Workbench

🇬🇧 English · 🇷🇺 Русский

Self-hosted dashboard to run, monitor and parameterize background jobs on Laravel.

Write a class with a run() method — Workbench gives it a web UI: one-click runs, a parameter form, live logs streamed over WebSocket, and input/output file handling. Scrapers, importers, converters, report generators — anything you'd otherwise run from cron or the CLI with no UI.

• 🧩 Write a class → get a UI. Subclass BaseServlet, implement run(), register it in JSON. No controllers, no Blade, no per-task admin page.
• 📡 Live logs over WebSocket. Each job streams its output into the browser in real time (Laravel Reverb) — no SSH-ing in to tail a log.
• 🗂 File I/O & param forms built in. Upload inputs, download results (single file or ZIP), forms rendered from a JSON definition.

⚠️ If you use Workbench for web scraping, respect each site's terms of service and the law. Scrape responsibly and only where permitted.

Interface

Run a servlet end-to-end — one click, logs streaming live over WebSocket, results ready to download:

Servlet list — live search (⌘/Ctrl + K), dark mode, sidebar by category:

More screenshots

Servlet page with live logs:

Servlet page (ready to run):

Login (Laravel Breeze):

Use cases

Workbench fits any "operator presses a button → task runs → watch the logs → grab the output" workflow:

• 🕷 Web scraping / price & feed monitoring — the headline example; ships with a demo crawler.
• 📥 Data imports — CSV/XLSX/feed → your DB or storage, with a form for source and credentials.
• 🔁 File conversions — batch-transform uploaded files (demo CSV converter included).
• 📊 Report / export generation — produce CSVs and exports on demand (demo report generator included), no one-off admin page.
• 🧰 Internal tools — give a non-developer a button + form to run a parameterized task themselves.
• ⏱ Cron + CLI with a UI — keep your scheduled scripts, but give them a dashboard, live logs and file handling.

Features

• 🧩 Servlets — the unit of work: subclass BaseServlet, implement run(). Three demos included (a network scraper, a file converter, a report generator).
• 📡 Live logs — log()/notice()/warn()/error()/debug() stream to the browser in real time (Laravel Reverb / WebSocket).
• 🗂 Files — upload inputs, download results (single or ZIP), clear in/out per servlet and per user.
• ⚙️ Parameters — text/password/checkbox forms declared in resources/data/servlets.json; values saved per user.
• 🔐 Auth — Laravel Breeze (login / register / password reset) on the same Inertia + Vue stack.
• 🌓 UI — responsive sidebar, dark mode, command palette (⌘/Ctrl + K), hide inactive servlets.
• 🛠 HTTP client for scrapers (HasHttpClient, optional) — Laravel HTTP + cookie jar, retry/backoff, Symfony DomCrawler, CSV & encoding helpers.

Stack

• Backend: PHP 8.4, Laravel 13
• Frontend: Inertia.js + Vue 3 + Tailwind CSS 4 (Vite)
• Realtime: Laravel Reverb (WebSocket)
• Auth: Laravel Breeze (Inertia/Vue)

Quick start

Local (requires PHP 8.4, Composer, Node 20+)

git clone https://github.com/jazz-max/workbench.git && cd workbench
cp .env.example .env

composer install
php artisan key:generate
touch database/database.sqlite
php artisan migrate --seed      # creates a demo user (below)

npm install
npm run build                   # or `npm run dev` while developing

# in separate terminals:
php artisan serve               # http://localhost:8000
php artisan reverb:start        # WebSocket server for live logs

Demo user from the seeder: test@example.com / password (or register a new one at /register).

Docker

cp .env.example .env
./vendor/bin/sail up -d
./vendor/bin/sail artisan key:generate
./vendor/bin/sail artisan migrate --seed
./vendor/bin/sail npm install && ./vendor/bin/sail npm run build

App: http://localhost:8083 (port configurable via APP_PORT).

Writing a servlet

1. Create a class in app/Servlets/ extending App\Servlets\BaseServlet and implement run(): bool.
2. Register it in resources/data/servlets.json — with a fully-qualified class name (backslash included), otherwise it won't appear in the UI.

