# Operate a Mind

Keep configuration, recovery, and observed behavior understandable.

Source: https://docs.minds.sh/docs/operations



An operationally useful Mind needs more than a responsive dashboard. Know which revision is running, how it is configured, how requests are authorized, and how you would recover its data.

## Start with these guides [#start-with-these-guides]

<Cards>
  <Card title="Configuration" href="/docs/operations/configuration">
    Understand service groups, listeners, data storage, and authority.
  </Card>

  <Card title="Backups and recovery" href="/docs/operations/backups">
    Separate data backups from runtime snapshots and learning checkpoints.
  </Card>

  <Card title="Troubleshooting" href="/docs/operations/troubleshooting">
    Diagnose failures by layer instead of repeating the same request.
  </Card>

  <Card title="Compatibility and launch status" href="/docs/operations/compatibility">
    See what was source-reviewed, observed live, and still needs verification.
  </Card>
</Cards>

For managed instances, use supported platform operations rather than modifying a provider resource behind the platform's back. For self-hosted instances, retain control of the deployed image, configuration, credentials, and data paths.