HTML to Markdown now uses our document engine with sandboxed file access, so semantic body markup becomes GFM without fetching remote includes.
From real runs
| Element | Result | What happens |
|---|---|---|
| Headings | Kept | The h1 and h2 elements became # and ## Markdown headings with their visible text intact. |
| Nested lists | Kept | The ordered checklist kept its numbering, and nested ul items became indented Markdown bullets. |
| Tables | Changed | The rowspan service cell became a blank first column in the second GFM table row. |
| Block quotes | Kept | The blockquote paragraph converted to a Markdown quote line without the cite attribute. |
| Code blocks | Kept | The pre/code sample became a fenced Markdown code block marked with js. |
| Inline styling | Changed | Strong became Markdown bold and del became strikethrough, while styled span, sup, and u stayed as raw HTML. |
| Images | Changed | The data-URI image stayed as a raw img tag with its alt text and inline border style. |
| Page chrome | Lost | The nav links and footer note were absent from the Markdown output. |
| Scripts and includes | Lost | The script content did not appear, and our document engine's sandbox prevents remote iframe or include fetching. |
| Warnings | Kept | The API forwarded X-Warnings as an empty JSON array for both tested HTML inputs. |
The h1 and h2 elements became # and ## Markdown headings with their visible text intact.
The ordered checklist kept its numbering, and nested ul items became indented Markdown bullets.
The rowspan service cell became a blank first column in the second GFM table row.
The blockquote paragraph converted to a Markdown quote line without the cite attribute.
The pre/code sample became a fenced Markdown code block marked with js.
Strong became Markdown bold and del became strikethrough, while styled span, sup, and u stayed as raw HTML.
The data-URI image stayed as a raw img tag with its alt text and inline border style.
The nav links and footer note were absent from the Markdown output.
The script content did not appear, and our document engine's sandbox prevents remote iframe or include fetching.
The API forwarded X-Warnings as an empty JSON array for both tested HTML inputs.
We converted our HTML sample (490 B) with Document engine on 2026-09-11. It took 14 ms and produced a 250 B Markdown file, 49% smaller than the input. The engine raised no warnings.
Input · sample.html · 490 B
Standalone document with 1 heading, 1 paragraph, 1 table, 0 links, 0 images
Output · sample.md · 250 B
1 heading, 1 table, 2 list items, 0 code blocks, 0 image references
First lines of the output
# Feedback Atlas — Q3 activity trail
**Prepared for:** Board of Directors
| Metric | Q2 | Q3 |
|--------|--------|--------|
| ARR | \$4.1M | \$4.9M |
| NRR | 108% | 113% |
- Enterprise pipeline doubled.
- CSAT held at **94%**.pandoc
pandoc input.html -f html -t gfm -o output.mdRun with sandboxing for untrusted HTML so iframes, remote includes, and external resources are not fetched during conversion.
The HTTP API and the MCP server run the exact engine this page uses (Document engine); the response carries the same warnings the web converter shows. Check support first with GET /api/v1/pairs?from=html&to=md.
curl -X POST https://filetypeconverters.com/api/v1/convert \
-H "Authorization: Bearer $FTC_API_KEY" \
-F "file=@input.html" -F "from=html" -F "to=md" \
--output output.mdYes. The updated route produced # headings, Markdown bullets, a GFM table, and a fenced code block from the rich HTML fixture.
It flattens the table for Markdown. In the tested rowspan table, the repeated worker cell became an empty first column on the second row.
No. The document engine runs sandboxed, the script body did not appear in the output, and remote include fetches are not allowed.
Our document engine keeps raw HTML when GFM has no clean equivalent or when inline style should not be guessed away. The rich run left styled span, sup, u, and the data-URI image as HTML.