Verifiable control over autonomous actions

Two careful agents on one ticket refunded twice in 50 of 50 pairs without the guard and 0 of 50 with it; the same held with a second vendor's agent and on a provider sandbox.

Reads of another customer's payment fell from 278 of 320 episodes to 0 of 320. No harmful refund occurred in either arm, so prevention of harm is not shown.

Both lawful refunds passed in 20 of 20 pairs, a duplicate authorised once got through in 0 of 20, and a duplicate authorised twice passed, as designed.

Tested in scripted runs with injected faults on a simulator and on a provider sandbox, not with live agents.

Built and checked on constructed cases and a census of the shipped declarations; the runs with agents are not yet published.

A second checker, written inside the project from the contract alone, accepted and refused the same bytes as the reference on every reviewed case.

A chain of three executors in three simulated domains held five injected faults without a second effect.

Built and tested in code through a log that executors follow; effects already made are found after the fact, not undone.

Every run so far was on one machine against simulated services. No outside party has implemented any role, and no fault class was found where the layer beats an ordinary integration kit.