Caution

Stop DO NOT USE.

This repository is a prototype / proof-of-concept only. It is NOT intended for production use and MUST NOT be used in any deployment. There is no official support, no guarantee of correctness, and no commitment to maintain this project. Use the official matlab-proxy package instead

matlab-proxy-go

A Go implementation of matlab-proxy — a web server that launches MATLAB and provides browser-based access to it.

This project is a from-scratch rewrite of the original Python package. It produces a single, statically-linked binary with no runtime dependencies (no Python, no Node.js, no npm), while preserving API compatibility with the original.

Key features:

• All three licensing types: existing license, Network License Manager (NLM), MathWorks Online License (MHLM)
• Attach mode: connect to an already-running MATLAB session without spawning a new one
• Token-based authentication with port-scoped session cookies
• Concurrent browser session protection with automatic crash recovery
• WebSocket proxy for MATLAB's Embedded Connector
• Built-in web terminal for shell access from the browser
• Idle timeout with automatic shutdown
• Xvfb/Fluxbox management for headless Linux environments
• Clean shutdown with artifact cleanup

Quick Start

Option 1: Install script

The install script automatically downloads the latest release for your platform.

Linux / macOS:

curl -fsSL https://raw.githubusercontent.com/prabhakk-mw/matlab-proxy-go/main/install.sh | sh

Windows (PowerShell):

powershell -ExecutionPolicy ByPass -c "irm https://raw.githubusercontent.com/prabhakk-mw/matlab-proxy-go/main/install.ps1 | iex"

To install a specific version or to a custom directory:

# Linux / macOS
curl -fsSL https://raw.githubusercontent.com/prabhakk-mw/matlab-proxy-go/main/install.sh | VERSION=0.5.1 INSTALL_DIR=/usr/local/bin sh

# Windows
$env:VERSION = "0.5.1"; $env:INSTALL_DIR = "C:\Tools"; irm https://raw.githubusercontent.com/prabhakk-mw/matlab-proxy-go/main/install.ps1 | iex

Option 2: Install from a package (Linux)

Native .deb and .rpm packages are available on the Releases page. Packages automatically pull in recommended dependencies (Xvfb, Fluxbox) and provide clean install/uninstall tracking via the system package manager.

# Download with gh CLI
gh release download --repo prabhakk-mw/matlab-proxy-go --pattern '*amd64.deb'   # Debian/Ubuntu
gh release download --repo prabhakk-mw/matlab-proxy-go --pattern '*amd64.rpm'   # RHEL/Fedora

# Or download with curl
VERSION=0.2.3
curl -fSLO "https://github.com/prabhakk-mw/matlab-proxy-go/releases/download/v${VERSION}/matlab-proxy_${VERSION}_amd64.deb"
curl -fSLO "https://github.com/prabhakk-mw/matlab-proxy-go/releases/download/v${VERSION}/matlab-proxy-${VERSION}.amd64.rpm"

# Debian / Ubuntu — using apt (also installs recommended deps: xvfb, fluxbox)
sudo apt install ./matlab-proxy_*_amd64.deb

# Debian / Ubuntu — using dpkg (binary only, no dependency resolution)
sudo dpkg -i matlab-proxy_*_amd64.deb

# RHEL / Fedora
sudo rpm -i matlab-proxy-*.amd64.rpm

# Skip optional dependencies if not needed (e.g. desktop systems with a display)
sudo apt install --no-install-recommends ./matlab-proxy_*_amd64.deb

Option 3: Download a pre-built binary

Download a binary for your platform from the Releases page:

# Download with gh CLI
gh release download --repo prabhakk-mw/matlab-proxy-go --pattern '*linux-amd64.tar.gz'    # Linux (amd64)
gh release download --repo prabhakk-mw/matlab-proxy-go --pattern '*darwin-arm64.tar.gz'   # macOS (Apple Silicon)

# Or download with curl
VERSION=0.2.3
curl -fSL "https://github.com/prabhakk-mw/matlab-proxy-go/releases/download/v${VERSION}/matlab-proxy-v${VERSION}-linux-amd64.tar.gz" | tar xz
curl -fSL "https://github.com/prabhakk-mw/matlab-proxy-go/releases/download/v${VERSION}/matlab-proxy-v${VERSION}-darwin-arm64.tar.gz" | tar xz

# Install
sudo mv matlab-proxy /usr/local/bin/

Option 4: Build from source

git clone https://github.com/prabhakk-mw/matlab-proxy-go.git
cd matlab-proxy-go
go build -ldflags "-s -w" -o bin/matlab-proxy ./cmd/matlab-proxy/

See Installation for all platforms and options.

Run

# MATLAB must be on PATH or set MWI_CUSTOM_MATLAB_ROOT
matlab-proxy

The server prints an access URL on startup. Open it in your browser.

Attach to a Running MATLAB Session

You can connect matlab-proxy to a MATLAB session that is already running, instead of having it start a new one.

Step 1: In your running MATLAB, run the setup script:

run('/path/to/matlab-proxy-go/scripts/enable_connect.m')

Step 2: Start the proxy with the printed connection info:

matlab-proxy --ec-port <port> --mwapikey <key>

See Attach Mode for details and known limitations.

Documentation

Document	Description
Architecture	System design, component map, request flows
Installation	Building from source, cross-compilation, Docker
Usage	Configuration, environment variables, CLI flags, attach mode
Web Terminal	Built-in browser terminal for shell access
Differences from matlab-proxy	What changed, what was removed, what was added
Releasing	Version strategy, tagging, pre-releases
TODO	Known gaps and planned improvements
Contributing	Development setup, building, testing, code style

When to Use This Instead of the Python Version

This Go rewrite may be a better fit than the original matlab-proxy in the following scenarios:

Single-binary deployment

The Go version compiles to a single ~10 MB binary with all assets embedded. There is no need to install Python, pip, Node.js, or npm. This is particularly useful for:

• Minimal container images (scratch / distroless / Alpine)
• Air-gapped environments where installing Python packages is difficult
• Appliance-style deployments where you want one file to copy and run

Lower resource footprint

Go's concurrency model (goroutines) and compiled nature result in lower memory usage and faster startup compared to the Python asyncio server. If you are running many proxy instances (e.g., one per user in a shared cluster), the per-instance overhead matters.

Simplified operations

No virtual environments, no pip install, no package.json. The binary is self-contained and can be managed with standard process supervisors (systemd, supervisord, Kubernetes).

Native package management

The Go version ships as .deb and .rpm packages with dependency declarations. Installing via apt or rpm automatically pulls in recommended system dependencies (Xvfb, Fluxbox), provides clean uninstall tracking, and integrates with enterprise fleet management tools (Ansible, Chef, Puppet). The Python version requires pip install which does not manage system-level dependencies.

Cross-compilation

A single GOOS=linux GOARCH=amd64 go build command produces a binary for any target platform. No need to worry about platform-specific Python wheels or native dependencies.

When the Python version is still the better choice

• You need matlab-proxy-manager for multi-instance management (shared/isolated MATLAB sessions for Jupyter kernels). The Go version does not include this component.
• You need the development/test mode with the fake MATLAB server for local testing without a real MATLAB installation.
• You need deep integration with Jupyter via the Python entry-point plugin system.
• You rely on third-party Python-based integrations that register custom configurations via matlab_proxy_configs entry points.

License

See LICENSE for details.

---

Copyright 2026 The MathWorks, Inc.