Drop the original PDF and the edited Markdown: you get a line-by-line verdict of what survived, what was reworded and what disappeared — including numbers and figure references. No reading the whole document again.
LAST REVIEWED 15 OCT 2026 · EVERY TOOL RUNS LOCALLY IN YOUR BROWSER
1 · BASELINE — ORIGINAL PDF
pdf, page by page
2 · EDITED FILE
md or txt
Skip the PDF? You can also use an earlier .md as the baseline: choose it in the second box, then swap the roles in the first one — or use the buttons below.
Everything runs in your browser — neither file leaves your device.
Converting a PDF to Markdown and then asking an editor, a template or an AI assistant to “just reorganise the layout” is the fastest way to publish study material. It is also the fastest way to lose a paragraph, a percentage or a figure nobody notices until a student asks about it. Proofreading page by page does not scale, and a plain text diff is useless once headings become bold lines and lists become tables.
This checker compares the content, not the formatting: it ignores case, accents, quotes, dashes, list markers, table pipes and section order, and reports what a reader would actually miss.
1.240, 1240 and 336,0 versus 336 are the same figure written differently, while 96,4% turning into 96% is a real change and gets flagged. and <img>) in the Markdown.Both files are read and compared inside your browser. Nothing is uploaded, stored or logged — which matters when the material is unpublished course content, a contract or an internal report. The tool keeps working offline once the page has loaded.
Choose the original PDF as the baseline and the edited .md as the candidate, then click Check Integrity. You get a verdict, a recovery percentage and explicit lists of missing text, reworded sentences, changed numbers, renamed headings and unreferenced figures.
No — and that is on purpose. Reformatting is exactly what a layout rework changes: headings become bold paragraphs, lists become tables, lines get rewrapped. The checker normalises case, accents, quotes, dashes, list markers, pipes and section order, then verifies that the content itself survived.
It counts the embedded raster figures in the PDF and compares that with the image references in the Markdown, so you know if any figure was left behind. What no automatic check can do is tell whether a chart was replaced by a different chart — for that, compare the pages visually. The tool tells you which pages contain figures so the manual pass takes seconds.
Not necessarily. A rewrite that keeps the meaning is fine; the report shows which words of the original are gone so you can decide. Deletions appear separately as missing text, because nothing replaced them.
Yes. Use the baseline button to load an earlier .md or .txt as the reference — the ideal baseline is the file produced by the PDF to Markdown converter, before the layout was reworked. Everything else works the same way.
No. The PDF is parsed with the pdf.js engine inside your browser and the edited file is read locally, so unpublished course material, contracts and internal reports stay on your device.