Build a Deployment Health Dashboard With ToolJet MCP

What Is a Deployment & Health Dashboard?

Engineering teams that run many services lose time when an alert fires, because on-call checks the metrics tool, the deployment pipeline and the incident tracker separately before knowing where to look. Here you build a deployment health dashboard with ToolJet MCP so one screen shows whether a problem began when something shipped. An agent generated the result as a structured ToolJet application, with data, queries and interface together, and the team keeps it running. You will see the timeline, the incident flow and the rollout page, and you can reproduce all of it from one prompt.

Page one: error rate and p95 per service on one 24-hour axis, with deploys as dotted lines and markers and incidents shaded across it, next to the open-incident list.

Page two: current rollouts with strategy, status, progress, health and any rollback note.

How to Find Which Deployment Caused an Outage

Find the deployment behind an incident by plotting error rate, response time, deploys and open incidents on one shared time axis so the order of events is visible at a glance. ToolJet MCP lets a coding agent build this as a working app with the data, queries and interface in place.

1. Collect error rate and response time for every service on one shared time axis
2. Mark every deployment on the timeline with what shipped, who shipped it and the commit range
3. Shade open incidents across the timeline so overlaps with deploys stand out
4. Show unhealthy and recently deployed services first and collapse the healthy ones
5. Acknowledge the incident and share a note with the affected services and recent deploys
6. Record any rollback with a reason and the person who made the call

How the Timeline, Deploy Markers and Incident Bands Work

The first page stacks two charts on a shared 24 hour axis, error rate on top and response time underneath, with one line per service. With 30 services on one axis the chart is unreadable, so only unhealthy, recently deployed and pinned services are drawn, and the rest collapse into a healthy count you can expand. Every deployment appears as a dotted line and a marker. Hovering a marker shows what shipped, who shipped it and the commit range, and clicking it filters the whole page to the hour before and two hours after that deploy. Incidents appear as shaded bands, and on-call acknowledges them from the list beside the chart.

• Error rate and response time per service on one shared 24 hour axis
• Deployment markers with hover details and a click that filters the page to that deploy
• Incident bands across the timeline and an acknowledge action that drafts an incident note
• Unhealthy and recently deployed services shown by default, with pinning for each engineer
• A rollouts page with a status filter and a rollback action reserved for release managers

Build a Deployment Health Dashboard With ToolJet MCP From One Prompt

Want to build it yourself? Start with the ToolJet MCP repository for setup instructions, supported agents, and everything you need to follow along.

The Consolidated Build Prompt

The real build took several passes, and the requirements are consolidated into one prompt that could have been used from the start. You can use it to reproduce the dashboard in a single shot.

Build a dense, dark service health app in ToolJet for the engineering team of a healthtech company running about 30 services, using ToolJet Database with sample data for services, metrics, deployments, incidents, pins and rollouts.

Keep it to two pages. Service health shows one chart with error rate stacked over response time on a shared 24 hour axis in Toronto time, one line per service, with every deployment marked by a dotted line and a marker. Hovering a marker shows what shipped, who shipped it and the commit range. Clicking it filters the whole page to the window from one hour before to two hours after that deploy, with a Clear window button. Rollouts lists current rollouts with strategy, status, progress, health and any rollback note, with a status filter and search.

By default show only unhealthy services, services deployed in the last 2, 4 or 8 hours, and each engineer's pinned services, up to three. Collapse the rest into a healthy count that expands to grey lines. Add an open incidents list and a services table under the chart, and a one line status bar instead of KPI tiles.

On-call engineers can acknowledge an incident, which opens an editable note listing the affected services, when the error rate started climbing and the deploys in the preceding hour. Confirming saves the note and copies it to the clipboard. Only release managers can mark a rollout as rolled back, with a required reason. All engineers can view the app. Create groups for engineers, on-call and release managers.

How ToolJet MCP Builds the Whole Application

ToolJet MCP works through the whole application rather than the screen, so the data model, the queries, the components and the wiring between them all land as one structured ToolJet application, as described in the ToolJet MCP documentation. Enterprise app building also covers data connectivity, workflows, permissions and ongoing change, and the generated application sits on a runtime that carries those where supported instead of handing the team a codebase to operate. From there you can keep using AI, switch to visual editing, or drop into code where a case needs it.

