Windows container images for AI agent testing environments.
This project provides minimal Windows container images designed for testing whether AI agents can install open-source tools from GitHub repositories. Unlike the docker-dev project which focuses on developer productivity, these images intentionally include only essential tools.
| Folder | Description | Docs |
|---|---|---|
wincore/ |
Windows Server Core 2022 with Chocolatey, Git | wincore/README.md |
⚠️ Windows containers require a Windows host
-
Install Windows in UTM:
- Download Windows 11 ARM from Microsoft
- Create a new VM in UTM with Windows 11 ARM ISO
- Complete Windows installation
-
Install Docker Desktop:
- Download Docker Desktop for Windows from docker.com
- Install and restart
-
Switch to Windows Containers:
- Right-click Docker Desktop tray icon
- Select "Switch to Windows containers..."
- Windows 10/11 Pro or Enterprise (Hyper-V enabled)
- Docker Desktop with Windows containers mode
# Clone this repository
git clone https://github.com/yourusername/docker-windows.git
cd docker-windows
# Build the image
docker build -t wincore -f wincore/Dockerfile wincore
# Run interactive session
docker run --rm -it wincore
# Run with mounted workspace
docker run --rm -it -v ${PWD}:C:\workspace wincore| Aspect | docker-dev | docker-windows |
|---|---|---|
| OS | Linux (Ubuntu) | Windows Server Core |
| Purpose | Developer productivity | AI agent testing |
| Tools | Many (Oh My Zsh, Vim, Node, Python) | Minimal (Git, Chocolatey) |
| Image size | ~2GB | ~5GB |
| Runs on | Any Docker host | Windows host only |
MIT License