Preview Markdown and understand the cleaned HTML export

Know which document you are keeping

Markdown Preview has three related outputs: a visible preview, cleaned HTML markup and the original Markdown source. These are useful for different tasks. The preview lets you read the document; HTML lets you transfer rendered structure; Markdown lets you keep editing the source.

The source download retains what you typed, including raw HTML or image references that the preview omits. The HTML export instead contains the sanitized rendering. Do not assume that downloading the original Markdown makes that source safe for every other renderer.

Try a heading, emphasis and a code fence

Paste this example and wait for the preview to update. The heading should be followed by a paragraph with bold text. The fenced block displays its contents as code, rather than treating its angle brackets as an HTML element.

## Release notes

Review the **output** before sharing.

```html
<strong>shown as code</strong>
```

<h2>Release notes</h2>
<p>Review the <strong>output</strong> before sharing.</p>

Tables, task lists and strikethrough use GitHub-flavored Markdown conventions. A task checkbox is disabled in the preview. Syntax highlighting and every GitHub-specific feature are not reproduced, so check the real destination if you are preparing a README for a particular service.

Test an omitted image and a normal link

![Architecture diagram](https://example.com/diagram.png)

[Read the notes](https://example.com/notes)

The image becomes [Image omitted: Architecture diagram]. Its remote URL is not fetched by the renderer. The ordinary HTTPS link remains a link; following it is a separate action you choose. Raw HTML images, scripts, event handlers, embedded media, forms, SVG and inline styles are removed from the preview and HTML export.

Marked parses Markdown, while a separate DOMPurify allowlist and additional link checks restrict the rendered output. The Marked documentation explains why parsing and sanitization are separate concerns. This implementation limits accepted link schemes, blocks protocol-relative and backslash-containing targets, and does not load embedded images.

Choose an export, then inspect it where it will be used

1. Use Copy HTML when another editor expects an HTML fragment. The copied markup does not include the site's visual theme.
2. Use Download HTML for a simple standalone HTML document with a UTF-8 declaration and viewport setting. It contains the cleaned content, without the tool's editor.
3. Use Download Markdown to preserve an editable draft. Save it before refreshing or closing the tab because the tool has no saved-document history.
4. Recheck headings, tables, links and code in the actual publishing system. Its renderer or sanitizer can make different choices.

Relative links deserve special care. A target such as notes/setup.html points somewhere different when you move a file to another folder or website. Review destinations rather than judging a link only by its label. A permitted HTTPS scheme is not an endorsement of the destination.

Normal paragraph line breaks also differ from hard breaks: use a blank line for a new paragraph, or Markdown hard-break syntax when you intend a line break within one. The tool processes at most 200,000 input characters and caps generated HTML size. Use smaller sections for large manuals, and keep the original source separately.