SafeSpark™By 4P3X AI™ · Variant Hub
Security and data

Security, data and production-hardening requirements

Current local-first security design, data boundaries and production security requirements for the SafeSpark™ Hub.

Public development demonstration using fictional school and learner data

Current demonstration design

  • Local-first browser storage and offline service-worker caching.
  • Separate role-App manifests and narrow scopes.
  • Fictional data and data-minimisation boundaries.
  • Local JSON export and controlled import demonstrations.
  • No public API keys, secret tokens or production credentials in the supplied source.

Current limitations

The local owner PIN is a browser-device demonstration control, not production authentication. Public data is not production encrypted because no real pupil data is permitted. The build has not been penetration tested or independently security certified.

Production requirements

  • Authenticated identities, least-privilege access and secure session management.
  • Encryption in transit and at rest.
  • Central audit logging, monitoring and incident response.
  • Backup, recovery, retention and deletion controls.
  • DPIA, threat modelling, code review and penetration testing.
  • Secure hosting, patching and vulnerability management.

Service-worker boundary

Each installed role App uses a narrow scope and unique cache. Cache activation removes only obsolete caches from the same SafeSpark™ App family. It does not clear localStorage, delete IndexedDB databases or unregister unrelated Apps.

Security position

No claim is made that the Hub is penetration tested, Cyber Essentials certified, ISO certified, fully secure or zero risk.

Adaptive-data boundary

Subject-scoped local profiles

The current 4PEX AI™ prototype writes a nested adaptive profile into each existing SafeSpark™ subject storage object. Maths cannot read the Spanish, French, Advanced Maths or Technology profile through the agent. The learner can reset adaptive preferences without deleting the surrounding progress record.

Production use would require authenticated identity boundaries, encryption, retention and deletion controls, audit logging, backup and recovery, security testing and organisation-approved rules governing who may inspect or change adaptive settings.