I built a forensic debugger for Azure Service Bus dead-letter queues — here's how it works

It's 2 AM. Your monitoring fires. 5,000 messages in the Dead-Letter Queue.
You need to know three things immediately:

• What's actually inside those messages
• Which failure pattern caused this
• Which messages are safe to replay right now

That investigation used to take me 6 hours. It now takes 45 minutes. Not because I got faster — because I built the right tool.

ServiceHub is an open-source, self-hosted web application that treats dead-letter queue debugging as the forensic investigation it actually is.

Here's what it does that a standard management console cannot:
Full message body inspection — click any message, active or dead-letter, and see the complete JSON with syntax highlighting, every system property, and the exact broker error description.

Client-side AI pattern detection — clusters your DLQ messages into error groups with confidence scores. Runs entirely in your browser. Zero data exfiltration. No API key needed.

Auto-replay rules — define conditions, set rate limits, add exponential backoff. The engine runs autonomously and shows live Pending → Replayed → Success counters.

30-day DLQ Intelligence — persists every scan to local SQLite. Know whether tonight's spike is a new incident or the same pattern from last Tuesday.

Correlation Explorer — paste any Correlation ID and trace that message across every queue, topic, and namespace in one search.

The entire thing is read-only by design. Uses PeekMessagesAsync — your consumers are never affected. Works with Listen permission only.

bash
git clone https://github.com/debdevops/servicehub.git
cd servicehub && ./run.sh

One command. Auto-installs .NET 10 and Node.js.

Open http://localhost:3000.

Or try the hosted demo in 30 seconds — no install:
👉 https://app-servicehub-prod.azurewebsites.net/

The hosted version authenticate via Microsoft Entra ID (Azure AD). No user database. No personal data stored. Connection strings AES-GCM encrypted at rest in your session.

What does your current DLQ investigation workflow look like? Drop it in the comments — I'm genuinely curious what the most painful part still is.

⭐ GitHub: https://github.com/debdevops/servicehub — MIT license, PRs welcome.