Purpose

This skill extracts clean, readable Markdown from any web page URL by stripping navigation menus, advertisements, sidebars, footers, cookie banners, and all other non-content elements. It produces token-efficient output that focuses exclusively on the meaningful content of the page.

When agents fetch web pages using standard tools, the raw output often includes hundreds of navigation links, promotional widgets, and boilerplate text that consume tokens without adding value. This skill eliminates that noise using Defuddle, a purpose-built content extraction tool, to isolate the article body, documentation text, or main content and return only what matters.

The skill gracefully degrades when Defuddle is not installed: it offers to install it automatically or falls back to standard web fetching with a clear note about the trade-off in output quality and token usage.

This is a universal skill — it works in any project, any terminal context, and does not require Obsidian or any specific project structure.

When to Use

Invoke this skill when:

webpage-reader

Purpose

When to Use