Skip to Content

Tavily

Expands Tavily search, previously available as Built-in Web Search in Chat, to an MCP Integration—enabling real-time web search results through Skills even in the Sandbox environment.

Basic Information

Authentication

No authentication required. Automatically handled by QueryPie AIP.

Key Features

  • Search the latest web information in real time and incorporate it into AI responses
  • Search only from selected sites or exclude specific sites
  • Search for recent information by specifying a time range (today, this week, this month, etc.)
  • Get search results optimized for specific domains such as finance and news
  • Read and analyze the content of specific web pages with AI

Supported MCP Tools

Performs real-time web search and returns highly relevant results. Supports advanced search depth, time range filters, and image search, providing results with clear citations beyond simple URLs/snippets.

Input Parameters:

  • query (required): Search query
  • searchDepth (optional): Search depth (default: basic)
    • basic: For simple questions needing quick answers
    • advanced: For complex topics, rare information, or in-depth analysis
  • includeDomains (optional): List of domains to include in results (e.g., ["nytimes.com", "bbc.com"])
  • excludeDomains (optional): List of domains to exclude from results
  • includeImages (optional): Whether to include related images (default: false)
  • timeRange (optional): Content publication time range — day (24h), week (7d), month (30d), year (365d). Set only when user explicitly mentions a time period
  • topic (optional): Search category
    • general: General search (default)
    • finance: Markets, investments, economic data, financial news
    • news: Politics, sports, major current events

extract

Extracts raw text from web pages at specified URLs. Supports comprehensive content extraction including tables and embedded elements.

Input Parameters:

  • urls (required): List of URLs to extract
  • extractDepth (optional): Extraction depth (default: basic)
    • basic: Fast extraction of main text content
    • advanced: In-depth extraction including tables and embedded elements
  • includeImages (optional): Whether to include extracted images (default: false)

Usage Examples

  • A user enters the prompt “Search for today’s top AI news”.
  • The LLM calls the search tool via MCP to perform a real-time web search.
  • It then responds in chat based on the tool call results.

Other usage examples:

  • “Find interest rate trends over the past month”
  • “Read the content of this URL: https://example.com/article 
  • “Search only on reuters.com for the latest economic news”

Notes

  • timeRange is only set when the user explicitly mentions a time period.
  • advanced search is more accurate than basic but may take longer to respond.
Last updated on