Cloudsway ScaleBox Sandbox
/install cloudsway-scalebox-sandbox
ScaleBox Sandbox Skill
Create and manage isolated cloud sandboxes for secure code execution.
When to Use
- Running untrusted code in an isolated environment
- Browser automation (browser-use, computer-use templates)
- Testing scripts that may have side effects
- Remote code execution without local resource impact
- Creating temporary development environments
Prerequisites
1. Get API Key
Register at ScaleBox Dashboard to get your API key.
Set environment variable:
export SCALEBOX_API_KEY="your-api-key-here"
2. Install CLI
Download the official CLI from ScaleBox Documentation:
# Visit official docs for platform-specific downloads
# https://www.scalebox.dev/docs/en/cli/installation
# After installation, authenticate:
scalebox-cli auth login --api-key $SCALEBOX_API_KEY --server-url https://api.scalebox.dev
Security Note: Always download CLI from official sources. Verify checksums when available.
3. Server URL
https://api.scalebox.dev
Port Access URL Format
Important: ScaleBox uses port prefixes, not port suffixes.
Method 1: Specify ports at creation time (Recommended)
Use API to specify custom_ports when creating the sandbox:
curl -X POST "https://api.scalebox.dev/v1/sandboxes" \
-H "X-API-Key: $SCALEBOX_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"template": "base",
"name": "my-sandbox",
"timeout": 3600,
"custom_ports": [
{"port": 8080, "name": "web-server", "protocol": "TCP"}
]
}'
Method 2: Add port to running sandbox
If you need to add a port after creation:
# CLI
scalebox-cli sandbox port add \x3Csandbox-id> --port 8080 --name web-server --protocol TCP
# API
curl -X POST "https://api.scalebox.dev/v1/sandboxes/{sandbox_id}/ports" \
-H "X-API-Key: $SCALEBOX_API_KEY" \
-H "Content-Type: application/json" \
-d '{"port": 8080, "name": "web-server", "protocol": "TCP"}'
Access URL Format
Format: https://{port}-{sandbox_domain}/path
Example:
- Sandbox domain:
sbx-abc123.k27xn5o3lnw5dan3x.scalebox.dev - Port: 8080
- Correct URL:
https://8080-sbx-abc123.k27xn5o3lnw5dan3x.scalebox.dev/ - ❌ Wrong:
https://sbx-abc123.k27xn5o3lnw5dan3x.scalebox.dev:8080/
Available Templates
| Template | CPU | Memory | Use Case |
|---|---|---|---|
| base | 2 | 512MB | Basic sandbox |
| code-interpreter | 2 | 1GB | Jupyter environment |
| browser-use | 2 | 2GB | Browser automation (VNC) |
| browser-use-headless | 4 | 4GB | Browser automation (headless) |
| desktop | 8 | 8GB | Desktop environment (VNC) |
| computer-use-preview | 2 | 2GB | Google Computer Use |
Three Ways to Use ScaleBox
| Method | Best For | Notes |
|---|---|---|
| CLI | File operations, command execution | Most powerful, cross-platform |
| REST API | Lifecycle management | Auth via X-API-Key header |
| Python SDK | Programmatic access | pip install scalebox-sdk |
Method 1: CLI Commands
1.1 Sandbox Lifecycle
# Create sandbox
scalebox-cli sandbox create --template \x3Cname> --name \x3Cname> --timeout \x3Cseconds>
# List sandboxes
scalebox-cli sandbox list
# Get sandbox details
scalebox-cli sandbox get \x3Csandbox-id>
# Pause sandbox (preserves state)
scalebox-cli sandbox pause \x3Csandbox-id>
# Resume paused sandbox
scalebox-cli sandbox resume \x3Csandbox-id>
# Terminate sandbox immediately
scalebox-cli sandbox terminate \x3Csandbox-id>
# Delete sandbox and resources
scalebox-cli sandbox delete \x3Csandbox-id>
Create parameters:
--template: Template name or ID (default:base)--name: Sandbox name (optional, auto-generated)--timeout: Timeout in seconds (default: 300)--cpu: CPU count (minimum: 2)--memory: Memory in MB--storage: Storage in GB--async: Fire-and-forget mode - returns immediately, sandbox may still be starting. You must poll status yourself.--auto-pause: Pause on timeout instead of terminate
Important: async vs sync:
- Sync mode (default, no
--async): Waits for sandbox to be fully running before returning. The returned sandbox is ready to use. - Async mode (
--async): Returns immediately after creating the request. The sandbox may still be starting. You must pollsandbox get \x3Cid>untilstatus: runningbefore using it.
