search

What This Skill Does

The search skill enables your OpenClaw agent to perform intelligent, LLM-optimized web searches using the Tavily API. Unlike standard search engines, this tool is specifically designed to feed relevant context directly into AI workflows. It provides comprehensive results, including page snippets, metadata, and relevance scores, allowing the agent to synthesize information, verify facts, or stay updated on current events without requiring manual intervention or code execution.

Installation

To integrate this skill, ensure you have the OpenClaw environment configured. First, install the skill using the command: clawhub install openclaw/skills/skills/barneyjm/search-2. After installation, you must obtain a valid API key from Tavily. Open your configuration file located at ~/.claude/settings.json and add your credentials under the env block as follows: {"env": {"TAVILY_API_KEY": "tvly-your-api-key-here"}}. Once saved, the agent will automatically utilize this key for all web-related queries.

Use Cases

This skill is highly versatile for any agentic task requiring external data. Common use cases include: summarizing complex technical documentation, researching market trends, retrieving real-time news updates, gathering comparative data for products, or performing deep-dive research into academic papers. By setting specific time_range or include_domains parameters, you can focus the agent's research scope to ensure the output remains relevant to your specific domain needs.

Example Prompts

1. "Search for the latest architectural design patterns for serverless microservices and provide a summary of the pros and cons found in top-tier technical blogs."
2. "Find recent news from the last week regarding advancements in LLM reasoning capabilities and compare the findings from at least three reputable AI research domains."
3. "Look up the current status and documentation for the latest release of the OpenClaw framework; list the most significant breaking changes."

Tips & Limitations

To maximize effectiveness, always prioritize descriptive queries. If the agent returns too much irrelevant data, use the include_domains or exclude_domains parameters to curate the source list. Keep in mind that max_results defaults to 5; for comprehensive reports, feel free to increase this up to 20. Note that this skill requires an active internet connection and a valid API subscription. While the tool is powerful, avoid using it for private or sensitive internal data, as queries are routed through the Tavily search infrastructure.