Run agents.Make them better.

Run agents in isolated sandboxes. Connect models and tools, automate workflows, and build products on Tangle.

Open the workbenchRead the docs

Create a sandbox. Run your code.

Add sandboxes to your app with the SDK and CLI.

SDK reference CLI
create.ts
import { Sandbox } from '@tangle-network/sandbox';

const tangle = new Sandbox({
  apiKey: process.env.TANGLE_API_KEY!,
  baseUrl: 'https://sandbox.tangle.tools',
});

const box = await tangle.create({
  environment: 'universal', maxLifetimeSeconds: 900,
});
await box.waitFor('running');
npm install @tangle-network/sandbox

Give every agent an isolated workspace.

Run coding agents interactively or in the background.

Claude Code$ claudeStart Codex$ codexStart OpenCode$ opencodeStart Hermes$ hermesStart OpenClaw$ openclawStart NanoClaw$ nanoclawStart Kimi Code$ kimiStart Pi$ piStart Prime Agent$ prime-agentStart
Intelligence

Understand every run.
Improve the next.

Explore Intelligence

Follow the work.

Claude Code
formatters.ts

Find what needs to change.

TangleFix unavailable uptime formatting
Tool callDuration
  1. Find files150msFind workspace files: starts 0.000 seconds after the first recorded tool; completed in 150 milliseconds.
  2. Read formatter41msRead formatter: starts 2.121 seconds after the first recorded tool; completed in 41 milliseconds.
  3. Read tests27msRead tests: starts 2.469 seconds after the first recorded tool; completed in 27 milliseconds.
  4. Baseline tests434msRun failing tests: starts 6.309 seconds after the first recorded tool; completed in 434 milliseconds; output reports two failing tests.
  5. Edit formatter89msEdit formatter: starts 11.026 seconds after the first recorded tool; completed in 89 milliseconds.
  6. Verify change296msTest the change: starts 13.715 seconds after the first recorded tool; completed in 296 milliseconds.
2 failing tests
NaN+Infinity→"0s"
Before16 passed · 2 failed
After18 passed · 0 failed

Review changes where you work.

GitHub

Handle non-finite durations

formatters.ts+3−0
@@ -22,0 +23,3 @@ export function formatUptime(valueMs: number) {+  // NaN/±Infinity mean "unavailable" here, same as in formatBytes above: NaN+  // fell through to "NaNs" and +Infinity to "Infinityd NaNh".+  if (!Number.isFinite(valueMs)) return "0s";

Build your product on Tangle.

One API key for compute, inference, and intelligence.

Sandboxes

Isolated computers for agents: shell, files, browser, any coding harness.

sandbox.tangle.tools

Intelligence

Analyze traces and compare runs before and after a change.

intelligence.tangle.tools

Inference

OpenAI-compatible access to multiple model providers through one API.

router.tangle.tools

Start with a free account.

The Free plan has no included credit. Add prepaid credit before using AI models. Sign in to compare paid plans, included usage, and limits before checkout.

Sandbox egress

Egress is outbound traffic from a sandbox, counted per calendar month (UTC). Inbound traffic is free.

PlanIncluded each monthBeyond the allowance
Free100 GiB per accountRefused, never charged
Pro2 TB per sandbox$0.05 per GB
Team2 TB per sandbox$0.04 per GB

On Pro and Team, every sandbox has its own 2 TB; allowances are not pooled across the account.

On Free, the account's sandboxes share one 100 GiB cap. Once it is reached, new sandboxes and prompts are refused until the next month, and nothing is charged.

TB and GB are decimal: 1 TB is 1,000 GB, and 1 GB is 1,000,000,000 bytes. GiB is binary: 100 GiB is 107,374,182,400 bytes.

Machine-readable terms: sandbox.tangle.tools/v1/pricing/egress

Sign in to see plansDiscuss an infrastructure workload

Build your next workflow with us.

Bring us the work you want to automate. We’ll build it with your team.

Talk to TangleStart building