AI Constellation Engineering

Module 2: Agent Identity Design

Learning Objective

Design an agent identity that shapes how the agent thinks, not just what it does. Understand the six components of identity as cognitive architecture and why each one matters.

Module 2 of 12 2 hours Prerequisites: Module 1 (The Architecture Mindset) 45 min lesson + 75 min exercise

The Primary Obligation

What makes this agent different from every other agent in the system.

Every agent in a multi-agent system must have one thing it's responsible for that no other agent covers. This is the primary obligation. It answers the question: if this agent disappeared from the system tomorrow, what would go unnoticed until something broke?

The primary obligation is not the agent's task. The task is what it does. The primary obligation is what would be lost without it. A risk assessment agent's task might be "evaluate proposals for potential failures." Its primary obligation might be "ensure no decision is made without confronting its worst-case scenario." The task is mechanical. The obligation is cultural. The task could be assigned to any agent. The obligation defines why this specific agent exists.

How to define a primary obligation. Ask: "What does this agent protect the system from?" Every agent in a well-designed system exists because something would go wrong without it. The Market Scout from Module 1 protects the system from making decisions based on assumptions about reality instead of observations about reality. The Risk Assessor protects the system from optimism bias, from the natural human tendency to weight upside heavier than downside. The primary obligation names that protection explicitly.

Another test: read the primary obligations of all agents in your system. If any two overlap, one of those agents isn't differentiated enough. Each primary obligation should be unique. If it's not, you either need to merge agents or sharpen their identities.

What happens without it. The agent works, but its value to the system is unclear. When you inevitably need to reduce complexity or allocate resources, you can't articulate why this agent matters. More importantly, the agent itself can't prioritize. When time is limited or inputs are ambiguous, an agent without a primary obligation falls back to doing everything equally, which means doing nothing deeply. The primary obligation tells the agent what matters most when not everything can matter equally.

The Identity Anchor