tavily-best-practices

What This Skill Does

The tavily-best-practices skill provides a structured framework and implementation patterns for integrating the Tavily Search API into agentic workflows. It is designed to help developers move beyond simple search queries to build robust, production-ready RAG systems and autonomous agents. The skill focuses on optimizing token usage, managing search context, and utilizing advanced features like content extraction, crawling, and research-grade synthesis. It bridges the gap between raw API documentation and practical, agent-first coding patterns, ensuring your integrations are efficient and resilient to the volatility of web data.

Installation

To integrate this skill into your environment, use the OpenClaw CLI:

clawhub install openclaw/skills/skills/barneyjm/tavily-best-practices

Ensure you have your Tavily API key ready. You should add it to your environment variables or your Claude configuration file as follows:

{ "env": { "TAVILY_API_KEY": "tvly-YOUR_API_KEY" } }

Once installed, you can access the reference materials and implementation templates directly within your IDE or agent workspace.

Use Cases

• Autonomous Research Agents: Utilize the research() method to handle multi-step information gathering and synthesis without manual orchestration.
• Optimized RAG Pipelines: Use extract() with semantic query reranking to feed high-quality, relevant context into your LLM, preventing context window bloat.
• Knowledge Base Crawling: Use crawl() to ingest documentation or technical manuals from specific sites, focusing on paths and depth to ensure only relevant content is retrieved.
• Real-time Data Monitoring: Leverage the search() method with 'news' topics to keep agents updated on rapid industry developments.

Example Prompts

1. "Use the tavily-best-practices skill to research current best practices for building scalable RAG pipelines using Tavily and LlamaIndex."
2. "I need to crawl the official documentation site at docs.example.com; show me how to configure the crawler to prioritize API routes and limit content to 3 chunks per page."
3. "Refactor my current web search implementation to use the extract() method instead of basic search, ensuring I only retrieve text from high-relevance URLs."

Tips & Limitations

• Token Management: Always set chunks_per_source when using extract or crawl to manage your token budget effectively.
• Search Depth: Use search_depth='advanced' for high-stakes information retrieval, but default to basic for quick, cost-effective tasks.
• API Limits: Keep in mind that search and extract have limits on result counts and URL batching; always validate your lists before invoking API methods.
• Environment Safety: Never hardcode your API key; always use environment variables as documented in the prerequisites.