Skip to main content

​
Search Integration with Tavily

Power your LLM with AI-driven search capabilities using Tavily - a search API optimized for AI agents and LLMs.

​
What It Does

The Tavily integration provides:
  • AI-optimized search - Results tailored for LLM consumption
  • Precise answers - Direct answers extracted from sources
  • Source citations - Full URLs and snippets for verification
  • Image search - Optional image results
  • Customizable depth - Basic or advanced search modes

​
Setup

​
1. Get Your Tavily API Key

  1. Sign up at Tavily.com
  2. Get your API key from the dashboard

​
2. Import the Configuration

The Tavily configuration is included in NUEVOS_HITOS.json: 
{
  "name": "Tavily AI Search",
  "code": "tavily",
  "baseUrl": "https://api.tavily.com",
  "authenticationType": "API_KEY",
  "apiKeyLocation": "IN_BODY",
  "apiKeyName": "api_key",
  "apiKeyValue": "<YOUR_API_KEY>",
  "customHeaders": {
    "Content-Type": "application/json"
  },
  "tools": [
    {
      "name": "Búsqueda Tavily",
      "code": "tavily-search",
      "description": "Realiza una búsqueda impulsada por IA optimizada para obtener respuestas y fuentes precisas.",
      "endpointPath": "/search",
      "httpMethod": "POST",
      "enabled": true,
      "isExportable": true,
      "parameters": [
        {
          "name": "api_key",
          "type": "STRING",
          "description": "Clave API de Tavily (requerida en el cuerpo de la petición).",
          "required": true
        },
        {
          "name": "query",
          "type": "STRING",
          "description": "La consulta de búsqueda a procesar.",
          "required": true
        },
        {
          "name": "search_depth",
          "type": "STRING",
          "description": "Profundidad de búsqueda ('basic' o 'advanced').",
          "required": false,
          "defaultValue": "basic"
        },
        {
          "name": "include_images",
          "type": "BOOLEAN",
          "description": "Si se deben incluir imágenes en los resultados (true/false).",
          "required": false
        }
      ]
    }
  ]
}
Import via API: 
curl -X POST http://localhost:8080/api/import/providers \
  -H "Content-Type: application/json" \
  -d @NUEVOS_HITOS.json
Tavily uses IN_BODY authentication - the API key is sent in the request body, not as a header.

​
Example Queries

User request:
“Search for the latest developments in AI agent frameworks”
LLM tool call: 
{
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "name": "Búsqueda Tavily",
    "arguments": {
      "api_key": "YOUR_API_KEY",
      "query": "latest developments AI agent frameworks 2026",
      "search_depth": "basic"
    }
  },
  "id": "msg_search_1"
}

​
Advanced Search with Images

LLM tool call: 
{
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "name": "Búsqueda Tavily",
    "arguments": {
      "api_key": "YOUR_API_KEY",
      "query": "HandsAI MCP protocol integration",
      "search_depth": "advanced",
      "include_images": true
    }
  },
  "id": "msg_search_2"
}

​
Response Format

Tavily returns structured search results optimized for AI consumption: 
{
  "query": "latest developments AI agent frameworks 2026",
  "answer": "In 2026, AI agent frameworks have seen significant advances including MCP (Model Context Protocol) standardization, improved tool discovery mechanisms, and better integration patterns for REST APIs.",
  "results": [
    {
      "title": "MCP Protocol Adoption in 2026",
      "url": "https://example.com/mcp-2026",
      "content": "The Model Context Protocol (MCP) has become the standard for AI agent communication...",
      "score": 0.95
    },
    {
      "title": "HandsAI: Universal API Bridge",
      "url": "https://github.com/Vrivaans/HandsAI",
      "content": "HandsAI bridges MCP clients with REST APIs dynamically...",
      "score": 0.89
    }
  ],
  "images": [
    {
      "url": "https://example.com/image1.png",
      "description": "MCP architecture diagram"
    }
  ],
  "response_time": 1.23
}

​
Key Features

Tavily preprocesses search results to extract the most relevant information for LLMs, reducing token usage and improving response quality.
The answer field provides a concise summary extracted from the search results, perfect for quick responses.
Every result includes the source URL and relevance score, enabling fact-checking and transparency.
  • Basic: Fast, sufficient for most queries
  • Advanced: Deeper crawl, more comprehensive results

​
Use Cases

Research

Gather information from multiple sources

Fact-Checking

Verify claims with cited sources

News Monitoring

Track latest developments in real-time

Market Research

Analyze trends and competitors

​
Best Practices

1

Be Specific

Use detailed queries for better results: “HandsAI MCP integration” instead of “API tools”
2

Use Search Depth Wisely

Start with basic for speed, upgrade to advanced only when needed
3

Cite Sources

Always include source URLs when presenting information to users
4

Cache Results

For repeated queries, consider caching results to reduce API calls

​
Next Steps

GitHub Integration

Automate GitHub workflows

Jules Agent

Delegate coding tasks to Google’s Jules

Build docs developers (and LLMs) love

Get started for freeTalk to us