):\nbash\ngit clone <repository-url>\ncd tavily-mcp-server\n\n\n3) Install dependencies (if present):\nbash\nnpm install\n\n\n4) Configure environment variable for Tavily API key:\nbash\nwrangler secret put TAVILY_API_KEY\n. When using the Cloudflare dashboard, set the environment variable TAVILY_API_KEY under Worker Settings > Variables.\n\n5) Deploy:\nbash\nwrangler deploy\n \nIf you are using the Cloudflare Dashboard deployment flow, simply copy the worker.js code into the online editor and Save & Deploy, then set the TAVILY_API_KEY in the environment variables before deploying.

Tips and notes:\n- The server runs as a Cloudflare Worker; ensure you have the correct account and permissions to deploy Workers.\n- Set TAVILY_API_KEY as a single or comma-separated list if you have multiple Tavily API keys for rotation. The server will use them as configured.\n- The MCP endpoint is /mcp; health checks are available at /health and service info at /.\n- For local testing with Wrangler, you can use wrangler dev to simulate the worker environment.\n- If you encounter authorization issues, verify that the environment variable is correctly configured and that the API keys are valid and active.\n- Logging in Cloudflare Dashboard helps diagnose issues in production; use wrangler tail for local or remote log streaming.