defuddle
Teach an agent to extract readable Markdown and selected metadata from web pages with the Defuddle CLI.
Overview
The Defuddle skill provides a small command-line workflow for turning an HTML page into readable content. Its instructions cover producing Markdown, saving output to a file and extracting individual metadata properties such as a page title or description. The intended use is to remove page navigation and other clutter before an agent works with the material. It packages guidance for an existing CLI rather than supplying a separate browsing service. The agent’s environment needs the Defuddle command and access to the page being processed. The original file includes the supported output flags and examples.
Getting started
Follow the skill repository’s installation instructions and make the Defuddle CLI available in the agent environment. Use the documented Markdown-output option when extracting a standard web page.
Compatibility and requirements
Requires the Defuddle CLI and command execution in the agent environment.
Things to know
This is an extraction workflow for web-page content. The captured instructions do not establish access to blocked, authenticated or otherwise unavailable pages.
Skill instructions
Sources and documentation
- https://github.com/kepano/obsidian-skills/blob/main/skills/defuddle/SKILL.md
- https://github.com/kepano/obsidian-skills
Related MCP servers and skills
- obsidian-cliAgent skill
- json-canvasAgent skill
- obsidian-basesAgent skill
- PatentRef MCPMCP server
- TestWell Lab Tests MCPMCP server
- Altss Public Profiles MCPMCP server