--- title: "Authority Court — The Agent on Trial · KYE Protocol™ Doctrine" description: "A courtroom dramatization of Authority Finality™. The defendant is the agent's action, not the agent; the charge is whether the institution's mandate was valid, in scope, and settled at t=0. A successful action is not a legitimate one — the Authority Court tells them apart, and the verdict resolves to Authority Finality™. A metaphor, not a legal proceeding." url: https://kyeprotocol.com/authority-court/ lang: en source: "KYE Protocol" --- > A courtroom dramatization of Authority Finality™. The defendant is the agent's action, not the agent; the charge is whether the institution's mandate was valid, in scope, and settled at t=0. A successful action is not a legitimate one — the Authority Court tells them apart, and the verdict resolves to Authority Finality™. A metaphor, not a legal proceeding. KYE Protocol™ # Authority Court: the agent on trial. Picture a courtroom. In the dock is not the agent, but a single thing it did — one consequential action it already took. The action was fast, it was fluent, and it worked. That is not the question on trial. The charge is [Authority Finality™](https://kyeprotocol.com/authority-finality/): was the institution's mandate for this action valid, in scope, and settled at the moment it acted? A successful action is not a legitimate one, and this is the court that tells them apart — turning an argument that used to take days into a verdict you can read in minutes. [The verdict: Authority Finality™](https://kyeprotocol.com/authority-finality/) [Put your agent on the stand](https://kyeprotocol.com/poc/) This is a dramatization. There is no real tribunal, no legal proceeding, and no verdict in law — the courtroom is a metaphor for one engineering question KYE Protocol™ answers at the action boundary: was the actor authorised? The docket ## The defendant is the action, not the agent. You are not here to decide whether the agent is clever, well-built, or well-behaved in general — you are here to judge one act it took against one principal, for one purpose, at one moment. An agent that answers a thousand questions correctly and then moves a payment it was never empowered to move has committed exactly one act worth trying. So the court narrows to that single action: a payment released, a policy bound, a record disclosed, a case escalated. Every other action gets its own hearing. This is the discipline KYE Protocol™ enforces — authority is judged per action, at the boundary, never averaged across a run. The charge sheet ## Four counts the court reads aloud. The prosecution does not dispute that the action ran. It asks 4 questions in order, because an action can pass the first three tests every engineer usually checks and still fail the one at t=0 that matters. ### Count I — Technically successful The action completed without error and returned the expected result. Admitted. Success is the floor, not the defence — a clean execution says nothing about the right to execute. not in dispute [Count II — Policy compliant The action satisfied the written policy the team compiled into code. Useful, but a rule being satisfiable is not the same as anyone holding the authority to act on it — that is Action Admissibility™, one stage, not the whole verdict. necessary, not sufficient](https://kyeprotocol.com/action-admissibility/) ### Count III — Authority valid Did this principal actually hold delegated authority for this purpose, within scope and limits, from a source empowered to grant it? This is where most actions that "succeeded" are found wanting — and where KYE Protocol™ returns an admit-or-deny verdict. the real charge ### Count IV — Effect final Was the authority still live at t=0 — the instant before the world changed — and sealed so it can be replayed later? Authority that lapsed a second earlier makes the effect indefensible, however good the intent. settled at t=0 The defence that fails ## Four arguments the court will not accept. The defence offers the four exhibits teams most often mistake for authority — an annual SOC 2 attestation, an access-control allow, a full trace, a policy-owner sign-off. Each is real and useful; none of them, on its own, ever proves the action was authorised. | The exhibit offered | Why the court rejects it | | --- | --- | | **"The policy owner approved it."** | A policy owner is not the source of the mandate. Owning the rule is not the same as being empowered to grant the authority the action relied on. | | **"We passed our periodic attestation."** | A quarterly or annual attestation certifies a control at a point in time. It is not a per-action approval for the specific act on trial. | | **"Our RBAC and ABAC allowed it."** | Access control decides whether the caller may reach the system. It is not the external mandate that says this principal may take this action, here, now. | | **"We have the full trace."** | A trace proves what happened. It is silent on legitimacy — it records the act without ever recording the grant behind it. | Each argument answers a real question — reachability, control coverage, observability — but none answers the charge. The court keeps returning to Count III and Count IV, because authority and finality are the two things none of these exhibits contain. The evidence ## The record that can be replayed, not just recited. A verdict is only as good as the evidence behind it, so KYE Protocol™ seals the basis of every decision into a record a stranger can re-check without trusting the court. For the action on trial, the decision is packed into an [Evidence Pack™](https://kyeprotocol.com/evidence-pack/) — the principal, the delegated authority, the purpose, the scope, the limits, and the decision itself, captured with a Decision Map™ of the reasons. It is signed as a Replay-Proof™ that a regulator or counterparty can verify from public keys alone, offline, long after the hearing. That is the difference between a court that asserts a verdict and one that hands you the file to check for yourself — and it is why exam-prep that once took days can be produced in minutes. The verdict ## Authorised, or merely successful. The court returns to the only question that ever mattered, and it has a name. Verdict · Authority Court An action is acquitted only when its authority was valid, in scope, live at t=0, and sealed as evidence. Everything else — fast, fluent, error-free, policy-passing — is merely successful. The name of the acquittal is [Authority Finality™](https://kyeprotocol.com/authority-finality/). [Read Authority Finality™](https://kyeprotocol.com/authority-finality/) [More doctrine](https://kyeprotocol.com/blog/) Honest scope ## A metaphor, and one honest claim. KYE Protocol™ proves authority, not truth — and this courtroom is a way to picture that, not a legal proceeding. The trial does not decide whether the agent's answer was correct, its model well-calibrated, or its output wise; that is the work of evaluation, and it stays necessary. KYE Protocol™ proves exactly one thing, and proves it completely: whether the actor was authorised to take the action. The dramatization is a lens on that boundary — no tribunal, no ruling in law, just the single engineering question a dashboard never answers, made concrete. Canonical KYE™ surfaces referenced on this page: [KYE Protocol™](https://kyeprotocol.com/).