Convert the text layer of a PDF into Markdown locally - headings, lists and paragraphs are detected from the structure. No upload.
Running locally on your device ...
0%
Your files never left your device
No. Everything runs in your browser - your file never leaves your device. How this is verifiable
Markdown is the format for notes, wikis, static sites and AI tools: plain, versionable and readable everywhere. Anyone wanting to reuse the content of a PDF there has so far had to copy it together laboriously. This tool reads the text layer of the PDF and returns it as Markdown with the coarse structure already in place - headings, bullet lists and paragraphs instead of one single wall of text.
The structure is detected with a heuristic: lines in a noticeably larger font than the body text become headings (#, ## or ###), lines starting with a bullet or number become Markdown lists, and a larger vertical gap between lines marks a new paragraph. From the bare text positions this produces a readable, reusable Markdown document - an ideal starting point that you refine further in your editor.
Honest and important: this is a structure approximation, not a faithful layout reconstruction. Tables, multi-column pages and fine formatting cannot be derived losslessly from the text layer. As with plain text extraction: a scanned or image-only PDF has no text layer - there the result stays empty and you need the text recognition "PDF OCR" instead. Everything runs locally in the browser via pdf.js; your file does not leave your device.
| Input formats | |
|---|---|
| Output format | MD |
| Batch processing | No |
| Processing | Locally in your browser (JavaScript) |
| File upload | None |
Limitations: Converts the existing text layer of a PDF into Markdown (UTF-8 .md). Headings, lists and paragraphs are detected heuristically from font size and spacing - a structure approximation, not an exact layout reconstruction. Tables and multi-column pages are only linearly approximated. Scanned or image-only PDFs have no text layer; the result stays empty there, use "PDF OCR" instead. Unlock password-protected PDFs first.
By font size: lines noticeably larger than the body text become #, ## or ### depending on the ratio.
No. The conversion is a structure approximation; tables and multi-column layouts are only linearly approximated, not faithfully rebuilt.
Your PDF is probably a scan or made of images and has no text layer. Use "PDF OCR", which recognises text from the pixels, instead.
No. The conversion runs entirely locally in the browser via pdf.js; the file does not leave your device.