# @push.rocks/smartkvm

Programmable browser-based visual KVM automation with frame capture and keyboard transport.

## Install

```sh
pnpm add @push.rocks/smartkvm
```

## Usage

```typescript
import {
  SmartBrowserKvm,
  SmartKvmTerminal,
  type IOcrEngine,
} from '@push.rocks/smartkvm';

const ocrEngine: IOcrEngine = {
  async recognize(frame) {
    // Plug in an OCR implementation here.
    return {
      text: '',
      confidence: 0,
    };
  },
};

const kvm = new SmartBrowserKvm({
  url: 'https://jetkvm.local',
  kind: 'jetkvm',
  username: 'admin',
  password: 'admin',
  headless: false,
  ignoreHttpsErrors: true,
});

await kvm.connect();

const terminal = new SmartKvmTerminal({
  kvm,
  ocrEngine,
  osHint: 'linux',
  shellHint: 'bash',
});

await terminal.bootstrap();
const result = await terminal.runCommand('uname -a');
console.log(result);

await kvm.disconnect();
```

## Scope

This package is transport-focused and AI-agnostic. It automates browser-based visual KVM devices by opening the KVM web UI, focusing the viewer, capturing frames, and sending keyboard input. OCR and AI/model integrations are intentionally pluggable and external.

## Manual Browser Test

Automated tests do not require real KVM hardware. To run the manual browser smoke test, set:

```sh
SMARTKVM_TEST_URL=https://your-kvm.local pnpm test
```

Optional variables are `SMARTKVM_TEST_USERNAME`, `SMARTKVM_TEST_PASSWORD`, and `SMARTKVM_TEST_HEADLESS=false`.