zipbox_
Claude CodeAnthropic

Claude Code with nothing held back.

Bypass permissions on, full root, a machine of its own. This is Claude Code the way it ships — minus the part where it can touch your laptop. The session on this page is real: write the server, run it, curl it, done.

Boot Claude CodeFree credit to start — no card →
⊞ Dashboarddapper-vole
── Claude Code v2.1.198 ──────────────
Welcome back!
claude-sonnet · API
Usage Billing
/workspace
What's new
Claude in Chrome is now generally available
Added background agent notifications in `claude agents` — sessions that need input or finish now fire the `Notification` hook (`agent_needs_input` / `agent_completed`)
Added `/dataviz` skill for chart and dashboard design guidance with a runnable color-palette validator
/release-notes for more
write a tiny node http server that replies "hello from my sandbox", run it, curl it, and show me the output. be brief.
Thought for 4s (ctrl+o to expand)
● Write(server.js)
└ Wrote 2 lines to server.js
1 const http = require('http');
2 http.createServer((_, res) => res.end('hello from my sandbox\n')).listen(3000);
● Bash(node /workspace/server.js &
sleep 0.3…)
└ hello from my sandbox
● Done.
✻ Cogitated for 11s
› ▮
⏵⏵ bypass permissions on (shift+tab to cycle) · ⌘ for agents
Claude Code · dapper-vole.zbox.sh⚙ 0%▤ 4%⛁ 8%⌘K to askPop outDelete

real session · captured live on zipbox.ai · sandbox deleted afterClaude Code source ↗

time to a live terminalseconds
your URLclaude.zbox.sh
price$0.0342/hr · $25/mo

How you set up Claude Code in the cloud

01

Sign in

Email if you want it simple, a crypto key if you want anonymity. No passwords — every session is cryptographically signed.

02

Boot Claude Code

A fresh Firecracker microVM boots and Claude Code installs and launches itself at first boot. On pay-per-use a metered key is injected; or paste your own key at boot.

03

It’s live in your browser

A full-root terminal opens in the browser and the box answers at its own zbox.sh subdomain. Pause parks it for free, delete wipes it for good, SSH is optional.

Questions, answered straight

Do I need an Anthropic API key to run Claude Code?
No. On pay-per-use, a metered key is injected when the sandbox boots and tokens are billed to your prepaid balance at cost plus 3% — Claude Code works immediately. Prefer your own billing? Paste your Anthropic key at boot and pay them direct.
What does running Claude Code in the cloud cost?
$0.0342 per hour while the sandbox runs — about $25 a month left on 24/7 — and $0 while it’s paused. You start with free credit — no card. Claude Code’s default machine is 2 vCPU · 4 GB RAM · 80 GB disk, and other sizes start at $10/mo.
What happens when I pause or delete a Claude Code sandbox?
Two buttons, honestly: Pause shuts the machine down and parks its disk for free — restore it later and pick up where the agent left off. Delete wipes the whole VM — disk, memory, network — permanently. You only ever pay while it runs.
Is it safe to let Claude Code run unattended with full permissions?
In a sandbox, yes — that’s the point. The box is a hardware-isolated Firecracker microVM with its own kernel and none of your files, keys, or cookies. Close the tab and it keeps working; the blast radius is one disposable machine.
Can I SSH into the sandbox?
Optionally. The browser terminal is the default and needs no setup. Add an SSH public key when you boot and you can also connect from your own terminal — same machine, same session.

every agent zipbox boots

PiClaude CodeCodexGrokHermesOpenClawopencodeClineCursor CLI

Claude Code, on a machine you can throw away.

One click boots Anthropic’s agentic coding tool on a disposable cloud machine — and you pay only while it runs.

Boot Claude CodeFree credit to start — no card →