Fetch content from an http or https URL.

Use this for reading current web pages, downloading files that need to be inspected,
or explicitly extracting structured data from RSS/Atom feeds and OPML subscription lists.

Notes:
- Authenticated or private URLs often fail in this tool. Prefer specialized MCP or authenticated tools for GitHub, Confluence, Jira, Google Docs, and similar services.
- Redirects to a different host are not followed automatically. If a page redirects cross-host, call `webfetch` again with the returned redirect URL.
- A small built-in docs whitelist can bypass approval for `webfetch` only. It is not a general network allowlist for other tools.
- `yolo` mode still skips approval, but URL safety validation and redirect restrictions continue to apply.
