Morning Run for Domotz

A shareable workflow that turns your Domotz environment into a one-prompt morning operations dashboard. Built for MSPs, internal NetOps teams, and anyone running Domotz at scale.

You ask Claude to "run it." Claude executes ~18 health checks against your Domotz MCP server, then renders a single screenshot-ready HTML dashboard you can scan in two minutes.

This package contains everything you need to set it up.

What you get

1. morning-run-prompt.md — the system prompt that tells Claude what to check, how to triage, and how to render the dashboard. This is the heart of the workflow.
2. morning-run-setup.md — three ways to install it: paste-it (works today, zero setup), Claude Project (recommended daily flow), and Claude Code slash command (for terminal users).
3. This README — what the package is, what it does, what you need.

What it actually does

When you run the prompt against your Domotz environment, Claude executes 18 checks organized into three tiers:

Tier 1 — Critical (act now): offline important devices, active critical alerts, unhealthy collectors, long-running unresolved alerts, credential failures on managed infrastructure, new or unrecognized devices.

Tier 2 — Operational (act this week): stale data on devices that appear healthy, orphaned alert rules, stale or failed config backups, important devices with no alert coverage, SNMP coverage gaps, devices left in unmanaged state.

Tier 3 — Strategic (act this quarter, or sell as a service): sites without baseline monitoring, customers with alert noise that needs tuning, end-of-life or out-of-date firmware, sites without config backup enabled, tag and documentation gaps, fleet trend changes.

The output is a single HTML artifact with:

• A header showing date, time, organizations scanned, and device count
• A priority strip at the top with the 3 to 5 things actually worth acting on today
• Eighteen color-coded check cards organized by tier
• A "billable opportunities" strip near the bottom, surfacing findings you could turn into customer conversations, QBR slides, or new recurring revenue
• A timestamp and total runtime

Total runtime is typically 60 to 120 seconds, depending on the size of your environment.

What you need before you start

1. A Domotz account with at least one collector and some devices under monitoring. The richer your environment, the more the dashboard surfaces.
2. Access to Claude on one of these surfaces:
   • claude.ai (web)
   • Claude Desktop (macOS or Windows)
   • Claude Code (CLI, for the slash-command flow)
3. The Domotz MCP connector enabled in your Claude environment. Setup is in Claude under Settings → Connectors → search "Domotz." You'll be prompted to authorize against your Domotz account.

That's it. No scripts to install, no API keys to manage, no servers to host. The MCP handles authentication and tool exposure.

How to use this package

Open morning-run-setup.md and pick a setup level:

• Level 1 (paste it): 0-minute setup. Works in the next 5 minutes. Good for trying it once before committing.
• Level 2 (Claude Project): 10-minute setup. The right daily home. Open the project, type "run it," done. This is what most people should do.
• Level 3 (Claude Code): 15-minute setup. Turns the morning run into a single typed slash command in your terminal. Worth it if you're already living in a CLI.

Each level is fully self-contained in the setup guide. Pick one, follow it, verify the dashboard renders, move on.

Customizing it for your environment

The prompt is intentionally generic so it works for any Domotz tenant on day one. Once you've run it a few times, you'll know what to tune.

Common customizations:

• Scope to specific organizations. If you don't want every org scanned, edit the Scope section in the prompt.
• Add a check. Insert a new numbered entry under the right tier. Match the existing format: question, Domotz tools to call, expected output.
• Remove a check. Delete the entry. The dashboard renders whatever checks you keep.
• Change tier definitions. If "critical" means something different in your environment (e.g., you treat all customer-facing services as Tier 1), edit the tier descriptions.
• Adjust the billable opportunities strip. Tune it to match the services you actually sell.

The setup guide has a "Tuning the prompt over time" section that walks through this in more detail.

What this looks like in practice

Most teams who run this end up using it in one of three ways:

1. Solo MSP owner: open Claude every morning, run it, triage, close the chat. Two minutes from coffee to "I know what's broken."
2. NetOps team lead: run it before standup, screenshot the priority strip into Slack, walk the team through the day.
3. Customer-facing engineer: run it before a QBR with a customer, use the billable opportunities strip as the agenda.

Nothing in the workflow is locked to one of these. It's the same prompt, the same dashboard, the same dependencies.

A note on data and privacy

Everything happens in your Claude session, against your Domotz MCP connection, scoped to your authenticated account. No data is sent to third parties. The dashboard is rendered locally as an HTML artifact in your chat. Nothing is persisted unless you save the artifact yourself.

If you're sharing dashboards externally (screenshots in customer emails, exports into QBR decks), the prompt does not redact device names, IPs, or hostnames. Strip or rename what you need to before sending.

Where this came from

This workflow was built by a network engineer running a real MSP, then generalized so any Domotz customer could use it. It's deliberately simple: one prompt, one dashboard, no infrastructure. The leverage comes from asking the right 18 questions in plain English instead of clicking through 18 dashboard views.

If you build something better on top of it, share it back. The model this is built on (an LLM with MCP access doing senior-engineer triage on a fleet) is just starting to be explored, and every operator using it differently teaches the rest of us something.

Quick start, if you want to try it right now

1. Open Claude (claude.ai or Desktop).
2. Confirm the Domotz connector is enabled (Settings → Connectors).
3. Open morning-run-prompt.md in this package.
4. Copy everything between the SYSTEM PROMPT and END OF SYSTEM PROMPT markers.
5. Paste into a new Claude chat. Send.
6. Wait ~90 seconds. The dashboard appears as an artifact.

If it works, go install Level 2 in the setup guide. You'll save the copy-paste step every morning.