What problem does it solve? Developers building AI agents and RAG systems need reliable access to real-time web data, but integrating search, extraction, and crawling APIs correctly requires knowing which endpoint to use, how to tune parameters, and how to avoid context window explosion and runaway costs. ## Core Features & Use Cases - Method Selection Guidance: Maps tasks to the right Tavily method—search() for web results, extract() for URL content, crawl() and map() for site-wide discovery, and research() for end-to-end AI-synthesized reports. - Parameter Optimization: Documents search depth tradeoffs, query and chunks_per_source targeting, domain and path filtering, and async parallel patterns for Python and JavaScript SDKs. - Framework Integrations: Provides ready-to-use patterns for LangChain, LlamaIndex, CrewAI, Pydantic AI, Agno, Vercel AI SDK, Google ADK, and OpenAI/Anthropic tool calling. - Use Case: An engineer building a research agent uses the search-then-extract pipeline with score-based filtering and chunked extraction to gather focused, citation-ready web content without overflowing the model's context. ## Quick Start Ask your coding assistant to implement a Tavily web search integration for your agent following the best practices in this skill.