<?php

namespace App\Servlets;

class GenerateReport extends BaseServlet
{
    public function run(): bool
    {
        $this->notice('Starting…');           // → live log in the UI

        $rows = (int) ($this->params->rows ?? 100);
        $path = $this->params->outputPath . 'report.csv';

        $h = fopen($path, 'w');
        fputcsv($h, ['id', 'value']);
        for ($i = 1; $i <= $rows; $i++) {
            fputcsv($h, [$i, mt_rand(1, 1000)]);
        }
        fclose($h);

        $this->notice("Done: {$rows} rows written");
        return true;
    }
}

{
  "My jobs": {
    "generate-report": {
      "classname": "App\\Servlets\\GenerateReport",
      "title": "Generate report",
      "url": "",
      "description": "What it does",
      "params": [
        { "name": "rows", "label": "Rows", "type": "text", "value": "100" }
      ]
    }
  }
}

For network jobs, add use App\Servlets\Concerns\HasHttpClient; for an HTTP client (cookie jar, retry, DomCrawler) — see app/Servlets/DemoBooksScraper.php.

Included demos: DemoBooksScraper.php (network scraper), DemoCsvConverter.php (file converter), DemoReportServlet.php (report generator, runs out of the box).

Useful BaseServlet / HasHttpClient API

Method	Purpose
log/notice/warn/error/debug($msg)	write to the live UI log (and work.log)
$this->params->{inputPath,outputPath,coockiePath}	servlet file paths
clearOutputDir() / clearin()	clear results / input files
loadLocalParams()/saveLocalParams()	persistent state (local_params.json)
initHttp($base) / fetchCrawler($uri) / httpGet/httpPost	HTTP + DomCrawler (HasHttpClient)
writeCsv($dir,$name,$rows) / clean() / processPrice()	processing & output (HasHttpClient)

For non-UTF-8 sources set protected ?string $sourceEncoding = 'windows-1251';, for output files protected string $outputEncoding = 'windows-1251'; (everything defaults to UTF-8).

How it compares

For scraping — most runners in this space are Python/Go; Workbench is the PHP/Laravel-native option:

	Workbench	Scrapyd / Gerapy / Scrapydweb	Crawlab
Scraper language	PHP (Guzzle, Roach PHP, DomCrawler)	Python (Scrapy)	any (CLI)
Stack	Laravel + Vue	Python	Go + MongoDB
Live logs	WebSocket, real-time	polling / files	yes
Extra runtime for a Laravel team	none	Python	Go cluster

For general task-running — Workbench is operator-facing, not a queue worker:

	Workbench	Laravel Horizon	cron + scripts
Trigger	operator clicks Run (with a form)	automatic, queued	scheduled
Built for	parameterized, on-demand tasks	background queue jobs	unattended scripts
UI / live logs / file I/O	yes	queue metrics only	none

Use Horizon for queued background jobs; use Workbench when a human needs to launch a task, fill in parameters, watch it run and download the result.

Live logs (Reverb)

Log streaming runs over Laravel Reverb. Fill REVERB_APP_* and VITE_REVERB_* in .env, run php artisan reverb:start, and make sure BROADCAST_CONNECTION=reverb. Channels are public (isolated by userId in the channel name), so no channel authorization is needed.

Remote access (MCP)

Companion project workbench-mcp — an MCP server that lets a Claude CLI on another machine read/write project files, search code, run allow-listed commands and git, and delegate coding to a Claude on the host. Handy for remote servlet development.

Optional proxy feature

A reverse-proxy-to-source example ships disabled by default (e.g. to open a source site's cart/checkout inside the UI). To enable: uncomment ServletController::proxy() and the servlet/*/proxy/* route, optionally validateCsrfTokens/trustProxies in bootstrap/app.php, and set VITE_ENABLE_PROXY=true. The skeleton is generic — adapt it to the target site.

Configuration

Everything is configured via .env (see .env.example). Defaults: SQLite database, DB-backed queue/cache/sessions, Reverb broadcasting.

Tests

php artisan test       # runs on in-memory SQLite
./vendor/bin/pint      # code style (Laravel Pint)

License

MIT. PRs and issues welcome — see CONTRIBUTING.md.