tavily

What This Skill Does

The tavily skill provides OpenClaw agents with a high-performance, AI-native web searching capability powered by the Tavily API. Unlike traditional search engines that return cluttered HTML or advertisements, Tavily is specifically engineered to provide structured, clean, and contextually relevant snippets ideal for LLM processing. This skill includes two primary utilities: a comprehensive search tool for answering complex questions and an extraction tool for gathering raw content from specific URLs. It is the go-to tool for any agent requiring up-to-date information, deep research, or event-based data.

Installation

To integrate this skill into your OpenClaw environment, execute the following command in your terminal:

clawhub install openclaw/skills/skills/frontrunnerrrr/tavily-web-searcher

Ensure you have a valid TAVILY_API_KEY obtained from the Tavily developer portal. This key should be configured within your OpenClaw environment variables to allow the agent to authorize requests successfully.

Use Cases

This skill is perfect for scenarios requiring real-time data ingestion. Use it to:

• Stay updated on industry trends using the news topic filter.
• Perform deep research on technical topics by triggering the --deep search mode, which synthesizes multiple sources.
• Gather specific research data by passing URLs to the extraction script to convert messy web pages into clean, readable text strings.
• Automate market sentiment analysis or monitor competitive landscapes by integrating search results into an analytical workflow.

Example Prompts

1. "Perform a deep search on the latest advancements in solid-state battery technology and summarize the top three breakthroughs."
2. "Search for the latest news on global AI regulation from the last 7 days and provide a list of key legislative developments."
3. "Extract the content from https://example.com/tech-article and generate a concise bulleted list of the technical requirements mentioned in the text."

Tips & Limitations

• Efficiency: For simple queries, avoid the --deep flag to save on latency and API credits.
• Limits: The -n flag defaults to 5 results; increase this for broader coverage but be aware of the 20-result maximum.
• News Topic: Remember that the --days parameter is only applicable when using --topic news.
• API Reliability: As this relies on an external API, ensure your environment handles potential network timeouts gracefully within your agent loops.