> ## Documentation Index
> Fetch the complete documentation index at: https://docs.getsly.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Environments (test vs. live)

> Full data isolation between sandbox and production.

Sly runs two logically separate environments:

|                    | Test (sandbox)                 | Live (production)          |
| ------------------ | ------------------------------ | -------------------------- |
| Base URL           | `https://sandbox.getsly.ai/v1` | `https://api.getsly.ai/v1` |
| API key prefix     | `pk_test_*`                    | `pk_live_*`                |
| Agent token prefix | `agent_test_*`                 | `agent_live_*`             |

Everything else — accounts, agents, wallets, transfers, streams, webhooks, events, exports — is fully isolated between them. They share *nothing* at the data layer except your tenant identity.

## What isolation means concretely

* A test account cannot receive a live transfer, or vice versa
* Webhooks sent for test events use your test webhook URLs; live events use live URLs
* Data exports download test or live, never mixed
* Reports aggregate over one environment at a time

## Key prefix mismatch

* Test keys against `api.getsly.ai` → 401
* Live keys against `sandbox.getsly.ai` → 401

The SDK handles this for you — it picks the correct base URL from the key prefix.

## When you can cross boundaries

There's only one place test and live overlap: **organization identity**. Your tenant, your team members, your branding — those are shared. But any data that represents actual activity is environment-scoped.

See [Get Started → Environments](/get-started/environments) for the full comparison including rate limits, KYC behavior, and faucet access.