Verax AI system defaults to denying requests with no explicit rules

Verax, an AI security system, operates on a default-deny policy for any request without a specific rule. Its initial policy explicitly permits only four named tools, blocking all others including those on connected servers. Refused attempts are cryptographically signed and logged locally, providing a clear audit trail of blocked actions. The system is designed so unauthorized attempts cannot bypass rules by using different tool names. However, the logging and signing mechanism has not yet undergone an independent security audit.