obs-defuddle

To extract clean markdown from a web page and save tokens, use the defuddle parse command with the --md flag. This removes navigation, ads, and clutter to produce readable, concise text for summaries.

Yes, you can get HTML or JSON output when parsing web pages. While markdown is the default format for token-saving content extraction, the tool supports optional HTML and JSON outputs to fit different processing needs.

The best way to convert web articles into concise text for LLM summaries is using a web page content extraction tool that strips clutter and outputs clean markdown. This reduces token usage and simplifies reading.

Yes, web page markdown extraction works on standard documentation pages, articles, and blog posts. It effectively removes site clutter to generate copyable, concise text suitable for downstream processing.

Yes, you need to install defuddle before extracting markdown from URLs. The Skill requires the defuddle tool to be present in your environment to execute the parse command and generate clean output.

Using markdown extraction for web pages instead of raw HTML significantly reduces token usage. It strips out non-essential structural tags and clutter, producing clean, readable text that is more efficient for language models.