You wrote what you thought was a bulleted list on GitHub, pushed, and the rendered page shows a single long line with asterisks floating in it. Surprise: blank lines matter in Markdown too.

This is a split-pane Markdown editor. Left side: whatever you are typing. Right side: rendered HTML that updates on every keystroke. Three parsing flavors are selectable -- CommonMark (the safe baseline), GFM (what GitHub actually renders, with tables, strikethrough, task lists, and URL auto-linking), and Markdown Extra / MultiMarkdown flavors if you need footnotes, definition lists, and abbreviations.

The most valuable feature is not the preview -- it is the dialect-mismatch warning. Paste a table written for GFM and switch the parser to CommonMark. The right pane instantly becomes a flat line of pipe characters. That visual shock teaches you more about which Markdown is which than any spec doc.

A few other behaviors that will save you a headache:

• HTML passthrough. Markdown lets you embed raw HTML. This previewer does too -- <div align="center">, <details>, <abbr> all work. If your target platform is GitHub, remember it strips style=, onload=, <script>, <iframe>, and some SVG attributes. This tool does not strip them so you can see what the parser intended. Always eyeball the target platform's sanitizer before shipping.

• Math ($...$ inline, $$...$$ block) is selectable via a toggle. KaTeX renders it. If you use bare $ for prices (e.g. "it costs $5 to ship"), disable math mode.

• Mermaid diagrams are supported under GFM. Fenced mermaid blocks become flowcharts, sequence diagrams, and Gantt charts. Mermaid syntax errors produce a visible error block instead of failing silently -- that alone is worth the price of admission.

• Task lists (- [x] Done, - [ ] Todo) render as checkboxes in GFM mode. In CommonMark they stay as literal [x] and [ ] strings, because that spec does not define them.

• Table of contents is auto-generated from ##-level headings. Skipping a level (jumping from ## to ####) breaks the nesting -- the editor flags that.

Export options: raw Markdown, rendered HTML, HTML with embedded GitHub-like CSS for pasting into a CMS, plain text with markup stripped (for word-count sanity), and browser Print-to-PDF for hard copies.

If you are maintaining a docs site, writing a CONTRIBUTING.md, or drafting a blog post, this catches parser-specific gotchas before they land on a real commit.