Adam Anwar
  1. Work
  2. MarkItDown Drop
  3. Docs

MarkItDown Drop

Local drag-and-drop web tool that converts documents, images and audio files into Markdown.

3 min read

Source on GitHub

A drag-and-drop web tool that converts documents into Markdown. It runs a local Flask server and uses the Microsoft markitdown library. It accepts PDF, Word, PowerPoint, Excel, images, audio, and structured data files.

Features

  • Drag and drop. Drop a supported file onto the target zone.
  • Split view. The raw Markdown sits beside a rendered HTML preview.
  • One-click actions. Copy the result to the clipboard, or download it as a .md file.
  • Batch queue. Convert several files in one batch, then move between the results.
  • Dark theme. A glassmorphism interface written in vanilla CSS.
  • Local only. The server runs on 127.0.0.1. No file leaves the machine.

Supported formats

app.py holds the authoritative list in SUPPORTED_EXTENSIONS.

Category Extensions Result
Office documents .docx, .pptx, .xlsx, .xls Keeps headings, bullet lists, tables, and layout structure
PDF .pdf Extracts text, headings, and tabular content
Images .jpg, .jpeg, .png Extracts EXIF metadata
Audio .wav, .mp3 Speech-to-text transcription
Structured data .csv, .json, .xml, .html, .htm Formats the data as Markdown tables and code blocks
Text .txt, .md Passes the text through
Archives and email .zip, .epub, .msg Unpacks the container and converts what it holds

Note: image conversion returns EXIF metadata. It does not read text out of the picture. Text extraction from an image needs an LLM client that this build does not configure.

Prerequisites

  • Python 3.10 or later.
  • macOS, Linux, or Windows.

Quick start

Run the launcher script:

./run.sh

The script creates a virtual environment in .venv, installs the dependencies, starts the server on http://127.0.0.1:5050, and opens a browser.

Manual install

  1. Create and activate a virtual environment:

    python3 -m venv .venv
source .venv/bin/activate

  2. Install the dependencies:

    pip install -r requirements.txt

  3. Start the server:

    python app.py

  4. Open http://127.0.0.1:5050.

To use a different port, set PORT before you start the server:

PORT=8080 python app.py

Dependencies

requirements.txt pins these packages:

markitdown[docx,pptx,xlsx,xls,pdf,outlook,audio-transcription]>=0.1.6
flask>=3.0.0
audioop-lts; python_version >= "3.13"

The extras list decides which converters exist. If you add a format, add its extra here as well.

API

Method and path Purpose
GET / Serve the interface
GET /api/formats Return the supported extension map
POST /api/convert Accept an upload and return Markdown

Project structure

app.py                              Flask server and conversion endpoints
requirements.txt                    Python dependencies
run.sh                              macOS launcher script
static/app.js                       Drag and drop logic, fetch calls, preview, tabs
static/style.css                    Dark theme design system
templates/index.html                Interface markup
MarkItDown Drop.app/                macOS application bundle that calls run.sh
LICENSE                             MIT

Notes

app.py runs Flask with debug=True. This is correct for local use.

CAUTION: Do not expose this server to a network while debug=True is set. The Werkzeug debugger runs code that a request sends it. Set debug=False and bind a real WSGI server first.

License

MIT. See LICENSE.

Back to top