MCP Tavily

中文文档

A Model Context Protocol (MCP) server implementation for Tavily API, providing advanced search and content extraction capabilities.

Features

Multiple Search Tools:
- search: Basic search functionality with customizable options
- searchContext: Context-aware search for better relevance
- searchQNA: Question and answer focused search
Content Extraction: Extract content from URLs with configurable options
Rich Configuration Options: Extensive options for search depth, filtering, and content inclusion

Usage with MCP

Add the Tavily MCP server to your MCP configuration:

json
{
  "mcpServers": {
    "tavily": {
      "command": "npx",
      "args": ["-y", "@mcptools/mcp-tavily"],
      "env": {
        "TAVILY_API_KEY": "your-api-key"
      }
    }
  }
}

Note: Make sure to replace your-api-key with your actual Tavily API key. You can also set it as an environment variable TAVILY_API_KEY before running the server.

API Reference

Search Tools

The server provides three search tools that can be called through MCP:

1. Basic Search

typescript
// Tool name: search
{
  query: "artificial intelligence",
  options: {
    searchDepth: "advanced",
    topic: "news",
    maxResults: 10
  }
}

2. Context Search

typescript
// Tool name: searchContext
{
  query: "latest developments in AI",
  options: {
    topic: "news",
    timeRange: "week"
  }
}

3. Q&A Search

typescript
// Tool name: searchQNA
{
  query: "What is quantum computing?",
  options: {
    includeAnswer: true,
    maxResults: 5
  }
}

Extract Tool

typescript
// Tool name: extract
{
  urls: ["https://example.com/article1", "https://example.com/article2"],
  options: {
    extractDepth: "advanced",
    includeImages: true
  }
}

Search Options

All search tools share these options:

typescript
interface SearchOptions {
  searchDepth?: "basic" | "advanced";    // Search depth level
  topic?: "general" | "news" | "finance"; // Search topic category
  days?: number;                         // Number of days to search
  maxResults?: number;                   // Maximum number of results
  includeImages?: boolean;               // Include images in results
  includeImageDescriptions?: boolean;    // Include image descriptions
  includeAnswer?: boolean;               // Include answer in results
  includeRawContent?: boolean;           // Include raw content
  includeDomains?: string[];            // List of domains to include
  excludeDomains?: string[];            // List of domains to exclude
  maxTokens?: number;                    // Maximum number of tokens
  timeRange?: "year" | "month" | "week" | "day" | "y" | "m" | "w" | "d"; // Time range for search
}

Extract Options

typescript
interface ExtractOptions {
  extractDepth?: "basic" | "advanced";   // Extraction depth level
  includeImages?: boolean;               // Include images in results
}

Response Format

All tools return responses in the following format:

typescript
{
  content: Array<{
    type: "text",
    text: string
  }>
}

For search results, each item includes:

Title
Content
URL

For extracted content, each item includes:

URL
Raw content
Failed URLs list (if any)

Error Handling

All tools include proper error handling and will throw descriptive error messages if something goes wrong.

Installation

Installing via Smithery

To install Tavily API Server for Claude Desktop automatically via Smithery:

bash
npx -y @smithery/cli install @kshern/mcp-tavily --client claude

Manual Installation

bash
npm install @mcptools/mcp-tavily

Or use it directly with npx:

bash
npx @mcptools/mcp-tavily

Prerequisites

Node.js 16 or higher
npm or yarn
Tavily API key (get one from Tavily)

Setup

Clone the repository
Install dependencies:

bash
npm install

Set your Tavily API key:

bash
export TAVILY_API_KEY=your_api_key

Building

bash
npm run build

Debugging with MCP Inspector

For development and debugging, we recommend using MCP Inspector, a powerful development tool for MCP servers.

The Inspector provides a user interface for:

Testing tool calls
Viewing server responses
Debugging tool execution
Monitoring server state

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

Fork the repository
Create your feature branch (git checkout -b feature/AmazingFeature)
Commit your changes (git commit -m 'Add some AmazingFeature')
Push to the branch (git push origin feature/AmazingFeature)
Open a Pull Request

License

This project is licensed under the MIT License.

Support

For any questions or issues:

Tavily API: refer to the Tavily documentation
MCP integration: refer to the MCP documentation

@kshern/mcp-tavily: MCP Tavily

Author

README