Markdown to HTML — Source & Preview

Convert one draft and export its source

1. Paste or type Markdown, or import one UTF-8 Markdown/text file. CRLF and CR become LF; content is not trimmed or Unicode-normalized. An imported file waits for Convert, and changing the draft removes the previous result.
2. Press Convert to parse and check the supported profile. Raw HTML stays text, and prohibited link or image destinations stop conversion rather than being silently removed. Use Source to inspect complete HTML or Preview for inert formatting when its separate limits allow it.
3. Copy source or download converted.html. Both use the exact same UTF-8 fragment, without an added page wrapper, CSS or byte-order mark. The preview is not exported, and downloaded HTML is not automatically opened.

Understand the format and the preview

The profile is markdown-it 15.0.3 default with tables and strikethrough, plus the tool's explicit input, nesting and URL restrictions. It supports ordinary headings, lists, quotes, emphasis and escaped code, but is not full GitHub-flavored Markdown. There are no task-list, math, diagram, footnote or highlighter plugins.

Raw HTML is rendered as text rather than enabled. For example, a script tag in the draft is not a script running in the preview. This behavior is not a general-purpose sanitizer for arbitrary HTML, and malformed Markdown can remain ordinary text instead of causing a syntax error.

Approved links and image references remain in the source fragment. Preview links are noninteractive spans and images are labelled placeholders; no resources are fetched there. A large preview can be omitted completely while the full source, Copy and download remain available. The preview is not a pixel-faithful hosted page.

Relative references are not resolved against this site or your imported file's location. Your publishing page chooses their eventual base. Opening or rehosting an exported fragment can therefore load its approved images or follow links, unlike the inert preview inside this tool.

Examples of structure, references and empty output

Use example fills a heading A, a blank line and Hello with world in bold, without converting automatically. Conversion should produce an h1 heading and a paragraph containing strong emphasis. The result's details explain the active profile, byte counts and separate preview behavior.

A relative link such as ../guide?q=x#part can stay a relative HTML reference. An HTTPS image can stay in source but appears only as an alt placeholder in Preview. A mailto address is permitted for a link, not an image; data images, script schemes and protocol-relative references are rejected.

A nonblank document containing only unused reference definitions can legitimately produce a zero-byte HTML fragment. That is different from incomplete blank input. The result explains the empty output and can still copy or download those exact zero bytes without inventing a paragraph.

Explicit syntax, resource and export boundaries

The received draft fits 200000 UTF-16 units; canonical UTF-8 and any imported file fit 512 KiB, with at most 20000 lines. NUL, malformed UTF-8 and unpaired surrogates are rejected. Exactly one initial UTF-8 file BOM is removed; a later U+FEFF is preserved. Oversize edits are rejected rather than partially truncated.

Nesting guards, 100000 parsed tokens, bounded destination/fence strings, a 4 MiB source-output cap and a 15-second operation deadline limit the job. Preview additionally fits 256 KiB, 20000 tokens and 10000 emitted elements. These are product controls, not speed, heap or device guarantees; there is no partial-success artifact on a failed conversion.

Links permit the explicit HTTP(S), mailto and relative-reference profile; images need an HTTP(S) or relative path. Unsupported schemes, credentials, ambiguous authorities and prohibited control/backslash encodings stop conversion. Allowed destinations are not a claim that a linked site or downloaded fragment is trustworthy.

There is no complete-page export, user CSS, raw-HTML mode, live editor, cloud/URL import, persistence or batch conversion. Copy and Download always use the stored current fragment, not the preview. Changing the input or cancelling retires old actions; a clipboard write or download already handed to the operating system cannot be retracted.

Local draft processing and current results

Parsing and rendering run in a local worker. This tool adds no draft/file upload, URL fetching, input-bearing address, stored history or content telemetry. Its isolated preview does not load images or activate links. Existing site services remain; local processing is not a certification of analytics vendors or deployed behavior.