Markdown Visualizer

What It Does

The Markdown Visualizer is a split-pane editor that renders your Markdown to HTML in real time. Left pane is raw text; right pane is the rendered output, updating on every keystroke. The parser is a zero-dependency micro-implementation written in ~100 lines of TypeScript - no network requests, no markdown-it, no remark, no dependencies at all.

It’s aimed at developers and writers who work in plain Markdown and want to verify their formatting quickly - without spinning up a local dev server, installing a VS Code extension, or opening GitHub’s preview tab. The shareable URL feature lets you share a draft document as a single link with no backend involved.

How to Use It

1. Type or paste your Markdown in the left pane. The right pane updates live.
2. Check word count and character count in the toolbar - useful for keeping READMEs under a target length.
3. Click Copy HTML to grab the raw HTML string the parser produced. Paste it directly into a CMS, email template, or HTML file.
4. Click Share to generate a shareable URL. The full Markdown is base64-encoded into the URL hash - anyone with the link sees the same document.
5. Click Download to save the source as a .md file.

The Math / How It Works

The parser is a line-by-line and inline-pass regex transformer, not a proper AST parser. It processes the input in two stages:

Block-level pass - each line is matched against block patterns in priority order: fenced code blocks, headings (# through ######), blockquotes, ordered and unordered lists, horizontal rules, and pipe tables. Matching lines are wrapped in the appropriate HTML element.

Inline pass - within each text block, a series of regex substitutions handles bold-italic, bold, italic, strikethrough, inline code, images, and links. The order matters: bold-italic (***) must be matched before bold (**) to avoid greedy match errors.

The shareable URL uses btoa(encodeURIComponent(markdown)) to produce a URL-safe base64 string stored in the hash. On load, decodeURIComponent(atob(hash)) restores it. This is purely client-side - nothing hits a server.

Why Developers Need This

Seeing rendered Markdown during authoring catches problems that are invisible in raw text. The most common: forgetting to close a code fence (everything after it renders as code), pipe table column misalignment (the table silently breaks), and link syntax mistakes ([text](url vs [text](url)).

The GFM vs CommonMark distinction matters more than most people realize. GitHub Flavored Markdown (GFM) supports pipe tables, strikethrough, and task lists - CommonMark does not. Many documentation platforms (Docusaurus, MkDocs, Hugo) use a specific Markdown dialect, and a table that renders perfectly in GitHub will break in CommonMark-strict mode if you have unescaped pipes in cells. This tool implements GFM-style tables so you can verify table syntax before committing.

For documentation pipelines and static site generators, this is also useful for one-off conversion checks: paste a chunk of docs, verify it looks right, copy the HTML for a CMS field, or confirm heading hierarchy before publishing.

Tips & Power Use

• Use the character count to manage long READMEs. GitHub truncates READMEs in the repository view beyond a certain render threshold; keeping under ~65,000 characters is safe.
• The “Copy HTML” button is particularly useful for CMS systems that accept HTML but not Markdown. Ghost, Substack, and some Webflow inputs accept pasted HTML well.
• Shareable URLs work in Slack and Discord - the full document is in the URL, so preview unfurls with the link, not the content. Share the link, not the rendered output.
• Verify table syntax here before committing to GitHub. GFM tables need a separator row (|---|---|) with at least one dash per cell. Missing separators silently render the table as plain text.
• The inline code pass escapes HTML before rendering - so pasting code samples with <div> tags won’t break the preview output.

Supported syntax

• Headings - # through ######
• Inline formatting - bold, italic, bold-italic, strikethrough, inline code
• Links and images - [label](url) and ![alt](url)
• Fenced code blocks - ```lang ``` (language class preserved for downstream highlighters)
• Blockquotes - >
• Unordered and ordered lists (including mixed)
• Tables - GFM-style pipe tables
• Horizontal rules - ---

Limitations

• No nested list indentation - multi-level list items flatten to one level.
• Single-line blockquotes only - multi-line > blocks are not merged.
• No footnotes, task lists, or math - these require a full parser (e.g., markdown-it or remark).
• HTML in source is escaped - raw HTML in the Markdown input is treated as text, not rendered. This is intentional (XSS prevention).
• URL length cap - very large documents may exceed browser URL limits (~8k characters after encoding). The app silently skips URL sync above that threshold.