openui-forge-php

OpenUI Forge — PHP

Build generative UI apps with a React frontend + Laravel backend. Streams the OpenAI API's native SSE response straight through response()->stream().

Activation Triggers

• "openui php", "openui laravel", "openui php backend"
• "generative ui php", "laravel streaming ui backend"

Prerequisites

• Node.js >= 22 (24 LTS recommended) + React >= 18.3.1 (19+ recommended) (frontend)
• PHP >= 8.3 + Laravel 13.x (backend; Laravel 13 requires PHP 8.3 minimum and supports 8.3 through 8.5)
• Composer; guzzlehttp/guzzle ships with Laravel and backs the Http facade (no extra dependency to call OpenAI)
• OPENAI_API_KEY environment variable set

Quick Start

1. Create the React frontend and install OpenUI deps:

npm install @openuidev/react-ui @openuidev/react-headless @openuidev/react-lang lucide-react zod

1. Generate the system prompt:

npx @openuidev/cli generate ./src/lib/library.ts --out backend/storage/app/system-prompt.txt

1. Create the Laravel backend (see Full Code below). On a fresh app, enable API routes once with php artisan install:api.
2. Run: php artisan serve on :8000, frontend on :3000

Full Code

Backend: composer.json (require block)

{
    "require": {
        "php": "^8.3",
        "laravel/framework": "^13.0"
    }
}

Laravel bundles guzzlehttp/guzzle, so the Http facade can call OpenAI with no extra package. The OpenAI SSE passthrough below needs nothing beyond the framework.

Backend: routes/api.php

<?php

use App\Http\Controllers\ChatController;
use Illuminate\Support\Facades\Route;

// `php artisan install:api` creates this file and prefixes it with /api,
// so this route is reachable at POST /api/chat.
Route::post('/chat', [ChatController::class, 'chat']);

Backend: app/Http/Controllers/ChatController.php

<?php

namespace App\Http\Controllers;

use Illuminate\Http\Request;
use Illuminate\Support\Facades\Http;
use Symfony\Component\HttpFoundation\StreamedResponse;

class ChatController extends Controller
{
    // Loaded once per worker process, then reused across requests.
    private static ?string $systemPrompt = null;

    private function systemPrompt(): string
    {
        if (self::$systemPrompt === null) {
            $path = storage_path('app/system-prompt.txt');
            if (! is_file($path)) {
                abort(500, 'system-prompt.txt not found at ' . $path);
            }
            self::$systemPrompt = (string) file_get_contents($path);
        }

        return self::$systemPrompt;
    }

    public function chat(Request $request): StreamedResponse
    {
        // Read config(), never env(), in a controller: after `php artisan
        // config:cache` (standard in production) env() returns null outside
        // config files. See the config/services.php block below.
        $apiKey = config('services.openai.key');
        if (! $apiKey) {
            abort(500, 'OPENAI_API_KEY not set');
        }

        $incoming = $request->validate([
            'messages'           => 'required|array|min:1',
            'messages.*.role'    => 'required|string',
            'messages.*.content' => 'required|string',
        ])['messages'];

        // Prepend the system prompt; never trust a client-sent system message.
        $messages = array_merge(
            [['role' => 'system', 'content' => $this->systemPrompt()]],
            array_map(
                fn (array $m) => ['role' => $m['role'], 'content' => $m['content']],
                $incoming,
            ),
        );

        $baseUrl = rtrim(config('services.openai.base_url'), '/');
        $model   = config('services.openai.model');

        // stream => true returns the Guzzle PSR-7 response with its body still
        // on the wire, so we read it chunk-by-chunk instead of buffering the
        // whole completion in memory.
        $upstream = Http::withToken($apiKey)
            ->withOptions(['stream' => true])
            ->acceptJson()
            ->post("{$baseUrl}/chat/completions", [
                'model'    => $model,
                'stream'   => true,
                'messages' => $messages,
            ]);

        if ($upstream->failed()) {
            abort($upstream->status(), 'OpenAI request failed: ' . $upstream->body());
        }

        $body = $upstream->toPsrResponse()->getBody();

        return response()->stream(function () use ($body): void {
            // Forward OpenAI's SSE bytes verbatim. OpenAI already emits
            // `data: {chunk}\n\n` frames plus a final `data: [DONE]`, which is
            // exactly what openAIAdapter() parses, so no re-framing is needed.
            while (! $body->eof()) {
                $chunk = $body->read(8192);
                if ($chunk === '') {
                    usleep(1000); // avoid a busy-wait if the stream momentarily has no data
                    continue;
                }
                echo $chunk;
                if (ob_get_level() > 0) {
                    @ob_flush();
                }
                flush();
            }
        }, 200, [
            'Content-Type'      => 'text/event-stream',
            'Cache-Control'     => 'no-cache',
            'Connection'        => 'keep-alive',
            'X-Accel-Buffering' => 'no',
        ]);
    }
}