What Broke During the Build and How Was It Fixed?

Clicking a deploy marker at first did nothing. The chart reported the clicked time without seconds, while the lookup that finds the deploy used keys that included them, so no deploy matched. The lookup now compares the time to the minute and checks the series name, and the click filters the page as intended.

What the Finished Build Contains

The table shows what the finished build contains.

The components ToolJet MCP created, in the ToolJet inspector

Which Industries Can Use This App?

Any engineering team that runs many services and has to check a metrics tool, a deployment pipeline and an incident tracker separately when something breaks. Healthtech, fintech, e-commerce and SaaS platform teams with an on-call rotation fit the same pattern.

Who Can View, Acknowledge and Roll Back

Three groups were created: engineers, on-call and release managers, each with view access to the app. In the app, the Acknowledge button is only enabled for on-call members and the Mark rolled back column is only shown to release managers. The groups have no members yet. Restricting the acknowledge and rollback queries on the server needs a ToolJet plan feature that was not available on this instance, so those two rules are enforced in the interface only.

Enterprise Features for Your Deployment Monitoring Dashboard

The dashboard holds incident notes, deploy history and the right to roll a release back. ToolJet covers that governance at the platform layer, so you configure it once, not per app.

• SSO and SCIM: sign in with SAML, OIDC or LDAP, and provision users automatically
• Role-based access control: scope permissions to the app, the data source, and each query
• Audit logs: track every login, edit, and approval decision for compliance review
• Air-gapped deployment: self-host on Docker or Kubernetes so your data stays in your network
• Multiplayer editing: several builders work on the same app, with versioning and Git sync
• ToolJet AI inside your own deployment: run the AI features in your tenancy rather than a shared service

You could add a notification layer with ToolJet Workflows. For example, a workflow could fire when an incident is acknowledged, post the generated note to the incident channel in Slack, email the release managers through Gmail, and write the outcome back to sh_incidents.

Final Takeaways

The dashboard answers the question on-call actually asks: did this problem start when something shipped. Putting deploys, error rate, response time and incidents on one time axis removes the tab switching, and the default view keeps 30 services readable by drawing only the ones that need attention. On-call acknowledges an incident with a note already written, and release managers record a rollback with a reason. Because the result is a real, editable ToolJet application, the team can change thresholds, pages or access as needs change. You can build it yourself on [ToolJet Cloud].

Try ToolJet MCP

Build your own deployment health dashboard with ToolJet MCP, then request a ToolJet demo to map it onto your own services. Start free on [ToolJet Cloud].

FAQs

What is ToolJet MCP?

ToolJet MCP is a server that lets a coding agent create ToolJet applications from a written description, including the data model, queries, pages and components. The agent builds the app inside ToolJet, so you open and edit it like any other ToolJet app.

How does a deployment health dashboard show whether a problem started with a deploy?

Every deployment is drawn as a dotted line and a marker on the same time axis as error rate and response time. Hover a marker to see what shipped, who shipped it and the commit range, or click it to filter to the surrounding window.

Can I reproduce the deployment health dashboard from a single prompt?

Yes, the consolidated prompt describes both pages, the six tables, the timeline behavior and the access rules. Give it to an agent connected to ToolJet MCP and review the result in the ToolJet editor.

Do I need to write code to build the dashboard with ToolJet MCP?

No code is needed to describe or generate it. The timeline logic lives in a JavaScript query that you can edit, or you can leave it alone and adjust the pages visually.

Can I change the dashboard after ToolJet MCP generates it?

Yes, the result is a normal ToolJet application, so you edit pages, queries and styling in the visual editor. Service limits live in a database table, so you change which services count as unhealthy by editing rows.

Can the dashboard read my real metrics, pipeline and incident tools?

Yes, ToolJet connects to databases and APIs, so each feed can come from a connected data source instead of a table. You map each source's fields to the ones the queries expect.

What happens to the dashboard after the agent finishes building it?

It stays a structured ToolJet application on a runtime that handles data connections, roles and releases where supported. When requirements change, you or the agent edit the same application instead of operating generated code.