Skip to content
Wardog

Trust // Security

Safe to leave running.

A system permitted to act inside production has to be engineered for restraint first. These are the controls that make WARDOG's autonomy reviewable.

Controls

Enforced in code, outside of any model.

  • Tenant isolation

    Scope is validated at the authority layer before any action can be executed against an asset.

  • Deterministic authority

    Policy evaluation is code, reviewable and testable, and defaults to deny.

  • Bounded blast radius

    Reach is computed and compared to limits before authorisation, not discovered afterwards.

  • Reversibility

    Reversible actions are preferred, and rollback paths are defined per action type.

  • Short-lived authorisation

    Signed tokens are scoped and expiring, so authority cannot be replayed later.

  • Complete audit

    Issue, refusal, execution, verification and expiry are all recorded against the incident.

Responsible disclosure

If you believe you have found a security issue in our website or product, contact us and we will respond. Please do not test against systems you do not own.

Report an issue

Reach us through the contact page and mark your message as a security report.

Talk to us