Command Center Dashboard
Use one read-only command to orient yourself: what ACC currently observes, what needs attention, which evidence exists, and what to inspect before a governed change.
The 1.2.x CLI context includes managed-server, runtime-health, incident and recovery projections. The repository website Command Center was completed after the sealed GA source tag and is a read-only presentation of generated evidence, not a hosted monitoring service or an operational control surface. Missing or invalid evidence remains unavailable; freshness and source identity must be inspected. The marketing Hero uses illustrative data.
What the dashboard answers
The dashboard is a concise view over the canonical dashboard context and state pipeline, not a separate source of truth.
What is the current state?
The top summary reports the observed overall state and health score, followed by runtime, cognitive, and recovery status.
What needs attention?
Active incidents are promoted ahead of background detail. A healthy snapshot says that no operator action is required for that observation.
What evidence is available?
The evidence section names incident, recovery, and event context available to support the displayed result.
What should I do next?
The final section recommends a non-mutating next step: inspect the relevant evidence before recovery or change.
Historical 1.1.0 formatter example
This example is the formatter output exercised by the v1.1.0 product-experience test. Values are point-in-time observations, not permanent guarantees.
ACC COMMAND CENTER
Observed 2026-08-25T00:00:00+00:00
State HEALTHY
Health 100/100
CURRENT SYSTEM STATE
Runtime HEALTHY
Cognitive ACTIVE
Recovery READY
ATTENTION
No active incidents. No operator action required.
EVIDENCE
Incident runtime-incident-test · RECOVERED
Recovery runtime-incident-runtime-mutation-demo · Lifecycle Complete
Events 5
SAFE NEXT STEP
Inspect runtime evidence before any mutating or recovery operation.Run the current view with acc dashboard. The command exits zero when rendering succeeds; interpret the state and evidence inside the output rather than treating exit zero as a healthy-system verdict.
How to interpret the sections
The order is deliberate: state, attention, evidence, then action.
- Check freshness. Start with
Observed; a past result may no longer describe the system. - Read attention before acting. A degraded context surfaces active incidents and recovery need.
- Follow the evidence identifiers. Use runtime and recovery commands to inspect the underlying records.
- Keep the next step safe. The dashboard recommends inspection; it does not silently approve or perform mutation.
Dashboard boundaries
A clear summary is useful only when its authority is equally clear.
- It does not continuously monitor the host or replace a metrics, logging, tracing, or alerting platform.
- A
HEALTHYsnapshot means the supplied context evaluated healthy; it is not a security certification or availability guarantee. - The displayed score is derived from current dashboard state. It is not an SLA, risk probability, or prediction.
- The dashboard does not grant approval, authorization, access, recovery, execution, or deployment authority.
Read-only server observation in 1.2.x
The pilot requires a non-root, least-privilege SSH user, an explicitly pinned host fingerprint, and dedicated known-hosts and identity files outside the repository with safe permissions. Trust is recorded as operator-approved TOFU with independently_verified=false; matching a pin does not upgrade it to independent verification.
Only built-in identity, inventory and health probes are accepted. Their normalized evidence supports server intelligence and dashboard context; arbitrary remote commands, service restart, mutation and recovery authority remain blocked. Offboarding disables future probes while preserving evidence. Follow docs/product/production-server-onboarding.md for the reviewed operator runbook.
Move from summary to evidence
Use the runtime guide to inspect governed records or the access-control guide to understand protected decision paths.