superserve 0.5.0

Python SDK for the Superserve sandbox API

Superserve Python SDK

Python SDK for the Superserve sandbox API — spin up isolated sandboxes, run commands in them, and tear them down from code.

Install

pip install superserve
# or
uv add superserve
# or
poetry add superserve

Requires Python 3.9+.

Quick start

import os
from superserve import Superserve

client = Superserve(api_key=os.environ["SUPERSERVE_API_KEY"])

sandbox = client.sandboxes.create_sandbox(name="hello-world")

try:
    result = client.exec.command(
        sandbox_id=sandbox.id,
        body={"command": "echo 'Hello from Superserve!'"},
    )
    print(result.stdout)
finally:
    client.sandboxes.delete_sandbox(sandbox.id)

Get an API key from the Superserve console.

Resource clients

The Superserve client exposes four resource groups:

• client.sandboxes — create, list, get, patch, pause, resume, and delete sandboxes
• client.exec — run commands inside a sandbox (command) or stream output (command_stream)
• client.files — upload and download files to and from a sandbox
• client.system — health check

Configuration

from superserve import Superserve
from superserve.environment import SuperserveEnvironment

client = Superserve(
    api_key="ss_live_...",                            # required
    environment=SuperserveEnvironment.PRODUCTION,     # optional
    base_url="http://localhost:8080",                 # optional, overrides environment
    timeout=60.0,                                     # optional, default 60
    httpx_client=None,                                # optional, inject a custom httpx.Client
)

Async client

AsyncSuperserve mirrors every method on Superserve with async/await:

import asyncio
import os
from superserve import AsyncSuperserve

async def main():
    client = AsyncSuperserve(api_key=os.environ["SUPERSERVE_API_KEY"])

    sandbox = await client.sandboxes.create_sandbox(name="async-example")
    try:
        result = await client.exec.command(
            sandbox_id=sandbox.id,
            body={"command": "uname -a"},
        )
        print(result.stdout)
    finally:
        await client.sandboxes.delete_sandbox(sandbox.id)

asyncio.run(main())

Error handling

from superserve import Superserve
from superserve.errors import BadRequestError, NotFoundError

client = Superserve(api_key=os.environ["SUPERSERVE_API_KEY"])

try:
    client.sandboxes.get_sandbox("sbx_missing")
except NotFoundError:
    print("sandbox does not exist")
except BadRequestError as err:
    print("invalid request:", err.body)

Available error classes: BadRequestError, UnauthorizedError, NotFoundError, ConflictError, InternalServerError.

Documentation

Full docs: docs.superserve.ai

License

Apache-2.0