Tavily Web

A solid wrapper around Tavily's API that gives you five distinct web research operations: basic search, content extraction from URLs, instructed crawling, site mapping, and structured research with custom schemas. The two-phase architecture is smart, using a sub-skill to handle API calls so you don't bloat your main conversation context with HTTP overhead. Supports both recent search (with date ranges) and deep extraction in markdown format. The crawl feature is the standout here, you can give it natural language instructions like "find all Python SDK pages" and it traverses intelligently. Setup is straightforward with either environment variables or a local .env file. If you're doing any kind of real-time research or need to pull structured data from the web programmatically, this covers the bases.