WebFetch takes a URL, downloads the page, converts it to Markdown, and processes it. Give Claude a link to a library's docs page and ask it to summarize the API. Point it at a GitHub issue to get the discussion. It works on any public URL.
WebSearch searches the web and returns results with links. Use it when you need something current: the latest version of a package, a changelog entry, or a Stack Overflow answer for an error you've never seen. I'll put it this way: WebFetch is better when you already know the exact URL, and WebSearch is better when you're not sure where the answer lives.