Create your agent
Design the identity and contract of your AI agent. Result in 10 minutes.
Welcome to Architecture Thinking. In this module you will learn to build YOUR OWN AI agent: first its identity (who it is, its personality, its values) and then its contract (the instructions it must always follow). Each class teaches you step by step, modeling how it is done, and you finish by writing your agent contract that the app evaluates with a rubric (a list of criteria that checks whether your answer meets what was asked). **The framework that ties it all together:** an AI agent is not just the model that thinks: it is **Model + Harness**. The model is the brain; the harness is the body and the safeguards: the identity (SOUL.md), the contract, the rules, the limits and the environment that controls it. This module teaches you to direct that body, just as you would with your real agent (Hermes, OpenCode or Claude Code) to build your own learning system.
Your agent identity
Define who your agent is: personality, tone, values and absolute limits.
Instructions contract
Write clear instructions: Role, Context, Objective, Constraints, Format and Specificity.
Rules
General rules (always) vs specific (only in context) and loop rules.
Limits and Never do this
What your agent must not do, safety limits and why they matter as much as the instructions.
