Convert files to Markdown in your browser.

More than .pdf and .docx

Dashed badges need a callback (vision for raster images, transcription for audio, video understanding for video) and are not registered by all(). SVG converts locally.

Use the same converter in your app.

$ npm install @mdgate/converters

// Everything
import { toMarkdown } from '@mdgate/converters';

const markdown = await toMarkdown(bytes, {
  path: filename,
});

// One format
import { toMarkdown } from '@mdgate/pdf';

const markdown = await toMarkdown(bytes);

// Your own converter
import { create } from '@mdgate/core';
import { pdf } from '@mdgate/pdf';

function myFormat() {
  return {
    id: 'my-format',
    sniff(bytes, hint) {
      if (looksLikeMine(bytes)) return 2;
      if (hint?.path?.endsWith('.mine')) return 1;
      return 0;
    },
    convert(bytes) {
      return { markdown: parseMine(bytes) };
    },
  };
}

const toMarkdown = create([myFormat(), pdf()]);
const markdown = await toMarkdown(bytes);

Pure TypeScript, Node.js, Cloudflare Workers, Edge, Browser, No Python, No native addons, No WASM, Zero third-party runtime dependencies

Give your agent files it can actually read.

Turn Word, spreadsheets, presentations, email, PDFs, and other real-world files into Markdown your agent can grep, search, chunk, index, cite, and reason over.

1. • .pdf
   • .docx
   • .xlsx
   • .pptx
   • .hwp
   • .pages
   • .msg
   • …
2. mdgate/converters
3. Markdown
4. • grep
   • search
   • chunk
   • index
   • cite

Why this converter

FAQ

Are files uploaded to mdgate?

No. The converter on this page runs in a Web Worker and locally supported files stay in your browser.

What file formats can it convert?

PDF, Word, Excel, PowerPoint, OpenDocument, Apple iWork, HWP, WPS, email, OneNote, Visio, ebooks, archives, data files, and more.

Can I use it in Cloudflare Workers?

Yes. The converters are written in TypeScript and run in Cloudflare Workers without Python, native addons, or WASM.

Can I install only one format?

Yes. For example, use @mdgate/pdf, @mdgate/docx, or @mdgate/hwp.

What about scanned PDFs, images, audio, and video?

Deterministic document content is parsed locally. Media that requires model understanding can be connected to a model callback supplied by your application.