METHODOLOGY AND EVIDENCE

Every product claim should be inspectable.

How MandateShield turns protocol evidence into deterministic controls, tests behavior and separates what is verified from what is not.

Who publishes this material

MandateShield publishes and maintains this product, its open execution-boundary contract, security documentation and developer resources. Product pages are written from the behavior of the running implementation, not from customer testimonials or fabricated deployment claims.

How the controls are derived

The engine normalizes payment-relevant facts that recur across agent-commerce protocols: human mandate, amount, currency, merchant, agent identity, consent, timestamps, intent binding and replay identity. Protocol-native signature verification remains the responsibility of the originating protocol or payment provider.

How behavior is tested

Automated tests cover purchases inside authority, overspend, merchant substitution, prompt-injection language, expired authority, missing consent, incomplete evidence, future timestamps, invalid amount precision and ordinary two-decimal amounts. Production health checks separately report policy engine, database and billing configuration.

What MandateShield does not claim

  • It does not execute or process payments.
  • It does not receive card or bank credentials.
  • It does not replace fraud, sanctions or payment-network controls.
  • An ALLOW decision is not a guarantee that a transaction is legitimate.
  • The open boundary is not represented as an accredited industry standard.
  • Revenue, savings and loss avoidance are never guaranteed.