Getting started with the Sandbox SDK is easy. Follow these steps to install the package, configure your credentials, and initialize a sandbox.

​
Prerequisites

  • Python 3.10 or higher
  • Pip package manager
  • Valid CLOUDCODE_API_KEY

​
1. Install via pip

Install the cloudcode package, which includes both Local and Sandbox SDKs:

pip install cloudcode

​
2. Set Up API Keys

The Sandbox SDK uses your CloudCode API key to authenticate and manage resources.

​
Recommended: Use CloudCode API Key

Set the environment variable in your shell:

export CLOUDCODE_API_KEY="your_cloudcode_api_key"

Or in Python:

import os
os.environ["CLOUDCODE_API_KEY"] = "your_cloudcode_api_key"

​
Alternative: Provider Keys (Legacy)

You can also pass OpenAI and Anthropic keys directly (not recommended):

from cloudcode import SandboxSDK

sdk = SandboxSDK(
    model="gpt-4.1",
    api_key=os.getenv("CLOUD_CODE_API_KEY")
)

​
3. Initialize the Sandbox SDK

Import and create an instance of SandboxSDK:

from cloudcode import SandboxSDK
import os

sdk = SandboxSDK(
    model="gpt-4.1",               # Choose your AI model
    api_key=os.getenv("CLOUDCODE_API_KEY"),
    sandbox_timeout=600              # Sandbox lifetime in seconds (default 600)
)

# Verify connection and sandbox creation
info = sdk.get_sandbox_info()
print(f"Sandbox ID: {info['sandbox_id']}")
print(f"Expires at: {info['end_at']}")

You’re now ready to work in an isolated sandbox environment.

For usage examples, see: