About PDF2MD

I built PDF2MD because I kept hitting the same wall in two parts of my own work, and the existing converters either uploaded my files to a server I didn't trust or asked me to install something I didn't want to install.

Why this tool exists

I'm an indie hacker. Most weeks I'm doing two things: shipping small web tools, and researching topics for articles I write.

For the websites, I lean on Claude Code a lot. The way I work is, I find an API I want to integrate, I grab the official docs, and I hand them to Claude Code as context. The problem is most API docs come as PDFs, or as web pages that paste into chat as a wall of broken formatting. Claude does a much better job when the input is clean Markdown — headings stay headings, code blocks stay code blocks, tables don't collapse into one long line. So I was constantly converting PDFs to Markdown before pasting.

For the writing research, I collect a lot of source material. A typical article folder has thirty or forty files: web clippings, reports in PDF, the occasional Word doc someone sent me. To actually read across them, take notes, and find quotes, I need everything in one format. Markdown is the format I read in, so everything gets converted to Markdown.

I tried a few existing tools. Some had good output but uploaded the file to a server. Some were free but capped at 10 pages. A couple worked offline but needed a Python install I didn't want to deal with on a fresh laptop. None of them did batch conversion the way I wanted: drag in twenty files, get twenty Markdown files back.

So I made one.

What it does

PDF2MD runs entirely in your browser. The PDF is parsed by PDF.js, the same engine Firefox uses to render PDFs natively, and the text is reflowed into Markdown on your machine. Nothing leaves your computer. There is no upload step. If you turn off your wifi after the page loads, conversion still works.

The features I personally use most:

• Batch mode. I drop in a research folder, get a zip back. This is the main reason I built it.
• Tables stay tables. Most converters flatten tables into "row1col1 row1col2 row2col1" mush. PDF2MD keeps the pipe-and-dash Markdown table syntax, which is what I want when I paste into Obsidian or hand it to Claude.
• No size cap. I've thrown 200-page reports at it. Conversion is slower, but it finishes.
• No account, no email field. I never understood why a converter would need my email.

What it's not good at, honestly: scanned PDFs. If your PDF is a photograph of a page (text is not selectable when you try in your browser), PDF2MD has no OCR and you'll get an empty result. For that, I run a separate OCR pass first. I'm thinking about adding OCR but it would mean shipping a fairly heavy WASM model, and I'd rather keep the page light.

Who I am

I'm Ben. I make small tools and write about AI. I'm not a company. PDF2MD is one of those tools. The everyday case — single files, basic batches, daily use — is free, because the marginal cost of letting people use a browser-side tool is basically zero. There's a paid tier for heavy use (ad-free, larger batch ZIPs, higher quotas) that helps me cover the time I put into improving it.

If something's broken or you want to suggest a feature, the issues page is at github.com/littleben/pdf2md, or you can email [email protected].

That's the whole story.

Last updated: February 18, 2026