Zero Trust Security Model
Relevant source files
The following files were used as context for generating this wiki page:
The Zero Trust Security Model in Nexus Orchestrator is a core pillar of the system's Governance Layer. It is designed to secure interactions between AI agents and operating environments by eliminating implicit trust. Instead of relying on traditional perimeter-based security, Nexus treats every agent and request as a potentially untrusted non-human identity that must be explicitly verified and authorized before accessing infrastructure resources.
This model functions as an autonomous safeguard within the "nervous system" of the orchestrator, synthesizing Open Policy Agent (OPA) governance with identity tracking to bridge the security gap between autonomous agents and the environments they manage.
Sources: README.md:1-12, README.md:18-18
Governance Layer Architecture
The security model is integrated into the Governance Layer, which works alongside the Connectivity and Orchestration layers. It specifically addresses the security requirements of non-human identities (AI agents) by enforcing policies via OPA.
Core Components
| Component | Description |
|---|---|
| OPA (Open Policy Agent) | The policy engine used to evaluate authorization requests based on predefined Rego policies. |
| Trust Registry | Managed via context/AGENTS.md, this component tracks known agents and their assigned trust levels. |
| Identity Provider | Handles the identification of non-human entities interacting with the daemon. |
Sources: README.md:18-18, context/AGENTS.md:1-10
Identity and Trust Management
Nexus maintains a persistent state of all known AI agents to facilitate the Zero Trust model. Every agent interacting with the orchestrator is assigned a unique identifier and a specific trust level, which determines the scope of its permissions.
Agent Tracking Schema
The system uses a structured format to track identities, ensuring that only registered agents can perform actions.
The diagram illustrates the relationship between the central Orchestrator and the individual agent identities it tracks.
Sources: context/AGENTS.md:5-10, README.md:18-18
Request Authorization Flow
When an AI agent attempts to interact with infrastructure—such as managing SSH connectivity or context files—the Zero Trust model intercepts the request. The workflow ensures that the agent's identity and the specific action are validated against the OPA policy engine.
The sequence diagram shows the mediation process where the Governance Layer validates requests before they reach the execution environment.
Sources: README.md:7-12, README.md:18-18
Policy Enforcement Logic
The security model relies on the principle of "never trust, always verify." This is implemented by checking the Trust Level defined in the agent registry against the sensitivity of the requested infrastructure operation.
- Identity Verification: The system checks if the
Agent IDexists in the local context registry. - Policy Evaluation: OPA evaluates if the agent's current
Trust Levelis sufficient for the target resource. - Contextual Integrity: The
context/AGENTS.mdfile acts as a local-first source of truth for these security decisions.
Sources: README.md:7-7, context/AGENTS.md:1-10
The Zero Trust Security Model ensures that the Nexus Orchestrator remains a secure "nervous system" for AI development, preventing unauthorized agents from performing potentially destructive infrastructure operations while maintaining autonomous efficiency.