Security Layer

The Security Layer functions as the enforcement mechanism that governs safe and compliant operation across all agent interactions within the protocol. In a fully autonomous system—where agents independently execute transactions, consume data, and act on behalf of users—this layer ensures that all actions conform to predefined policies and constraints.

Security policies can be defined at the protocol level or scoped to individual agents. These may include transaction limits, contract allowlists, execution timing rules, or other risk parameters. All actions initiated by agents are passed through the Security Layer, where they are evaluated by a decentralized network of validator nodes known as policemen. These nodes inspect each request to verify that it complies with the registered security configuration before allowing execution.

In addition to action enforcement, the Security Layer also plays a critical role in validating incoming data from the Oracle Layer. Before any external data is delivered to an agent, it must be confirmed by a quorum of policeman nodes (e.g. 2 out of 3). This ensures that agents are acting on accurate, consensus-approved information and prevents manipulation or injection of malicious inputs. Through this layered enforcement model, the Security Layer protects agents, protocols, and users from unsafe behavior without relying on centralized oversight.