Skip to main content
The Command Center is the first screen you see after logging in. It gives you a live view of everything happening across your environment — active incidents, agent activity, pending approvals, recent problems, and your system’s health pulse at a glance.

The ECHO orb

The large animated orb at the top of the Command Center is ECHO’s visual presence. The orb itself doesn’t change color for health — instead:
  • Severity-colored markers orbit the orb — one per pre-verification active incident (up to 8, with an overflow count beyond that). An incident that has reached verification or closure still shows in the incident list but gets no orbit marker, giving you an at-a-glance read of what’s still open and unresolved, and how severe.
  • The page takes on a subtle border glow keyed to your highest-severity incident that hasn’t reached verification yet. Incidents already in verification or closure don’t drive the glow — a board where the worst open incident is confirming recovery stays calm.
Below the orb, ECHO’s briefing card surfaces a short summary of your current incident landscape. This is a deterministic, client-side message templated from your incident counts and severities — not an LLM call — so it updates instantly as your incident landscape changes.

System Status panel

The System Status panel shows your key operational KPIs for the current environment:
  • Open — total open incidents (a plain count; there is no severity breakdown in this tile)
  • Today — an approximation of incidents resolved today, computed as total incidents minus currently-active incidents rather than a true rolling 24-hour window
  • MTTR — mean time to resolution, with a trend indicator vs. the prior period
  • Approval Rate — the percentage of incidents resolved, shown as a plain resolved-vs-total ratio (not specifically a measure of HITL approval behavior)
There’s no streak indicator on the Command Center currently — that metric isn’t available from the backend today, so it’s hardcoded to zero and the row never renders.

Active Now panel

The panel titled Active Now lists the first 4 open incidents for the selected environment, newest first. Each row shows a severity badge, the incident ID, any flapping/held badges, the incident title, and the current lifecycle stage — there’s no elapsed-time field on the row. When there are more than 4 open incidents, a View all N incidents button appears below the list and takes you to the full Incidents list. Clicking an incident row navigates to its Incident Hub where you can review agent analysis and take action.
The Active Now panel refreshes every 60 seconds. For a push-updated list, your incident Slack war-room receives real-time agent posts throughout the lifecycle.

Pending Approvals panel

When an AI agent reaches a decision point that requires human sign-off — for example, approving a proposed fix before it runs — an approval card appears here. The panel shows:
  • The incident the approval belongs to
  • The action being requested (e.g., “Execute rollback of payment-service to v1.4.2”)
  • The agent’s risk assessment
  • A Review button that opens a dialog with the full request, and an inline reject control for a quick dismissal without opening the dialog
Approvals you act on here are immediately reflected on the Incident Hub.
An approved action cannot be undone. Review the agent’s recommendation and risk level before approving. If you are uncertain, open the Incident Hub to read the full diagnosis context before deciding.

Agent Activity Feed

The Agent Activity Feed shows a live stream of what your AI agents are doing right now — across all open incidents. Each entry shows:
  • Which agent is acting (Incident Commander, Diagnosis Specialist, Resolution Specialist, Verification Specialist)
  • Which incident it belongs to
  • A short description of the current action
  • Timestamp
This feed auto-refreshes every 30 seconds. It is useful for confirming agents are progressing and spotting any stalls.
The Agent Activity Feed is controlled by an org-level setting and can be turned off for your organization. If you don’t see it, check with an administrator.

Recent Incidents, Recent Actions, and Recent Problems

Three compact panels below the main grid round out the dashboard:
  • Recent Incidents — your 3 most recently created incidents, regardless of current status. This is a quick way to jump back into something that just happened, not a “recently closed” list.
  • Recent Actions — a short history of completed approval decisions, so you can see what was recently approved or rejected without opening each incident.
  • Recent Problems — your 3 most recently created problem records, including resolved ones — not just active problems. This gives you a quick way to check whether a current incident has a linked problem record.

Recurrence Hotspots widget

The Recurrence Hotspots widget flags services with 2 or more problem records in the trailing 30-day window — regardless of whether those problems are formally linked into the same recurrence chain. A service appearing here is a signal that a structural fix may be warranted — click through to review the recurrence chain, which links out to the individual problem records and their root cause analyses.

Monthly Burn widget

For organizations on bring-your-own-key AI billing, the Monthly Burn widget shows LLM cost accrued this calendar month against your configured budget. The widget is visible to users with the settings:read permission.

Switching environments

Not every panel on the Command Center is scoped to the environment you have selected in the context switcher. Environment-scoped — these refresh to show the selected environment’s data: the System Status panel, the Active Now panel, Recent Incidents, and Recent Problems. Not environment-scoped — these show data across all your environments regardless of which one is selected: Pending Approvals, the Agent Activity Feed, Recent Actions, and the Recurrence Hotspots widget. Changing the environment doesn’t trigger a full page reload — the environment-scoped panels above refetch in place.

Real-time updates

The Command Center subscribes to real-time incident events. If the real-time feed drops, a full-width banner appears at the top of the page letting you know live updates are unavailable. Independent of that feed’s health, every panel also polls on its own schedule (roughly every 30–120 seconds depending on the panel) — so data keeps refreshing even when the banner isn’t showing. You do not need to refresh the page.