tavily-best-practices

The tavily-best-practices skill provides structured guidance for building production-ready integrations with the Tavily search API. Tavily is designed to give large language models (LLMs) access to real-time web data, enabling applications to go beyond static training knowledge and retrieve up-to-date information from the internet. This skill focuses on helping developers implement reliable, scalable, and efficient workflows using Tavily within AI-driven systems, such as agents, retrieval-augmented generation (RAG) pipelines, and autonomous research tools.

The skill covers the core capabilities of the Tavily ecosystem, including web search, URL content extraction, full-site crawling, URL mapping, and AI-powered research synthesis. It also highlights important SDK usage patterns for both Python and JavaScript, including client initialization, async usage for parallel requests, and project-based tracking. Developers can fine-tune behavior using parameters such as search depth, domain filtering, extraction depth, crawling limits, and structured output options. For research workflows, the API supports asynchronous job execution with polling and model selection (e.g., mini, pro, or auto) to balance speed and depth of analysis.

This skill is intended for developers building LLM-powered applications that require reliable external knowledge retrieval. Typical use cases include autonomous agents that perform multi-step research, RAG systems that enrich context with live web data, competitive analysis tools, documentation indexing pipelines, and AI assistants integrated into development environments like Claude Code or Cursor. It is particularly valuable for teams that need structured best practices for combining LLM reasoning with external, real-time information sources.