Templates

Good for: Most agent tasks: running code the model wrote, calling APIs, processing files and small data jobs.

from wbxsandbox import Sandbox

with Sandbox.create(template="base") as sb:
    sb.run("pip install feedparser")
    r = sb.run_code("""
import feedparser
feed = feedparser.parse("https://hnrss.org/frontpage")
for e in feed.entries[:5]:
    print(e.title)
""")
    print(r.stdout)

Good for: Quick shell commands, short scripts and running tests – where you want many sandboxes at once and do not need much memory.

Has less memory than base – installing large packages (pandas, PyTorch) or building big projects may run out of RAM. Use base for those.

from wbxsandbox import Sandbox

with Sandbox.create(template="mini") as sb:
    r = sb.run("python3 -c 'import sys; print(sys.version)'")
    print(r.stdout)

Good for: Let an assistant analyse spreadsheets and data, compute statistics and draw charts you can show to the user.

from wbxsandbox import Sandbox

with Sandbox.create(template="code-interpreter") as sb:
    sb.files.write("/home/user/sales.csv", open("sales.csv").read())
    sb.run_code("""
import pandas as pd, matplotlib.pyplot as plt
df = pd.read_csv("/home/user/sales.csv")
df.groupby("month")["revenue"].sum().plot(kind="bar")
plt.savefig("/home/user/chart.png", dpi=150)
""").check()
    open("chart.png", "wb").write(sb.files.read("/home/user/chart.png"))

Good for: Reading pages that need JavaScript, filling in forms, taking screenshots and testing websites – without running a browser on your own machine.

from wbxsandbox import Sandbox

with Sandbox.create(template="browser") as sb:
    sb.run_code("""
from playwright.sync_api import sync_playwright
with sync_playwright() as p:
    page = p.chromium.launch().new_page()
    page.goto("https://example.com")
    page.screenshot(path="/home/user/shot.png", full_page=True)
    print(page.title())
""").check()
    open("shot.png", "wb").write(sb.files.read("/home/user/shot.png"))

Good for: Agents that write or test JavaScript/TypeScript, build web front ends or run npm tooling.

from wbxsandbox import Sandbox

with Sandbox.create(template="node") as sb:
    sb.run("git clone https://github.com/user/project && cd project && npm ci", timeout=300).check()
    r = sb.run("npm test", cwd="/home/user/project", timeout=300)
    print(r.exit_code, r.stdout[-2000:])

Good for: Agents that build and test whole web applications – a backend, a front end and a database together.

from wbxsandbox import Sandbox

with Sandbox.create(template="fullstack", timeout=1800) as sb:
    sb.run("npx create-vite@latest app --template react-ts && cd app && npm install", timeout=600).check()
    print(sb.run("cd app && npm run build", timeout=600).stdout[-1000:])

Good for: Agents that write, compile and test Go programs.

from wbxsandbox import Sandbox

with Sandbox.create(template="go") as sb:
    sb.files.write("/home/user/main.go", 'package main\nimport "fmt"\nfunc main(){ fmt.Println("hello") }\n')
    print(sb.run("go run /home/user/main.go").stdout)

Good for: Agents that write Rust or need to check that a crate builds and its tests pass.

Compiling Rust is memory hungry, so this template has more RAM and CPU than the others.

from wbxsandbox import Sandbox

with Sandbox.create(template="rust", timeout=900) as sb:
    sb.run("cargo new hello", cwd="/home/user").check()
    print(sb.run("cargo run --release", cwd="/home/user/hello", timeout=600).stdout)

Good for: When you need a standard Ubuntu system and want the agent to install whatever it needs with apt.

from wbxsandbox import Sandbox

with Sandbox.create(template="ubuntu") as sb:
    sb.run("apt-get update && apt-get install -y ffmpeg", timeout=300).check()
    print(sb.run("ffmpeg -version | head -1").stdout)