MCP ConfigsApr 6, 2026·2 min read

Tavily — Search API Built for AI Agents & RAG

Search API designed specifically for AI agents and RAG pipelines. Returns clean, LLM-ready results with content extraction, no HTML parsing needed. Official MCP server available. 5,000+ stars.

MC
MCP Hub · Community
Quick Use

Use it first, then decide how deep to go

This block should tell both the user and the agent what to copy, install, and apply first.

Add to your .mcp.json:

{
  "mcpServers": {
    "tavily": {
      "command": "npx",
      "args": ["-y", "tavily-mcp"],
      "env": {
        "TAVILY_API_KEY": "tvly-your-key"
      }
    }
  }
}

Get your free API key at tavily.com. Restart Claude Code and ask: "Search the web for the latest MCP server releases."


Intro

Tavily is a search API built specifically for AI agents and RAG pipelines with 5,000+ GitHub stars. Unlike Google or Bing APIs that return HTML snippets, Tavily returns clean, LLM-ready text with source attribution — no scraping or parsing needed. It includes an official MCP server, Python/JS SDKs, and integrations with LangChain, LlamaIndex, and CrewAI. Best for developers building AI agents that need real-time web knowledge. Works with: Claude Code, Cursor, LangChain, CrewAI, any MCP client. Setup time: under 1 minute.


Why Tavily Over Google API

Feature Google Search API Tavily
Output HTML snippets Clean text, LLM-ready
Content extraction Manual scraping needed Built-in
Source quality All results Filtered for relevance
Token efficiency Wasteful (HTML tags) Optimized for LLMs
Pricing Per query 1,000 free/month

MCP Tools Available

tavily_search

"Search for best practices for deploying Claude Code in enterprise"

Returns: clean text results with URLs and relevance scores.

tavily_extract

"Extract the main content from https://docs.anthropic.com/en/docs/agents"

Returns: clean markdown from any URL.

Python SDK

from tavily import TavilyClient

client = TavilyClient(api_key="tvly-...")

# Basic search
results = client.search("latest AI agent frameworks 2026")

# Advanced search with content extraction
results = client.search(
    query="Claude Code best practices",
    search_depth="advanced",
    include_raw_content=True,
    max_results=5
)

for r in results["results"]:
    print(f"{r['title']}: {r['content'][:200]}")

Integration with Agent Frameworks

# LangChain
from langchain_community.tools import TavilySearchResults
tool = TavilySearchResults(max_results=5)

# CrewAI
from crewai_tools import TavilySearchTool
search = TavilySearchTool()

Key Stats

  • 5,000+ GitHub stars
  • 1,000 free searches/month
  • LLM-optimized output
  • MCP server + Python/JS SDKs
  • LangChain, CrewAI integrations

FAQ

Q: What is Tavily? A: Tavily is a search API built for AI agents that returns clean, LLM-ready text results instead of HTML, with built-in content extraction and source attribution.

Q: Is Tavily free? A: Free tier includes 1,000 searches/month. Paid plans for higher volume.

Q: How is Tavily different from Firecrawl? A: Tavily is a search engine — it finds relevant pages. Firecrawl is a scraper — it extracts content from URLs you specify. They complement each other.


🙏

Source & Thanks

Created by Tavily. Licensed under MIT.

tavily — ⭐ 5,000+

Thanks for building the search engine AI agents actually need.

Discussion

Sign in to join the discussion.
No comments yet. Be the first to share your thoughts.

Related Assets