markdown

|

The page as clean Markdown, ready for LLM context windows. It resolves to a string:

const markdown = await microlink.markdown('https://example.com')

Headings, links, lists, tables, and code blocks are preserved; navigation chrome, scripts, and styles are dropped. Documents work too: point it at a PDF with a text layer or an office file (docx, xlsx, pptx, odt, rtf, epub) and the content is converted the same way.

Options

selector <string> — scopes the conversion to the first element matching the CSS selector.
selectorAll <string> | <string[]> — scopes the conversion to every matching element, resolving to an array with one Markdown string per match.
type <string> — overrides how the extracted value is normalized, per the rules grammar.

Any shared option applies too. prerender and waitForSelector matter most here: client-side rendered content only exists after JavaScript runs.

Examples

Keep just the article body, leaving out headers, footers, and sidebars:

const markdown = await microlink.markdown('https://example.com/blog/post', {
  selector: 'article'
})

One string per comment on a discussion page:

const comments = await microlink.markdown('https://news.ycombinator.com/item?id=1', {
  selectorAll: '.comment'
})

console.log(comments.length)

Wait for a client-side rendered page to finish before converting:

const markdown = await microlink.markdown('https://app.example.com/docs', {
  prerender: true,
  waitForSelector: 'main h1'
})

Under the hood the call is an extraction rule with attr: 'markdown'; use extract when you want Markdown for one field and other values alongside it.

See URL to Markdown in the content conversion guide for scoping strategies, document support, and LLM pipelines.