Command
Read current attention and posture, then open the workspace that owns the next action.
Command is the authenticated landing workspace at /dashboard. It summarizes tenant activity and routes operators into the workspace that owns each record.
What it owns
Command owns the cross-product overview. Source records remain authoritative in Agents, Processes, Decision Desk, Incidents, Assurance Center, Audit Trail, and Control Mapping.
- A priority exception route names the single most urgent item and the workspace that owns it.
- A posture summary counts pending Decision Requests, policy changes, open incidents, and recent runs.
- Recent activity and recommended applications open the owning surface.
- Global search reaches records across workspaces.
- Workspace shortcuts preserve available record context when opening an owning surface.
Command is one page. It has no tabs.
Command has general authenticated access. Each destination applies its own route and resource permissions.
Attention lifecycle
flowchart LR S["Signal"] --> C["Command triage"] C --> D["Decision Desk"] C --> I["Incidents"] C --> A["Assurance Center"] D --> E["Evidence ready"] I --> E A --> E
Command is an aggregator. Decisions, incident transitions, remediation changes, and evidence creation run through the owning APIs.
Analyst availability
The authenticated shell includes the Analyst launcher and drawer when the model runtime is enabled and the principal holds assistant:chat. Installations without a model provider show “Analyst is disabled” at /assistant and omit the launcher and drawer. Policy enforcement, approvals, and evidence remain available in this installation mode.
Failure and access behavior
Loading, empty, unavailable, and stale states are shown independently for each data source. A destination can still deny access when the current user lacks its permission.
