Workbench is a dashboard built specifically for BullMQ queues. It runs as a native macOS app, embeds into any Node.js server, or deploys as a standalone Docker container. All three share the same open-source UI and connect directly to your existing Redis instance.

The core problem it solves: you shouldn't need to SSH into Redis or grep through logs to understand what your workers are doing. Workbench auto-discovers queues from any Redis URL and keeps everything updated live as jobs move through states.

Key capabilities:

• Live counters – completed, failed, active, and waiting counts per queue, with p50/p95 latency and throughput sparklines updated in real time
• Run inspector – a virtualized table of every job with status colors, timestamps, and full payload expansion without leaving the row; filter by status or search by ID
• Error triage – failures grouped by error class, ranked by frequency, with a 24-hour trend so you catch regressions the moment they spike rather than hunting through logs
• FlowProducer DAG – parent/child job relationships rendered as a real graph with per-node status and duration; drill into any node or replay subtrees
• Scheduler management – see active cron and delayed jobs, when they last ran, and what's next; pause, resume, or edit cron expressions without redeploying
• Enqueue from the UI – compose a payload, pick a queue, and fire it with ⌘↵; validates against your schema and shows the run immediately
• Editor integration – click any line in a failed job's stack trace and Workbench opens that file directly in Cursor or VS Code
• Full keyboard control – ⌘K to search, ⌥1–9 to switch queues, R to retry; every common action has a binding

The desktop app stores Redis credentials in the OS keychain and auto-updates via signed releases. The embedded version works with Hono, Express, Fastify, NestJS, Next.js, Nuxt, Bun, and a dozen other Node frameworks. It's a more focused alternative to something like Bull Board, with deeper BullMQ-specific features like flow visualization and scheduler editing built in from the start.

Licensed MIT. Apple-signed on macOS.