Tavily Search

Tavily Search is a specialized AI search API designed for LLMs and RAG applications. It provides:

• Real-time web search with AI-summarized results
• Built-in content extraction
• Date range filtering
• Domain filtering
• Free tier: 1000 searches/month

Configuration

Before using, you need to configure your Tavily API key:

1. Get your API key from https://tavily.com/
2. Set it in your environment: export TAVILY_API_KEY=your_api_key_here
3. Or add it to OpenClaw config at tools.tavily.apiKey

Core Capabilities

1. Basic Search (tavily_search)

Search the web with optional depth control and topic filtering.

Parameters:

• query (required): Search query string
• max_results (optional, default 5): Number of results (1-10)
• search_depth (optional): basic or deep (default basic)
• topic (optional): general or news (default general)
• days (optional): Time range in days back from current date

2. Search with Date Range (tavily_search_date_range)

Search with specific start and end dates.

Parameters:

• query (required): Search query string
• start_date (required): YYYY-MM-DD
• end_date (required): YYYY-MM-DD
• max_results (optional, default 5)
• search_depth (optional): basic or deep

3. Q&A Search (tavily_answer)

Get a direct AI answer to your question with citations.

Parameters:

• query (required): Question to answer
• max_results (optional, default 5)
• search_depth (optional): basic or deep

4. Extract Content (tavily_extract)

Extract clean, readable content from one or more URLs.

Parameters:

• urls (required): Array of URLs to extract

Usage Example

# Search for recent news about OpenClaw
python scripts/tavily_search.py "openclaw github" --days 1 --max_results 5

Output Format

All search commands return JSON with:

• results: Array of search results with title, content, url, score
• answer: AI-generated answer (for answer mode)
• response_time: Query response time

scripts/

references/