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

# Verify Devices

> Check that your devices are connected, eligible for workloads, and reporting earnings.

Work through three checks after you install the agent or integrate an SDK: the device is **connected**, it's **eligible** for work, and it's **earning**.

## 1. Connected

The device has joined the network with your SDK key.

> [tab] Android
>
> Construct the SDK with `enableLogs = true`, keep the app in the foreground, and watch logcat:
>
> ```bash theme={null}
> adb logcat -s ZeroGpuSdk:* ZeroGpuWebSocket:*
> ```
>
> The device is connected once you see `WebSocket ready for tasks`.

> [tab] Chrome Extension / Browser / Telegram
>
> Turn on console logs (`enableConsoleLogs="true"` on the script tag, or `enableConsoleLogs: true` in `offscreen.js`) and open the console. For a Chrome extension, that's **Inspect views: offscreen.html** on `chrome://extensions`. The SDK logs the device registering, loading a model, and becoming ready.
>
> On a website, the handle returned by `initZeroGpuSdk` reports `status === "running"`. For an extension, `npx zerogpu doctor` checks the integration itself.

> [tab] macOS app
>
> ```js theme={null}
> await worker.status(); // { running: true, ready: true, ... }
> ```
>
> `ready: true` means a model is loaded and the Mac is connected.

> [tab] Agent
>
> ```bash theme={null}
> zerogpu-edge status
> ```
>
> The status shows the service registered and serving. The ZeroGPU Edge app shows the same under **Status**. If it doesn't, run `zerogpu-edge doctor`.

> [tab] Server / npm
>
> ```bash theme={null}
> curl -s http://127.0.0.1:8080/healthz
> ```
>
> `"ready": true` means a model is loaded and the worker is connected.

Then open **Analytics** in the [Edge Operator Portal](https://edge.zerogpu.ai). The **Device Fleet** map shows devices seen in the last 24 hours, and **Unique Devices** counts them over time.

## 2. Eligible for workloads

A connected device receives work when all of these hold:

| Check | Where to look |
| - | - |
| The app or fleet is **Approved**, not Pending Review, Suspended, or Rejected | **My Apps** in the portal |
| The platform you integrated is one of the app types or device types you registered | **My Apps** → your app's setup |
| The device has the resources to run a model: free memory and storage, and on phones, a battery above 20% or charging | SDK logs (`degraded` events), `zerogpu-edge doctor` |
| It's allowed to run right now: the app is open (mobile and browser), the Mac is on power (macOS desktop app SDK policy), or the agent machine is inside its `--resource-policy` window | Your integration, `zerogpu-edge status` |
| The user hasn't turned resource sharing off | Your opt-out setting |

> [!NOTE]
>
> A registration that's "politely refused", or a device that sits ready without tasks, is normal. Work follows network demand, and new devices get less work at first. Keep devices running, and keep their stored identity (app data, `ZG_DATA_DIR`, the `/data` volume, or the agent's state folder): a reset device starts over as a newcomer.

## 3. Earning

Once a device serves its first request, it shows up in **Edge Requests** and **Earnings** on the portal's **Analytics** page. See [Earnings and payouts](https://docs.zerogpu.ai/edge/earnings.md).

## Platform guides

> [card] [Android](https://docs.zerogpu.ai/edge/sdk/android.md)

> [card] [Chrome Extension](https://docs.zerogpu.ai/edge/sdk/chrome-extension.md)

> [card] [Browser](https://docs.zerogpu.ai/edge/sdk/browser.md)

> [card] [Telegram Mini Apps](https://docs.zerogpu.ai/edge/sdk/telegram.md)

> [card] [macOS desktop apps](https://docs.zerogpu.ai/edge/sdk/electron.md)

> [card] [Server / npm](https://docs.zerogpu.ai/edge/sdk/server.md)

> [card] [Windows agent](https://docs.zerogpu.ai/edge/agent/windows.md)

> [card] [macOS agent](https://docs.zerogpu.ai/edge/agent/macos.md)

> [card] [Linux agent](https://docs.zerogpu.ai/edge/agent/linux.md)


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.