mcp-tavily-search

View original on GitHub 

A Model Context Protocol (MCP) server for integrating Tavily’s search API with LLMs. This server provides intelligent web search capabilities optimized for high-quality, factual results, including context generation for RAG applications and direct question answering.

Features

• 🔍 Advanced web search capabilities through Tavily API
• 🤖 AI-generated summaries of search results
• 🎯 Domain filtering for higher quality results
• 📊 Configurable search depth and parameters
• 🧠 Context generation for RAG applications
• ❓ Direct question answering capabilities
• 💾 Response caching with TTL support
• 📝 Multiple response formats (text, JSON, markdown)
• 🔄 Structured result formatting optimized for LLMs
• 🏗️ Built on the Model Context Protocol

Configuration

This server requires configuration through your MCP client. Here are examples for different environments:

Cline Configuration

Add this to your Cline MCP settings:

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

Claude Desktop with WSL Configuration

For WSL environments, add this to your Claude Desktop configuration:

{
	"mcpServers": {
		"mcp-tavily-search": {
			"command": "wsl.exe",
			"args": [
				"bash",
				"-c",
				"source ~/.nvm/nvm.sh && TAVILY_API_KEY=your-tavily-api-key /home/username/.nvm/versions/node/v20.12.1/bin/npx mcp-tavily-search"
			]
		}
	}
}

Environment Variables

The server requires the following environment variable:

• TAVILY_API_KEY: Your Tavily API key (required)

API

The server implements three MCP tools with configurable parameters:

tavily_search

Search the web using Tavily Search API, optimized for high-quality, factual results.

Parameters:

• query (string, required): Search query
• search_depth (string, optional): “basic” (faster) or “advanced” (more thorough). Defaults to “basic”
• topic (string, optional): “general” or “news”. Defaults to “general”
• days (number, optional): Number of days back to search (news topic only). Defaults to 3
• time_range (string, optional): Time range for results (‘day’, ‘week’, ‘month’, ‘year’ or ‘d’, ‘w’, ‘m’, ‘y’)
• max_results (number, optional): Maximum number of results. Defaults to 5
• include_answer (boolean, optional): Include AI-generated summary. Defaults to true
• include_images (boolean, optional): Include related images. Defaults to false
• include_image_descriptions (boolean, optional): Include image descriptions. Defaults to false
• include_raw_content (boolean, optional): Include raw HTML content. Defaults to false
• include_domains (string[], optional): List of trusted domains to include
• exclude_domains (string[], optional): List of domains to exclude
• response_format (string, optional): ‘text’, ‘json’, or ‘markdown’. Defaults to ‘text’
• cache_ttl (number, optional): Cache time-to-live in seconds. Defaults to 3600
• force_refresh (boolean, optional): Force fresh results ignoring cache. Defaults to false

tavily_get_search_context

Generate context for RAG applications using Tavily search.

Parameters:

• query (string, required): Search query for context generation
• max_tokens (number, optional): Maximum length of generated context. Defaults to 2000
• search_depth (string, optional): “basic” or “advanced”. Defaults to “advanced”
• topic (string, optional): “general” or “news”. Defaults to “general”
• Other parameters same as tavily_search

tavily_qna_search

Get direct answers to questions using Tavily search.

Parameters:

• query (string, required): Question to be answered
• include_sources (boolean, optional): Include source citations. Defaults to true
• search_depth (string, optional): “basic” or “advanced”. Defaults to “advanced”
• topic (string, optional): “general” or “news”. Defaults to “general”
• Other parameters same as tavily_search

Domain Filtering

The server supports flexible domain filtering through two optional parameters:

• include_domains: Array of trusted domains to include in search results
• exclude_domains: Array of domains to exclude from search results

This allows you to:

• Target specific trusted sources for academic or technical searches
• Exclude potentially unreliable or irrelevant sources
• Customize sources based on your specific needs
• Access all available sources when no filtering is specified

Example domain filtering:

{
	"include_domains": ["arxiv.org", "science.gov"],
	"exclude_domains": ["example.com"]
}

Development

Setup

1. Clone the repository
2. Install dependencies:

pnpm install

3. Build the project:

pnpm build

4. Run in development mode:

pnpm dev

Publishing

The project uses changesets for version management. To publish:

1. Create a changeset:

pnpm changeset

2. Version the package:

pnpm changeset version

3. Publish to npm:

pnpm release

Contributing

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

License

MIT License - see the LICENSE file for details.

Acknowledgments

• Built on the Model Context Protocol 
• Powered by Tavily Search API