# Sync mode (recommended) - sandbox is ready when command returns
scalebox-cli sandbox create --template base --timeout 600
# Output includes sandbox ID, ready to use
# Async mode - need to wait
scalebox-cli sandbox create --template base --timeout 600 --async
# Returns immediately, sandbox may still be starting
# Must poll: scalebox-cli sandbox get \x3Cid> until status is "running"
1.2 File Operations
# Upload file
scalebox-cli sandbox upload \x3Csandbox-id> \x3Clocal-path> \x3Cremote-path>
# Upload directory recursively
scalebox-cli sandbox upload \x3Csandbox-id> ./project /workspace/project --recursive
# Download file
scalebox-cli sandbox download \x3Csandbox-id> \x3Cremote-path> \x3Clocal-path>
# List files
scalebox-cli sandbox ls \x3Csandbox-id> \x3Cremote-path>
Note: Remote path must be absolute (e.g., /workspace/file.py).
1.3 Command Execution
# Execute command
scalebox-cli sandbox exec \x3Csandbox-id> "\x3Ccommand>"
# With working directory
scalebox-cli sandbox exec \x3Csandbox-id> "python3 script.py" --cwd /workspace
# With timeout
scalebox-cli sandbox exec \x3Csandbox-id> "npm install" --timeout 120
Method 2: REST API
2.1 Authentication
All requests require X-API-Key header:
curl -H "X-API-Key: $SCALEBOX_API_KEY" https://api.scalebox.dev/v1/sandboxes
2.2 Sandbox Endpoints
| Operation | Method | Endpoint |
|---|---|---|
| Create sandbox | POST | /v1/sandboxes |
| List sandboxes | GET | /v1/sandboxes |
| Get sandbox | GET | /v1/sandboxes/{id} |
| Delete sandbox | DELETE | /v1/sandboxes/{id} |
| Pause sandbox | POST | /v1/sandboxes/{id}/pause |
| Resume sandbox | POST | /v1/sandboxes/{id}/resume |
Note: File operations (upload/download/exec) are only available via CLI or SDK.
2.3 API Example
# Create sandbox
curl -X POST "https://api.scalebox.dev/v1/sandboxes" \
-H "X-API-Key: $SCALEBOX_API_KEY" \
-H "Content-Type: application/json" \
-d '{"template": "base", "name": "my-sandbox", "timeout": 600}'
Method 3: Python SDK
3.1 Installation
pip install scalebox-sdk
Module name: scalebox (not scalebox_sdk)
3.2 Basic Usage
import scalebox
import os
# Create sandbox
sandbox = scalebox.Sandbox.create(
template="base",
timeout=600,
api_key=os.environ.get("SCALEBOX_API_KEY")
)
print(f"Sandbox ID: {sandbox.sandbox_id}")
# Get signed URLs for file operations
upload_url = sandbox.upload_url(path="/workspace/test.py")
download_url = sandbox.download_url(path="/workspace/output.json")
# Kill sandbox
sandbox.kill()
Typical Workflow
# 1. Create sandbox
sandbox_id=$(scalebox-cli sandbox create --template code-interpreter --name my-task --async | grep "Sandbox ID:" | awk '{print $3}')
# 2. Upload code
scalebox-cli sandbox upload $sandbox_id ./script.py /workspace/script.py
# 3. Execute
scalebox-cli sandbox exec $sandbox_id "python3 /workspace/script.py"
# 4. Download results
scalebox-cli sandbox download $sandbox_id /workspace/output.json ./output.json
# 5. Cleanup
scalebox-cli sandbox delete $sandbox_id
Security Notes
- Sandboxes are completely isolated from local machine
- All file paths must be absolute (no
..traversal) - Internet access enabled by default (disable with
--internet=false) - Credentials never stored in database
- Minimum CPU: 2 cores
- Always download CLI from official sources
Error Handling
| Error | Solution |
|---|---|
CPU count must be at least 2 |
Use --cpu 2 or higher |
health check timeout |
Check status with sandbox get |
Sandbox not found |
Verify sandbox ID |
Timeout exceeded |
Create new sandbox |
Links
- Official Website: https://www.scalebox.dev
- Documentation: https://www.scalebox.dev/docs
- CLI Installation: https://www.scalebox.dev/docs/en/cli/installation
- API Reference: https://www.scalebox.dev/docs/en/api
Version
- Skill: 1.0.5
- CLI: v1.3.4
- SDK: v0.1.9
- API: v1
Last updated: 2026-03-21
- 确保已安装 OpenClaw(本地或 Docker 部署)
- 在对话框中输入安装命令:
/install cloudsway-scalebox-sandbox - 安装完成后,直接呼叫该 Skill 的名称或使用
/cloudsway-scalebox-sandbox触发 - 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
Cloudsway ScaleBox Sandbox 是什么?
Create and manage isolated cloud sandboxes for secure, remote code execution, browser automation, and temporary development environments via CLI, API, or Pyt... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 119 次。
如何安装 Cloudsway ScaleBox Sandbox?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install cloudsway-scalebox-sandbox」即可一键安装,无需额外配置。
Cloudsway ScaleBox Sandbox 是免费的吗?
是的,Cloudsway ScaleBox Sandbox 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。
Cloudsway ScaleBox Sandbox 支持哪些平台?
Cloudsway ScaleBox Sandbox 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。
谁开发了 Cloudsway ScaleBox Sandbox?
由 gloriathepenguin(@gloriathepenguin)开发并维护,当前版本 v1.0.5。