Tavily Block

The Tavily block taps into Tavily's AI-powered search engine to find relevant information from across the web, and can extract and process content from specific URLs with customizable depth. Reach for it when a workflow needs fresh, web-sourced facts or the readable content behind a known link.

Overview

When to Use

• Run an AI-optimized web search and feed the results or synthesized answer into a downstream agent.
• Gather up-to-date information that an LLM's training data would not include.
• Extract the clean, readable content from a specific URL for summarization or analysis.
• Pull source titles and URLs to cite or link back to in generated output.
• Choose between basic and advanced extraction depth depending on how thorough the content parse needs to be.
• Process multiple URLs in a single Extract Content call and receive structured error reports for any that fail.

Configuration

Operation

Dropdown selecting the action to perform. Defaults to Search.

Search Query

Long text input for the search terms sent to Tavily. Required when the operation is tavily_search. Supports {{blockName.field}} references from earlier blocks.

Max Results

Short input limiting the number of search results returned (1–20). Applies to tavily_search only. Optional; Tavily defaults to 5 when omitted.

URL

Long text input for the URL (or URLs) to extract content from. Required when the operation is tavily_extract. A single URL string is accepted; Tavily's API also accepts arrays. Supports {{blockName.field}} references.

Extract Depth

Dropdown controlling how thoroughly Tavily parses the page. Applies to tavily_extract only. Defaults to basic.

API Key

Required, password-protected field for your Tavily API key. Use {{TAVILY_API_KEY}} to reference an environment secret.

Inputs & Outputs

• Inputs:

  • operation (string) — Operation to perform (tavily_search or tavily_extract)
  • apiKey (string) — Tavily API key
  • query (string) — Search query terms (Search only)
  • maxResults (number) — Maximum number of search results (Search only)
  • urls (string) — URL to extract content from (Extract Content only)
  • extract_depth (string) — Extraction depth level: basic or advanced (Extract Content only)

• Outputs:

  • results (json) — Search results data (Search) or successfully extracted content per URL (Extract)
  • answer (string) — Synthesized search answer returned by Tavily (Search only)
  • query (string) — The query string that was executed (Search only)
  • content (string) — Extracted page content (Extract only)
  • title (string) — Page title of the extracted URL (Extract only)
  • url (string) — Source URL (Extract only)

Tools

Tavily Search (tavily_search) — Performs an AI-powered web search via POST https://api.tavily.com/search. Returns an array of results (each with title, url, snippet, and optional raw_content), the echoed query, and response_time. Key params: query (required), max_results (optional, 1–20), apiKey (required).

Tavily Extract (tavily_extract) — Extracts raw text content from one or more web pages via POST https://api.tavily.com/extract. Returns a results array (each with url and raw_content), an optional failed_results array (each with url and error), and response_time. Key params: urls (required — string or array of strings), extract_depth (optional, basic or advanced), apiKey (required).

YAML Example

tavily_1:
  type: tavily
  name: "Tavily"
  inputs:
    operation: "tavily_search"
    query: "{{trigger.userQuery}}"
    maxResults: 5
    apiKey: "{{TAVILY_API_KEY}}"
  connections:
    outgoing:
      - target: agent-1