Install on Your Desktop

LEAF runs as a local web app. You install it once, then launch it whenever you need it. Your data never leaves your computer.

[Screenshot: LEAF home page in a browser after running leaf serve]

Requirements

Operating system	macOS (Apple Silicon), Linux (x86_64), or Windows (x64)
Python	3.12 or newer — download from python.org
Disk	~500 MB for LEAF + room for your RAW files
RAM	8 GB minimum, 16 GB recommended for large datasets
Browser	Any modern browser (Chrome, Firefox, Safari, Edge)

Windows users

Windows additionally needs the .NET 8.0 SDK to read Thermo RAW files. Download .NET 8.0. macOS and Linux do not need .NET — LEAF uses a built-in Rust reader.

Install

LEAF ships as a single Python wheel per platform. Pick your preferred installer:

uv (recommended)

# Install uv if you don't have it: https://docs.astral.sh/uv/
uv tool install leaf

pip

pip install --user leaf

pipx

pipx install leaf

The wheel bundles everything LEAF needs: the Python backend, the Rust extraction engine, the web frontend, and the .NET reader (Windows). No separate downloads.

Launch

Open a terminal and run:

leaf serve

You should see:

LEAF v0.5.0 ready at http://127.0.0.1:18008
Press Ctrl+C to stop.

Open the URL in your browser and you'll see the LEAF home page.

[Screenshot: terminal showing leaf serve startup output]

Stop

Press Ctrl+C in the terminal window. Closing the browser tab does not stop LEAF — it keeps running until you stop the terminal.

Troubleshooting

Problem	Fix
command not found: leaf	The install location isn't on your PATH. With uv tool install, run uv tool update-shell. With pip install --user, add ~/.local/bin (macOS/Linux) or %APPDATA%\Python\Scripts (Windows) to PATH.
Port 18008 already in use	Another process is using the port. Run leaf serve --port 18009 (or any free port).
pythonnet errors on Windows	.NET 8.0 SDK is missing. Install from dotnet.microsoft.com.
RAW file fails to load	The Thermo file may be from an unsupported instrument firmware. See Troubleshooting.

Next step

→ Run your first analysis (5 minutes)

Or if your lab has a hosted MINT instance with LEAF already installed:

→ Use the hosted version instead