Security Layer

The Security Layer is the enforcement engine of the protocol. When an AI agent proposes a transaction, it must be validated by a decentralized network of nodes known as Policeman Nodes. These nodes act as co-signers on the agent’s wallet, verifying that every proposed action is both compliant and reasonable before it is executed.

The verification process is twofold. First, each Policeman Node performs Policy Verification, a deterministic check to ensure that the action does not violate any of the rules defined in the Identity Layer. If the proposal passes, it then undergoes Logic Verification, where a constrained large language model evaluates the reasoning behind the action. This model checks for logical coherence, alignment with user goals, and awareness of potential risks. Only if both checks succeed, and a quorum of nodes agree, is the transaction approved.

Transparency is fundamental to the Security Layer. Every validator’s decision is committed on-chain and later revealed in full, allowing anyone to audit their work. This radical openness ensures that nodes remain accountable, while also enabling community auditors to verify the correctness of every decision.