Start with your workspace.

1. Get the app.

Builds support macOS 13+ on Apple Silicon and Windows x64/ARM64. Find downloads and installation details on GitHub Releases.

2. Choose your model setup.

Local is the first-run default. Choose and download a model that fits your hardware. The inference runtime is included. Model downloads and hardware requirements vary.

For a shared deployment, an administrator runs Veil and Veil Console on your infrastructure. Select Company server in Settings and use the enrollment code they provide.

Connect a cloud provider.

Select ChatGPT or Claude in the sidebar and connect your own account through its official runtime. Provider access and usage limits are managed by that provider. The latest development version includes these workspaces; check the release notes for packaged availability.

3. Open a workspace.

Give Umbratic a task. It can work with local files, use skills, run workspace tools, and ask for approval when required by your settings.

4. Review the result.

Inspect code changes, review documents, and add feedback. Use Git worktrees to keep a task’s changes separate while you work.

Build or self-host.

The repository includes the desktop app, Rust runner, Veil model gateway, and Veil Console. Development and server guides live alongside the code.