Integration · Research and analyze

Tavily and Meteor AI agents

Connect Tavily to a Met: search the web, extract a page's content, map and crawl a site and build reports with citations.

Book a demo See every integration

How does Tavily connect to Meteor?

With the Tavily ready connector, your Met searches and reads the web with your Tavily account. It has five tools: search current sources when you do not know where the answer is, extract the clean content of a page you already know, map a site's structure, crawl many pages of the same site and produce a research report with citations. Every call spends API credits from your Tavily account; if you have several keys, the one named mcp_auth_default is the one the connection uses.

What it automates

What a Meteor Met does with Tavily

Answer with what a page says

If a customer asks about something published on a specific page, such as a policy or a spec sheet, the Met extracts its content and answers with what it says, with the link.

Review a competitor's site

The Met maps a competitor's site, crawls the pricing or services section and summarizes what they have published.

Build a report with sources

For a question that needs several sources, the Met asks for a research and gives you a comparison with the citations for each point.

5 tools

Everything your Met can ask Tavily for

Each one is a real action in the connector. You can switch off the ones you would rather it never used.

Tool	What it does
`tavily_search`	Searches current web sources when you do not know where the answer is.
`tavily_extract`	Extracts the clean content of one or several pages whose address is already known.
`tavily_map`	Maps a site's structure and lists its pages before crawling it.
`tavily_crawl`	Crawls and reads many pages of the same site starting from an address.
`tavily_research`	Produces a synthesis with citations, such as a report, a comparison or a decision ready answer.

The limits

What the Tavily integration does NOT do

We would rather say it here than have you find out during rollout.

It does not query for free. Every search, extraction, crawl or research spends API credits from your Tavily account, and crawling many pages spends more.
It does not open pages behind a username and password. It reads the public web; whatever lives behind a login does not show up.
It does not save anything anywhere. The five tools read the web and return text; saving the result needs another integration.
It does not guarantee the source tells the truth. The Met shows where it got the fact; checking it before an important decision is still on you.

Before you start

What you need to connect Tavily

1

A Tavily account with credits

A Tavily account with an API key and credits. If you have several keys, name the one you want the connection to use mcp_auth_default, in your account or your team's.
2

Connect it in Met

In Met, open Connections → Tavily → Connect. A Tavily window opens so you sign in and authorize Met.

Frequently asked questions

Questions about Tavily and Meteor

Extracting reads pages whose address you already know. Crawling starts from an address and reads many pages of the same site; mapping first helps avoid crawling too much.

If your account or your team's has a key named mcp_auth_default, the OAuth connection uses that one. That way you control which key spends the credits.

Yes, if it is public. The Met can extract the page where the answer is and reply with what it says, instead of answering from memory.

You do, with your Tavily account credits. Met does not charge for the queries your Met makes to Tavily.

Step by step