DOCX Creator

{{ sourceFile ? sourceFile.name : 'Drop an optional source file' }}

{{ sourceFile ? fileMeta : 'or browse for one HTML, Markdown, or text file up to 500 KiB' }}

Source format:

Markdown raw HTML remains inert text. HTML is reduced through a structural allowlist.

Source content:

Paste one source. Scripts, styles, forms, remote media, fields, macros, and embedded active content are never packaged.

{{ formatCount(sourceText.length) }} / 512,000 characters

Document title:

Leave blank to use generic document properties without inserting a title paragraph.

Download filename:

.docx

Unsafe filename characters are replaced before download.

Page size:

A4 is the default; choose Letter for US-oriented editing workflows.

Margins:

Margins are written as fixed WordprocessingML page settings.

Base typeface:

The receiving editor may substitute the typeface when it is unavailable.

Base text size:

Heading styles keep their own readable hierarchy.

Headings and lists:

Plain mode retains visible heading and list text without Word heading or numbering styles.

Reviewed hyperlinks:

Relative, file, UNC, data, script, and other schemes become plain text with a review warning.

Tables:

Plain rows are useful for receivers that handle Word table styles poorly.

Code blocks:

The source text remains unchanged; this choice affects Word paragraph styling.

Document subject:

Optional and blank by default; this adds metadata only.

Document author:

Optional and blank by default; source and metadata are never query-shared.

{{ row.label }}: {{ row.value }}

Review before download

{{ warning }}

The chart renderer is unavailable. The same counts remain available in the structure ledger.

Signal	Count	Meaning	Copy
{{ row.label }}	{{ row.value }}	{{ row.detail }}

A DOCX file is more than text with a Word filename. It is a ZIP container of related XML parts that describe paragraphs, heading styles, lists, tables, document properties, page settings, and hyperlinks. A useful conversion therefore has to decide which source structure becomes editable Word structure and which material should be flattened or left out.

Markdown, HTML, and plain text carry different amounts of structure. Markdown uses compact markers for headings, emphasis, links, lists, tables, and code. HTML can express those features along with scripts, forms, embedded media, and styling that do not belong in a restrained word-processing document. Plain text has line and paragraph breaks but no reliable semantic heading or table markup.

Source formats and the structure they can contribute to a DOCX file
Source	Structure that can be retained	Important boundary
Markdown	Headings, emphasis, lists, simple tables, links, and fenced code	Raw HTML remains inert text rather than being interpreted as markup.
HTML	Supported text blocks, headings, lists, simple tables, links, and code	Scripts, styles, forms, images, remote media, and embedded objects are omitted.
Plain text	Paragraphs and line breaks	Headings and lists cannot be inferred safely from ordinary text alone.

Preserving structure makes a document easier to navigate and edit, but fidelity has limits. Web layouts, complex CSS, nested page furniture, interactive fields, macros, and embedded media do not translate cleanly into a small, reviewable Word document. A warning about omitted or simplified content is more trustworthy than a file that appears complete while silently changing meaning.

Conversion is best treated as the start of an editing pass. Open the resulting document in the word processor that will be used for delivery, inspect headings and lists, check every important link, and review page breaks. Typeface substitution, pagination, and table width can vary between editors even when the underlying DOCX structure is valid.

How to Use This Tool:

Start with one reviewed source and choose the format that describes it accurately. Selecting HTML for Markdown, or plain text for structured content, changes what can be recognized.

Choose Markdown, HTML, or Plain text, then paste the source or select a matching text file. The source must contain supported visible text after unsupported material is removed.
Set the document title, page size, margins, base typeface, and text size. Leave the title blank when the source already begins with its own top-level heading or no title paragraph is wanted.
Choose whether headings and lists keep Word structure, safe links remain clickable, tables become editable grids, and code receives shaded monospace styling. Use the plain alternatives when the receiving editor handles those structures poorly.
Create the DOCX, then read the Structure ledger and every review warning before downloading. A nonzero omitted count means some source material did not enter the document as active or structured content.

Interpreting Results:

The paragraph, heading, list-item, table, and link counts describe the recognized document structure. They are useful checks against the source, not proof that the final pages look right. A heading count of zero is expected for plain text but deserves review when a structured Markdown document should contain sections.

Warnings identify simplification or omission. Unsafe or unsupported links become plain text, later links become plain text after the hyperlink limit, and active or embedded HTML is removed. Open the DOCX and confirm that the remaining text still communicates the intended meaning before sharing it.

Technical Details:

A DOCX without macros uses Office Open XML parts for the document body, styles, numbering, relationships, and properties. Conversion first reduces the chosen source format to supported blocks and inline runs. Those records are then written as Word paragraphs, heading styles, numbered or bulleted items, grid tables, code paragraphs, and hyperlink relationships.

Transformation Core:

DOCX conversion stages and governing rules
Stage	Transformation	Result
Parse	Recognize supported paragraphs, headings, lists, tables, code, emphasis, and links for the selected source format.	A bounded sequence of document blocks and styled text runs.
Reduce	Remove active, styling, form, metadata, image, and embedded HTML; keep unsupported link labels as ordinary text.	Visible content that can be represented without active behavior.
Map	Apply Word heading and numbering styles, table or plain-row treatment, code styling, page settings, and document properties.	Editable WordprocessingML with explicit structure choices.
Package and verify	Assemble the required document parts and relationships, then reopen the package structure for checks.	A DOCX without macros, ready for an editor-level review.

Rule Core:

Source text is limited to 500 KiB, 1,500 document blocks, and 1,000 table cells.
At most 250 HTTP, HTTPS, or mailto links become clickable. Relative paths, network paths, file links, data links, and script-like schemes remain plain text with a warning.
A4 uses 1-inch margins by default; US Letter is available. Narrow margins are 0.5 inch. Page size and margins are written as fixed document settings.
The generated document body has a 4 MiB safety limit. Input that leaves no supported text or structure is rejected instead of producing an empty document.

Character and paragraph styling is intentionally limited. Bold, italic, underline, safe hyperlinks, and inline code can survive where supported, but complex layout and cascading web styles are outside the transformation.

Privacy and Content Safety:

The source is processed in the current browser tab and is not placed in a shareable query. The document intentionally excludes scripts, macros, forms, remote media, and embedded active content. This reduces risk, but it does not certify the meaning, accuracy, licensing, or confidentiality of the text you provide.

Review sensitive text before conversion and inspect the downloaded DOCX in a trusted editor. Safe-link filtering checks the link scheme and basic shape; it does not prove that a permitted website or email address is trustworthy.

References:

ECMA-376 Office Open XML, Ecma International.
Create a word-processing document, Microsoft Learn, January 21, 2025.