> ## Documentation Index
> Fetch the complete documentation index at: https://aidocs.ethanbragdon.icu/llms.txt
> Use this file to discover all available pages before exploring further.

# Control Center

> Operate Starfire AI through the platform administration and observability Control Center.

# Control Center

Control Center is the privileged operating surface for Starfire AI. Its purpose is to let authorized operators manage the platform without treating direct database edits, source-code changes, or ad-hoc production commands as normal administration.

## Control Center 4.0

The NEXUS-era Control Center expands the admin experience into a broader platform operating system.

The administration model spans areas such as:

```text theme={null}
Overview
Users & Identity
Organizations
AI / Models / Providers / Routing
Intelligence
Developer Platform
Billing
Infrastructure / Health
Security
Configuration
Audit
Incidents
Diagnostics
```

Individual surfaces can roll out independently during Alpha 1.6.3.

## Operating principles

### Least privilege

Administrators should receive only the platform permissions required for their job. Support access, billing access, model administration, and security administration do not need to be the same role.

### Audit important changes

Actions that affect users, plans, credits, permissions, models, platform configuration, or security should be attributable to the administrator who performed them when the active subsystem supports auditing.

### Prefer observable state

Use request IDs, run IDs, billing records, health checks, incident timelines, and diagnostic state instead of guessing from symptoms.

### Keep private content separate from operations

Control Center needs enough metadata to operate the platform, but operational administration should not become unrestricted browsing of private user conversations or artifact contents.

## Typical investigation flow

1. Identify the affected resource: user, organization, request, run, subscription, model, provider, artifact, or component.
2. Confirm the active account/organization and entitlement context.
3. Review recent state, logs, health, and audit information.
4. Determine whether the problem is local or platform-wide.
5. Apply the smallest administrative change needed.
6. Verify the resulting state.
7. Record or update an incident if the problem affects broader platform availability.

## Global search

The Control Center architecture includes a global-search/command-palette concept so administrators can locate operational resources by identifiers such as user email, request ID, run ID, artifact ID, subscription/price identifiers, or other supported administrative references.

## Control Center is not the user app

Organization administrators and developers should be able to manage their own appropriate resources without receiving platform-wide Control Center access.

<CardGroup cols={2}>
  <Card title="Users & Identity" icon="user-gear" href="/admin/users">
    Operate account state, sessions, access, credits, and user-level diagnostics.
  </Card>

  <Card title="RBAC & Support Mode" icon="user-lock" href="/admin/rbac-support">
    Understand granular admin permissions and safer support access.
  </Card>

  <Card title="Models & Routing" icon="route" href="/admin/models-routing">
    Manage providers, health, fallback, and routing controls.
  </Card>

  <Card title="Intelligence" icon="brain" href="/admin/intelligence">
    Operate research, builds, artifacts, Knowledge, retrieval, and tools.
  </Card>

  <Card title="Billing" icon="credit-card" href="/admin/billing">
    Reconcile customers, subscriptions, plans, credits, and usage.
  </Card>

  <Card title="Security & Audit" icon="shield-halved" href="/admin/security-audit">
    Investigate sessions, events, admin actions, abuse, and risk signals.
  </Card>
</CardGroup>