WBX Sandbox – Python SDK

Python SDK for WBX Sandbox: isolated cloud sandboxes for AI agents and code execution. Each sandbox is a Firecracker microVM with its own Linux kernel, disk and network rules. No dependencies – only the Python standard library (Python 3.9+).

Install

pip install wbxsandbox

Configure

Create an API key in the portal (API keys) and set it in the environment:

export WBXSANDBOX_API_KEY=...
# optional, defaults to https://sandbox.wbx.cz
export WBXSANDBOX_URL=https://sandbox.wbx.cz

Both can also be passed as url= and api_key= arguments.

Quick start

from wbxsandbox import Sandbox

with Sandbox.create(template="base", timeout=300) as sb:
    r = sb.run("python3 -c 'print(1 + 1)'")
    print(r.exit_code, r.stdout)          # 0 "2\n"
    sb.files.write("/home/user/data.txt", "hello")
    print(sb.files.read_text("/home/user/data.txt"))
# the sandbox is destroyed when the block ends

Background jobs and batches:

from wbxsandbox import run_job

job = run_job(command="python3 -c 'print(42)'", template="base", timeout_seconds=600)
job.wait()
print(job.summary()["status"])

Interactive terminal in a running sandbox (installed as the wbx command):

wbx connect sb_...

Documentation: https://sandbox.wbx.cz/docs · Pricing: https://sandbox.wbx.cz/pricing

Versioning

The SDK has its own version. Its major.minor follows the platform release line (SDK 0.15.x works with platform 0.15.x); the patch number is raised only when the SDK itself changes. pyproject.toml and wbxsandbox.__version__ always carry the same version (checked by tests/test_sdk.py).

Tests

cd sdk/python && python3 -m unittest discover -s tests

The tests do not need network access.

License

MIT – see LICENSE.