Local sandboxes arrive in raytrace-cli 0.5.0

Local sandboxes are now part of raytrace-cli 0.5.0. Run your coding agent in a sandbox on your own machine and review its sessions and execution evidence in the local dashboard, without a RayTrace account.

From install to your first run

The local sandbox setup walkthrough shows each step, from installing the npm package to opening the dashboard. You can also watch it full screen.

npm install -g @raytrace-cli/cli@latest
raytrace --version
raytrace doctor
cd your-project
raytrace sandbox create --local --name demo-box
raytrace connect demo-box
claude

Use Node.js 22.13 or newer, and check that RayTrace is 0.5.0 or newer. Approve the Lima download if asked. The first run sets up a Linux VM and installs the sandbox runtime; allow several minutes. Sign in to Claude Code inside the sandbox if prompted.

In another terminal on your machine, run raytrace open to review the recorded sessions, tool calls, commands, file changes, and connections.

Your platform, your setup

• macOS 13.5+: uses built-in virtualization, with Lima downloaded by the CLI if needed.
• Linux: requires QEMU and access to /dev/kvm.
• Older macOS: requires QEMU.
• Windows: experimental; install Lima yourself.

Sandbox files and captured sessions stay on your machine. Agent calls use the provider account you sign in with inside the sandbox. Optional dashboard AI features use your own provider key.

Find the full requirements and copyable steps in the local sandbox guide, or explore the raytrace-cli source on GitHub.