tavily

What This Skill Does

The Tavily AI Search skill provides OpenClaw agents with a sophisticated, LLM-optimized web browsing interface. Unlike standard search APIs, Tavily parses search results specifically to minimize noise and maximize utility for Large Language Models. It enables agents to access real-time information, perform deep research, and retrieve clean text extracts from the web. By utilizing advanced search modes, agents can balance speed and depth, ensuring they always have the most relevant information for any given task.

Installation

To integrate this skill into your OpenClaw environment, execute the following command in your terminal: clawhub install openclaw/skills/skills/bert-builder/tavily

Use Cases

• Fact-Checking & Research: Quickly verify claims or gather authoritative data for long-form research papers.
• Current Events Monitoring: Use the 'news' topic filter to summarize breakthroughs or events from the last seven days.
• Technical Documentation: Filter by specific domains (e.g., official docs) to ensure the agent only retrieves high-quality technical code snippets and best practices.
• Data Augmentation: Provide the agent with real-time web content to enrich AI-generated reports and summaries.

Example Prompts

1. "Perform a research query on the impact of decentralized finance on traditional banking, and provide a summary of the pros and cons found in the top 5 search results."
2. "Search for the latest 2026 AI industry trends, restricting results to high-quality tech news outlets."
3. "Find the official documentation for the latest React hook updates, excluding tutorial blogs and forums."

Tips & Limitations

• Depth Settings: For complex inquiries involving multiple sub-topics, always default to 'advanced' mode. While it takes longer, the context window usage is significantly more efficient due to cleaner, more relevant extraction.
• Domain Constraints: Use the include/exclude domain features to eliminate noise. This significantly improves response accuracy when searching for highly technical or professional information.
• Token Management: While Tavily is optimized for LLMs, remember that fetching raw content can increase token usage. Use the summary feature when you need broad overviews rather than raw data.
• Privacy: Note that this skill involves external network requests; ensure your organization's security policy allows for outward-facing API calls.