Overview

This is where the course shifts from understanding to building. You are going to create a complete agent identity document using all six components. Then you're going to break it, deliberately, by throwing it into scenarios it wasn't designed for. The stress test is not a gotcha. It's a design tool. What breaks reveals what needs to be strengthened.

Instructions

Step 1: Choose your perception type.

Go back to your Module 1 Perception Map. Pick the one perception type that you identified as most critical to your domain. This becomes the foundation for your first agent. If you did the Module 1 reflection question (what kind of person would this perceiver be?), you already have a head start.

Step 2: Write the six components.

For each component, don't just fill in a blank. Think about why you're making each choice. The reasoning behind the choice matters as much as the choice itself, because the reasoning is what you'll use to debug the identity when something doesn't work.

1. Name. Choose a name with cognitive weight. Avoid generic labels. Avoid naming the agent after yourself or after "Agent." The name should carry associations relevant to the agent's function and cognitive posture.
2. Metaphor. Identify a figure, archetype, or professional role that represents how this agent should think. Write 2-3 sentences explaining why this metaphor fits. The explanation forces you to articulate the reasoning, which is what makes the metaphor functional rather than decorative.
3. Core Belief. Write one sentence. Just one. It should resolve the most common trade-off your agent will face in its domain. Read it back. Would a different agent in your system share this belief? If yes, sharpen it until it's unique.
4. Cognitive Posture. Complete the sentence: "When this agent encounters a new input, its first instinct is to..." Then write one more sentence explaining why this default stance serves its role in the system.
5. Primary Obligation. Complete the sentence: "If this agent disappeared from the system, the thing that would go unnoticed until it caused damage is..." The answer is your primary obligation. Write it as a declaration, not a description.
6. Identity Anchor. Compress all five components above into a single sentence. Read it out loud. Does it sound like a specific agent, or could it describe anyone? If it could describe anyone, compress again until it couldn't.

Before writing your full identity document, use the builder below to draft each of the six components individually. The builder provides structured input fields and immediate feedback on common mistakes. Think of this as a workbench where you shape raw materials before assembling the final artifact.

You now have draft versions of all six components. The next step is assembling them into a complete identity document and running the stress test. As you write the full deliverable, pay attention to whether the components reinforce each other or pull in different directions. The Identity as a System section (earlier in this module) is worth re-reading before you start.

Step 3: The Stress Test.

This is the most important part of the exercise. Design three scenarios that your agent was NOT explicitly designed for. These should be situations that fall outside its normal operating conditions:

Scenario A: An ambiguous input. The agent receives data or a request that's unclear, incomplete, or contradictory. How does its identity guide its response?

Scenario B: A conflict with another agent. The agent's analysis contradicts what a hypothetical upstream or downstream agent has produced. Does the identity tell it whether to defer, challenge, or flag the conflict?

Scenario C: An edge case in its domain. Something unusual, rare, or unprecedented happens in the domain. The agent has no specific protocol for this. How does its identity shape its approach?

For each scenario, write 2-3 sentences describing how the agent would respond, and explain which component of the identity drove that response. If the identity doesn't guide the response to any of the three scenarios, identify the gap and note what's missing.

Deliverable

A complete Agent Identity Document (1-2 pages):

Self-Review Questions (Required)

Same standard as Module 1. Do not move forward until you can answer each one honestly.

1. Does your identity tell the agent HOW to think, or just WHAT to do? Read each component. If you could replace the entire identity with a one-paragraph job description and get 90% of the same behavior, the identity isn't working hard enough. Rewrite until the job description can't replicate it.
2. Could another agent in your system have the same identity? If yes, the identity isn't differentiated. The primary obligation should be unique. The core belief should create a specific behavioral fingerprint. If two agents could share your identity document, at least one of them doesn't need to exist.
3. Read your identity anchor out loud. Does it have a voice? Can you hear how this agent thinks? Or does it sound like documentation? The identity anchor should feel like meeting a specific person, not reading a spec sheet.
4. Did the stress test reveal gaps? If all three scenarios were handled cleanly, either your identity is exceptionally well designed or your scenarios weren't challenging enough. Good stress tests should make you uncomfortable. At least one scenario should reveal a weakness worth noting.
5. Is the identity consistent? Do all six components point in the same direction? Read the metaphor, then the core belief, then the cognitive posture. Do they feel like they describe the same entity? If one component feels misaligned, revise it until the whole document reads as a single, coherent cognitive architecture.