AI-Powered Cloud Services / Sandbox SDK — Deploy Secure Environments / Browser Agent — Automated Testing / Tax Agent — AI Tax Service / Try it free at ai.tangle.tools / AI-Powered Cloud Services / Sandbox SDK — Deploy Secure Environments / Browser Agent — Automated Testing / Tax Agent — AI Tax Service / Try it free at ai.tangle.tools /
Coming Soon

Cloud in a
command.

Spin up isolated environments, run AI agents inside them, stream results in real-time. One npm install, zero DevOps.

$ npm install @tangle-network/sandbox
Documentation Join Waitlist
import { Sandbox } from "@tangle-network/sandbox"

const sandbox = new Sandbox({ apiKey: "sk_sandbox_..." });

// Create a sandbox with any Docker image
const box = await sandbox.create({
  image: "node:20",
  env: { NODE_ENV: "production" },
});

// Execute commands directly
const result = await box.exec("npm install && npm test");
console.log(result.stdout);
console.log(result.exitCode); // 0
Create. Connect. Execute.
From zero to running sandbox in under 10 seconds. SSH in, run code, stream output — all from one CLI.
Terminal — tangle sandbox
$ tangle sandbox create \ --image ubuntu:22.04 \ --gpu a100 --memory 16gb sandbox-sdk v0.4.0 Provisioning sandbox... ✓ Sandbox created sbx_7k3m9x2p · region: us-east · 3.2s $ tangle sandbox ssh sbx_7k3m9x2p Connecting to sbx_7k3m9x2p... root@sbx_7k3m9x2p:~# python3 train.py --epochs 50 --batch-size 64 Epoch 1/50 ████████████████████ loss: 0.834 acc: 0.72 Epoch 2/50 ████████████████████ loss: 0.412 acc: 0.85 Epoch 3/50 ████████████████████ loss: 0.198 acc: 0.93 $ tangle sandbox list ID IMAGE GPU STATUS UPTIME COST sbx_7k3m9x2p ubuntu:22.04 a100 running 4m 12s $0.76 sbx_a1b2c3d4 node:20 — paused 22m $0.22
TypeScript SDK. CLI. Full control.
Docker Images
Any image. Full root. GPU support.
Node, Python, Ubuntu, CUDA — or your own custom Dockerfile. Every sandbox gets full root access, custom dependencies, and optional GPU acceleration.
$ tangle sandbox create --image pytorch/pytorch:2.1-cuda12 \ --gpu a100 --memory 32gb --disk 100gb ✓ Sandbox created · sbx_9f2k1m · GPU: A100 80GB · 4.1s
AI Agent Runtime
Built-in LLM proxy.
Run Claude Code, Codex, or custom agents with tool access and file I/O inside isolated containers.
$ tangle agent task sbx_id \ "Build an API with auth" task complete · 42s
SSH & Terminal
Connect your way.
SSH directly, use the browser terminal, or control programmatically via the SDK.
$ tangle sandbox ssh sbx_id root@sbx:~# nvidia-smi A100 80GB · 0% util
Snapshots
Checkpoint anything.
Save state, restore instantly. Bring your own S3.
$ tangle sandbox snapshot \ --name "pre-deploy" saved · 2.1 GB · 0.8s
Isolation
Locked down by default.
Dropped capabilities, no privilege escalation. gVisor isolation.
[security] caps: ALL dropped [runtime] gVisor enabled sandbox isolated
File Operations
Upload. Download. Stream.
Full filesystem access — upload training data, download results, stream logs in real-time.
$ tangle sandbox upload ./data :/workspace 2.4GB uploaded in 8.1s $ tangle sandbox exec sbx_7k3m "python train.py" Epoch 1/10 ████████░░ 80% $ tangle sandbox download :/workspace/model.pt ./ model.pt (1.8GB) saved
The decentralized cloud is coming.
Create your first sandbox, run an AI agent, ship faster. Free tier included.