Markdown to DOCX Converter

The easy way to produce a .docx from Markdown is to drop the raw text into a single Word paragraph and call it converted. The result opens, but every heading is still a line beginning with a hash, the tables are still rows of pipes, and none of Word's features apply to any of it. This converter does the harder thing: it rebuilds the document as Office Open XML.

1. Parse to a syntax tree. Your Markdown is tokenised into a nested tree of blocks and inline nodes — the same parse that drives the live preview, so what you see is what gets converted.
2. Map each node to its Word equivalent. A level-two heading becomes Word's built-in Heading 2 style. A table becomes a w:tbl with a repeating header row. An ordered list becomes paragraphs bound to a shared numbering definition, so Word renumbers them correctly when you insert an item.
3. Package as a .docx. The document parts are zipped into the standard container and saved straight to your downloads folder.

The practical difference shows up the moment you open the file. Word's navigation pane lists your headings. Changing the document theme restyles every heading at once. Exporting to PDF produces a real bookmark tree. Generating a table of contents works. None of that is possible when headings are merely large bold text.

This input:

## Release notes

1. Fixed the export path
   - Windows only
2. Added **table** support

> Upgrade before Friday.

| Change | Status |
| --- | ---: |
| Export | Done |

produces a Word document containing a Heading 2, a two-item numbered list with a bulleted sub-item at level two, a bold run inside the second item, an indented and ruled quotation, and a two-column table whose header row is bold and whose second column is right-aligned. Every one of those is a real Word object: click into the list and Word offers list controls; click into the table and Word offers table tools.

Press Load example in the converter above to run a longer version of this through the tool and inspect the result yourself.

The document is assembled by JavaScript in this tab and handed to the browser's download mechanism as a local blob. Nothing is posted, logged or retained. This site is statically hosted and has no endpoint that could receive your text even if the code tried to send it.

You can confirm this the same way you would for any such claim: watch the network panel while converting, or disconnect from the internet after the page loads and convert anyway. Unpublished drafts, client work and internal documents can go through this tool without leaving your machine.

• Images are placeholders, not pictures. Embedding an image means fetching it, and fetching it means a network request — which would break the guarantee above. Each reference is kept as visible alt text for you to reinsert in Word.
• Raw HTML is flattened. Markdown permits arbitrary HTML, and Word has no equivalent for most of it. HTML blocks are reduced to their visible text; scripts, iframes and styling are discarded.
• No custom templates yet. The output uses Word's default styles. You can apply your own template afterwards — because the headings are real styles, restyling is a single operation in Word rather than a manual pass.
• Footnotes and citations are not converted. Markdown footnote syntax is not part of GitHub-flavored Markdown and is rendered as literal text.
• Math is not rendered. LaTeX or MathJax expressions pass through as plain text rather than becoming Word equations.
• Nesting beyond five list levels is flattened to the fifth level, which is the depth the numbering definition covers.