> This page is for Workforce.

> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.merge.dev/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.merge.dev/_mcp/server.

# Tavily

**Authentication:** API key. See [Magic Link](/merge-agent-handler/features/magic-link) for the runtime auth flow.

## Sample use cases

* Search the web for coverage of our Series B announcement from the past week and summarize the top five articles.
* Crawl `https://docs.acme.com` and list every page that mentions rate limits.
* Extract the main content from these three competitor pricing pages and compare their tiers.

## Available Tools

#### get_usage

Read current Tavily key and account credit usage and limits. Optionally scope to a project ID. Use this before expens...

#### list_logs

Read recent per-request usage logs with date, endpoint, project, and API-key filters. Requires a paid Tavily plan or ...

#### get_organization_usage

Get aggregated Tavily usage for an organization you own. Requires the owner's personal API key and exact organization...

#### submit_feedback

Submit actual quality feedback for a Tavily search request or session. Use the request ID from search_web or a known ...

#### create_research

Start an asynchronous cited research task for synthesis beyond search_web. Returns a request ID; use get_research to ...

#### get_research

Get status, cited sources, content, and usage for a research task. Use the request ID returned by create_research. Ca...

#### search_web

Search the live web when source URLs are unknown. Basic depth costs 1 Tavily credit; advanced costs 2. Returns ranked...

#### extract_content

Extract full or query-focused content from 1–20 chosen URLs. Use search_web first when URLs are unknown. Basic costs ...

#### map_site

Discover URLs and structure on a site without extracting page content. Use before crawl_site to identify relevant pat...

#### crawl_site

Explore a site and extract page content when many pages are needed. Use map_site first to choose paths; set a reasona...