The city the agent lives in
System-level patterns address the holistic, external architecture around individual agents. Each answers a core production question:
Key concept · The questions that define production readiness
- How do services and agents discover each other dynamically? → Tool and Agent Registry
- How do we enforce security, identity, and access control for autonomous agents? → Agent Authentication and Authorization
- How can we monitor and enforce compliance in real time? → Real-Time Compliance Monitoring
- How does the system react to external events asynchronously and at scale? → Event-Driven Reactivity
Discovery, security, governance, scale
| Pattern | Solves | Why it matters |
|---|---|---|
| Tool and Agent Registry | Dynamic discovery | Lets the ecosystem grow and evolve without constant code changes — modular, resilient. |
| Agent Authentication & Authorization | Identity & access | The non-negotiable first step — without it, trust is impossible. |
| Real-Time Compliance Monitoring | Automated governance | Embeds rule/regulation enforcement into the system's fabric — no black boxes in regulated industries. |
| Event-Driven Reactivity | Async responsiveness | Keeps the system responsive and scalable, reacting to outside events in near-real time. |
Key concept · Security is the foundation
An agent's identity and permissions are not optional. Agent Authentication and Authorization is the non-negotiable first step for any system where agents perform meaningful actions on behalf of a user or organization — without it, there is no basis for trust. Security is treated as the foundation the other patterns build on.
How these patterns earn production readiness
Decoupling enables scale and evolution. Production systems are dynamic: Event-Driven Reactivity makes the system responsive and scalable by reacting to events (a new email, a status change) asynchronously rather than polling, while the Tool and Agent Registry lets the ecosystem add and swap agents/tools without rewiring code — a modular, resilient architecture.
Key concept · Governance must be automated
In high-stakes, regulated environments, "hope is not a strategy." Real-Time Compliance Monitoring embeds governance directly into the system's fabric so that autonomous actions adhere to rules and regulations automatically and continuously — not through after-the-fact audits. Agents in finance, healthcare, and law cannot be black boxes.
Collectively, these patterns represent a shift in mindset. They are the deliberate architectural decisions that address the non-functional requirements — security, scalability, governance, and reliability — that distinguish a robust application from a fragile experiment, and provide the vocabulary for the foundational infrastructure of any serious agentic initiative.
Chapter 10 quiz
Fifteen questions on the four production questions, the system-level patterns, and what makes a system production-ready. Answer first, then expand Show answer.
Part A · Multiple choice
Which pattern answers "how do services and agents discover each other dynamically?"
- A Agent Authentication & Authorization
- B Tool and Agent Registry
- C Real-Time Compliance Monitoring
- D Event-Driven Reactivity
Show answer
B — Tool and Agent Registry. It enables dynamic discovery so the ecosystem can evolve without code changes.
Which pattern is described as the non-negotiable first step for any serious agentic system?
- A Event-Driven Reactivity
- B Tool and Agent Registry
- C Agent Authentication & Authorization
- D Real-Time Compliance Monitoring
Show answer
C. Without identity and permissions, trust is impossible — security is the foundation.
Which pattern keeps a system responsive and scalable by reacting to outside events in near-real time?
- A Event-Driven Reactivity
- B Real-Time Compliance Monitoring
- C Tool and Agent Registry
- D Single Agent Baseline
Show answer
A — Event-Driven Reactivity. Asynchronous reaction rather than polling.
In regulated industries, agentic systems cannot be "black boxes." Which pattern addresses this?
- A Tool and Agent Registry
- B Real-Time Compliance Monitoring
- C Event-Driven Reactivity
- D Multimodal Sensory Input
Show answer
B. It embeds automated, continuous governance into the system's fabric.
What is the overall focus of system-level patterns (versus agent-level patterns)?
- A The internal reasoning of a single agent
- B The external architecture — the "city" — that lets agents communicate, act securely, and scale
- C Prompt engineering
- D Fine-tuning datasets
Show answer
B. They move beyond the agent's mind to the surrounding production infrastructure.
How does decoupling (via the Registry and Event-Driven patterns) help a production system?
- A It removes the need for security
- B It enables scale and evolution — the ecosystem grows without constant code changes
- C It slows the system down
- D It eliminates governance
Show answer
B. Decoupling makes the architecture modular, resilient, and adaptable.
Part B · True or false
An agent's identity and permissions are optional nice-to-haves.
Show answer
False. They are non-negotiable — without authentication and authorization, meaningful action can't be trusted.
Compliance can be handled with after-the-fact audits rather than real-time monitoring.
Show answer
False. In high-stakes environments governance must be automated and continuous — embedded in the system's fabric.
Production readiness is fundamentally an architectural choice.
Show answer
True. These patterns are the deliberate decisions addressing non-functional requirements.
The Tool and Agent Registry requires developers to hardcode every new agent into the core system.
Show answer
False. The registry lets the ecosystem grow and evolve without constant code changes.
System-level patterns target non-functional requirements like security, scalability, governance, and reliability.
Show answer
True. These distinguish a robust application from a fragile experiment.
Part C · Short answer
Name the four system-level patterns.
Show answer
Tool and Agent Registry, Agent Authentication & Authorization, Real-Time Compliance Monitoring, and Event-Driven Reactivity.
Match each pattern to its production question (discovery, security, compliance, async events).
Show answer
Discovery → Tool and Agent Registry; security/identity → Agent Auth; compliance → Real-Time Compliance Monitoring; async events → Event-Driven Reactivity.
Which pattern is "the foundation," and why?
Show answer
Agent Authentication & Authorization — identity and permissions are the basis of trust; without them meaningful autonomous action is unsafe.
Name the four non-functional requirements these patterns collectively address.
Show answer
Security, scalability, governance, and reliability.