---
title: "A Permit Is Not Authority | KYE Protocol™ blog"
description: "A permit says the action is allowed. Authority says this principal may take it. Where runtime guardrails stop and Authority Finality™ begins."
url: https://kyeprotocol.com/blog/a-permit-is-not-authority/
lang: en
source: "KYE Protocol"
---

> A permit says the action is allowed. Authority says this principal may take it. Where runtime guardrails stop and Authority Finality™ begins.

18 July 2026 · KYE Protocol™ · Doctrine · ― Is Not Authority

# A Permit Is Not Authority.

A runtime guardrail can allow or block an agent's action in milliseconds. Useful — but a permit checks the action, not the right to take it.

[See action admissibility](https://kyeprotocol.com/action-admissibility/) [More doctrine](https://kyeprotocol.com/blog/)

A permit says this action is allowed. Authority says this principal may take it, here, now, for this purpose.

## What a permit gives you

An action firewall enforces a rule at the moment of the call.

- Allow, block, or ask on a tool call or side effect.
- Fast policy checks against the shape of the action.
- A hard stop for actions no one should ever take.

This is real enforcement, and every serious agent needs it. But a permit reasons about the action in front of it — not about who is behind it.

## Where it stops

A permit answers "is this action allowed?" — never "may this principal take it?"

The same wire transfer can be admissible for a treasury agent under one delegation and forbidden for a support agent under another. A permit that only inspects the action treats both callers alike; it has no model of the delegation chain, the purpose, or the expiry behind the request. Allowed-in-general is not authorised-for-this-actor, and the gap between them is exactly where an over-scoped agent slips through.

- It scopes the action; it never scopes the actor.
- It is stateless about delegation — no grant, no purpose, no expiry.
- A permitted action by the wrong principal is still a breach.

## The line

Authority binds the action to the principal's actual delegation, at the boundary.

KYE Protocol™ resolves who the actor is acting for, what authority that principal delegated, and whether this action falls inside its purpose and scope before returning admit or deny — then seals the verdict into a replay-verifiable [Evidence Pack™](https://kyeprotocol.com/evidence-pack/). That is Authority Finality™: a permit clears the action, and Authority Finality™ decides whether this actor was ever entitled to attempt it.

## Honest scope

KYE Protocol™ proves authority, not safety.

It does not replace your guardrails; a dangerous action should still be blocked by a permit regardless of who asks. What KYE Protocol™ adds is the authorisation verdict a firewall cannot form — bound to the principal, provable after the fact. Keep the permit. Add the authority.

> A permit says this action is allowed. Authority says this principal may take it, here, now, for this purpose.