What it does

Web search built for agents rather than people: results come back with extracted content and optional answer synthesis, which removes a second fetch step. Requires an API key and has a free tier that runs out fast on research workloads. Better than scraping for synthesis, worse for exact-match debugging.

Install

Terminal

npx -y tavily-mcp

Client configuration

Paste this into your MCP client config (Claude Desktop, Cursor, or any client that reads mcpServers), then restart the client.

mcp.json

{
  "mcpServers": {
    "tavily": {
      "command": "npx",
      "args": [
        "-y",
        "tavily-mcp"
      ]
    }
  }
}

Use it for

Research agents that need search results plus extracted page content

Skip it if

Anything that must work without a paid API key

When to use it

Reach for this when the task needs search inside the agent loop. Skip it if a plain shell command or a one-off script is enough — every server you add costs context and adds a failure mode.

Other TypeScript servers

Filesystem — Gives an agent scoped read, write, and search access to directories you explicitly allow
Memory — A persistent knowledge graph stored in a local JSON file, exposed through entities, relations, and observations
Sequential Thinking — Gives the model a scratchpad for structured, revisable reasoning: numbered thoughts that can branch, revise, and conclude

Tavily MCP