Clawk Gives Coding Agents a Disposable Linux VM, Not Your Laptop
Clawk runs coding agents in isolated, disposable Linux VMs instead of your host machine, using hypervisor-level isolation instead of prompt rules.
Clawk is a new open-source tool that runs autonomous coding agents like Claude Code or Codex inside an isolated, disposable Linux VM instead of your actual machine. It targets a common tradeoff: either babysit every command an agent wants to run, or use flags like --dangerously-skip-permissions and hope nothing critical gets deleted or leaked.
Running clawk inside a project directory boots a Linux VM with the repo mounted, root access for the agent, and no permission prompts. The isolation is enforced at the hypervisor level rather than through prompt-based rules: the host filesystem is never mounted, outbound network access is allow-listed, and ssh-agent is forwarded so operations like git push work without keys ever entering the VM.
On macOS 14+ with Apple silicon, clawk relies on Apple's Virtualization.framework directly, with no Docker, qemu, or sudo dependency; Linux support via firecracker is still experimental. If a VM gets wrecked, clawk destroy && clawk rebuilds it in seconds while preserving the repo and prior agent conversations. It also supports multi-repo ticket workflows using git worktrees and automated PR creation across repositories.
The project is explicitly pre-1.0, with breaking changes expected between releases, an important caveat for teams considering early adoption.