Markdown Renderer

What it is for

Markdown is portable in theory and inconsistent in practice. A document that looks right in one editor loses its tables in a wiki, drops its footnotes on a static site, or renders its maths as literal dollar signs somewhere else. The usual way to discover this is to publish and find out.

Markdown Renderer is a viewer that shows the document accurately, and then goes a step further: it examines the source and reports what is likely to break elsewhere, and why. The aim is not just to display the output, but to explain the difference between what you wrote and what another renderer will make of it.

Reading a document

Full Markdown rendering through markdown-it-py, with intelligent table handling — Auto, Markdown or HTML render modes, and per-table diagnostics explaining which was chosen and why. KaTeX renders both inline and display maths. A sidebar table of contents scrolls to any heading, live search highlights matches as you type, zoom runs from 50 to 200% and persists between sessions, and local images resolve properly. Parser profiles — Strict, GFM, Document — change how the backend actually parses, not merely how it complains.

The full check

Beyond preview, the app can audit a document for the things that make Markdown travel badly:

• Flavor and portability — superscripts, footnotes, task lists, strikethrough, definition lists, admonitions, custom heading IDs, TeX-style maths delimiters.
• Structure — multiple H1s, repeated heading text, skipped heading levels, heading markers missing their space.
• Lists — suspicious indentation, mixed markers, missing spaces, and Unicode bullets that look like lists but are not.
• Links and footnotes — missing, duplicate and unused definitions; malformed inline link and image syntax.
• Code and malformed source — unclosed fences, and the other cases where one broken block quietly swallows the rest of the file.
• Tables — raw HTML tables, suspicious pipe structure, and tables complex enough to require HTML rather than a portable pipe form.
• Encoding and copied text — invisible Unicode characters and the wrapping patterns that PDF, OCR, email and chat-copy workflows leave behind.

Cleanup that does not overwrite anything

Cleanup runs in two layers. A safe pass always runs, reparsing the document and writing it back in a consistent layout for headings, lists, blocks and tables. On top of that, a conservative set of optional fixes maps directly to diagnostics the check already reported — heading markers missing a space, list markers missing a space, fake Unicode bullets, invisible spacing characters.

Crucially, cleanup does not touch the file on disk. It updates the preview and the in-memory document so you can read the result first, with a summary in the lower pane and a saveable cleanup log. If you want to keep it, you export it. If cleanup or repair changes exist only in memory, the app asks before you open another document or quit.

Guardrails

Large documents raise a warning, expensive automatic resource checks are skipped, and files above the safe open limit are blocked rather than allowed to hang the window. Raw HTML in the source is sanitized before both preview and HTML export, so a document from somewhere else cannot execute anything.

What it does not do