Workbench is now available for h3

Workbench now ships a first-party adapter for h3 — the minimal HTTP framework powering Nitro, SolidStart, Analog, and a growing slice of the unjs world. Drop the same open-source BullMQ dashboard you already get on the desktop straight into your existing h3 server, behind whatever auth and routing you already run.

No new processes to manage, no parallel Redis connections, no copy of your queue config drifting out of sync. The dashboard reads from the same Redis your workers do, mounts at a path of your choice, and gets out of the way the rest of the time.

Install in your h3 app

Run the CLI from your project root — it detects h3, installs @getworkbench/h3, and wires the mount into h3 app for you:

Prefer to wire it up by hand? Install the adapter and add the snippet below to the file you already use for your h3 server bootstrap:

import { createApp, toNodeListener } from "h3";
import { createServer } from "node:http";
import { Queue } from "bullmq";
import { workbench } from "@getworkbench/h3";

const emails = new Queue("emails");

const dashboard = workbench({
  queues: [emails],
  basePath: "/jobs",
});

const app = createApp();
app.use("/jobs", dashboard);
app.use("/jobs/**", dashboard);

createServer(toNodeListener(app)).listen(3000);

That's the whole integration. Visit http://localhost:3000/jobs and the dashboard renders with every queue you passed in. The adapter is published as @getworkbench/h3 and the runnable example lives at examples/with-h3 in the repo.

What you get

• Live counters, p50/p95 latency, and throughput sparklines per queue.
• Virtualised runs table with status filters, full payloads one click away, and keyboard-driven retry.
• FlowProducer DAG view for parent/child jobs, with per-node duration and status.
• Scheduler timeline for cron + delayed jobs — pause, resume, edit cron without a redeploy.
• Error triage grouped by exception class with 24h trend lines.
• Click any line in a failed job's stack trace to jump straight to Cursor or VS Code.

Why not bull-board for h3?

Bull Board is the dashboard most h3 teams reach for first, and for plenty of projects it's fine. The three places Workbench tends to win are coverage, polish, and operations — here are the high-order bits:

• A native desktop app, not just an embeddable UI

  Bull Board is a server-side dashboard you mount into your app. Workbench is both — embed it into your server (same one-liner), run the native macOS app pointed at your Redis URL with zero server changes, or deploy the standalone Docker image when you want a separate container.

• Wider framework coverage, with first-party adapters

  Bull Board ships adapters for Express, Fastify, Hapi, and Koa. Workbench adds Hono, Elysia, NestJS, AdonisJS, Next.js, TanStack Start, Astro, Nuxt, Bun.serve, and h3 — thirteen officially-supported integrations, each with a smoke-tested example app.

• Built for production triage, not just queue inspection

  Error grouping with 24h trend lines, FlowProducer DAGs, scheduler timelines, p50/p95 latency, and ⌘-click to jump from a failed job's stack trace into Cursor or VS Code. The UI is keyboard-driven end-to-end.

The short version for h3 users specifically: Bull Board does not ship an official h3 adapter, so you'd be on a community wrapper or bridging through Express. Workbench's @getworkbench/h3 is a first-party package with a smoke-tested example app and the same one-command installer as every other supported framework.

For the full side-by-side, see Workbench vs Bull Board.

Get started

Adapter docs and a complete runnable example live in the @getworkbench/h3 README. If anything trips up, open an issue on GitHub — every supported framework has a CI smoke test, so reproductions move fast.