Multi-Agent AI Systems in C#

Building Autonomous Agents with Harness, Hermes, and Loop Engineering

Introduction: The Agent Revolution

1. Why Multi-Agent?
2. The Three Pillars
3. Who This Book Is For
4. What You Will Build
5. How This Book Is Organized
6. A Note on Code Examples
7. The Road Ahead

Chapter 1: The Age of Multi-Agent Systems

1. What Are AI Agents and Why Multi-Agent?
2. From Chatbots to Autonomous Systems: The Evolution
3. Introducing Harness, Hermes, and Loop Engineering
4. The Three-Layer Architecture: Runtime, Orchestration, Autonomy
5. What This Book Will Build
6. Chapter Summary

Chapter 2: Development Environment Setup

1. Prerequisites and Toolchain Installation
2. Creating Your First .NET Agent Project
3. Installing Microsoft Agent Framework Packages
4. Configuring Model Providers
5. Your First Running Agent in Five Lines of Code
6. Understanding the Project Structure
7. Configuring Dependency Injection
8. Verifying Your Setup
9. Troubleshooting
10. Chapter Summary

Chapter 3: Theoretical Foundations of Multi-Agent Systems

1. Agent Architecture Patterns: Reflex, Deliberative, Hybrid
2. Communication Models: Message Passing, Shared Memory, Blackboard
3. Coordination Theory and Emergent Behavior
4. The BDI Model: Belief, Desire, Intention in AI Agents
5. Why Multi-Agent Systems Outperform Single Agents
6. Chapter Summary

Chapter 4: Agent Harness Fundamentals

1. What Is an Agent Harness and Why It Matters
2. The HarnessAgent Class and Default Capabilities
3. Creating and Configuring a Harness Agent in C#
4. Enabling Context Compaction for Long Conversations
5. Customizing and Disabling Features
6. The Plan/Execute Workflow
7. Chapter Summary

Chapter 5: Tools, MCP, and Function Calling

1. Function Calling and Tool Invocation in MAF
2. Building Custom Tools with C# Methods
3. Model Context Protocol: Connecting to MCP Servers
4. Building Your Own MCP Server in C#
5. Exposing Agents as MCP Tools
6. Chapter Summary

Chapter 6: Memory, Context, and State Management

1. AgentSession and Multi-Turn Conversations
2. Context Providers and Custom Memory
3. Compaction Strategies for Long-Running Agents
4. Retrieval-Augmented Generation with Vector Stores
5. Structured Outputs and JSON Schema
6. Chapter Summary

Chapter 7: Hermes Agent Architecture

1. What Is Hermes Agent and Its Design Philosophy
2. Hermes Architecture: The AIAgent Orchestration Engine
3. Multi-Agent Orchestration with the Kanban Board
4. Scheduling, Cron, and Autonomous Task Execution
5. Integrating Hermes from C# via API Server
6. Chapter Summary

Chapter 8: Building Multi-Agent Workflows with Hermes

1. Defining Agent Roles and Specialization
2. Orchestrator Pattern: Decomposing Complex Tasks
3. Parallel Execution and Resource-Aware Scheduling
4. Shared Context and Selective Information Sharing
5. Building a Research Pipeline with Multiple Hermes Agents
6. Chapter Summary

Chapter 9: The Loop Engineering Paradigm

1. From Prompting to Loop Design: A Paradigm Shift
2. The Five Primitives of Loop Engineering
3. Automations: The Heartbeat of Autonomous Agents
4. Skills and Plugins: Codifying Project Knowledge
5. State Management: Memory That Survives the Loop
6. Chapter Summary

Chapter 10: Building Self-Prompting Agent Loops in C#

1. Implementing Agent Loops with LoopEvaluators
2. Goal-Driven Execution: Run Until Done
3. Scheduled Automations with Background Services
4. Self-Healing and Error-Recovery Patterns
5. Building a Complete Autonomous Agent Loop
6. Chapter Summary

Chapter 11: Multi-Agent Orchestration Patterns

1. Sequential Orchestration: Step-by-Step Pipelines
2. Concurrent Execution: Fan-Out and Fan-In Patterns
3. Handoff Orchestration: Contextual Agent Delegation
4. Group Chat: Multi-Agent Dialogue Spaces
5. Magentic Orchestration: Dynamic Sub-Agent Management
6. Chapter Summary

Chapter 12: Durable Workflows and Human-in-the-Loop

1. Making Workflows Durable with Durable Task Scheduler
2. Checkpointing and Recovery from Failures
3. Human-in-the-Loop: Approval Gates and Request Ports
4. Hosting on Azure Functions for Serverless Scaling
5. Conditional Routing and Sub-Workflows
6. Shared State in Workflows
7. Chapter Summary

Chapter 13: Agent Communication Protocols

1. Agent-to-Agent (A2A) Protocol: Cross-Framework Communication
2. AG-UI: Real-Time Streaming to Frontends
3. Building a Multi-Agent Dashboard with AG-UI
4. Interoperability Between MAF, Hermes, and Other Frameworks
5. Chapter Summary

Chapter 14: Security, Guardrails, and Governance

1. The OWASP Agentic Top 10 Threat Landscape
2. Middleware-Based Security Guards
3. Content Filtering and Harmful Output Prevention
4. Authentication, Authorization, and Agent Identity
5. Policy Enforcement with the Agent Governance Toolkit
6. Chapter Summary

Chapter 15: Testing, Debugging, and Observability

1. Unit Testing Agents and Executors
2. Integration Testing Workflows End-to-End
3. DevUI: Interactive Agent Testing Without Overhead
4. OpenTelemetry Tracing for Multi-Agent Systems
5. Metrics, Logging, and Performance Monitoring
6. Chapter Summary

Chapter 16: Deployment, Scaling, and Production Best Practices

1. Choosing Your Deployment Target
2. Deploying to Azure Container Apps with Aspire
3. Auto-Scaling and Performance Optimization
4. Cost Management: Token Budgets and Rate Limits
5. Production Runbooks and Incident Response
6. Chapter Summary

Chapter 17: Building a Complete Multi-Agent Application

1. Architecture Design: The Customer Support Platform
2. Building Specialized Agents
3. Implementing the Orchestration Workflow
4. Adding RAG, MCP Tools, and Human Approval Gates
5. Deploying and Monitoring the Complete System
6. Chapter Summary

Conclusion: The Road Ahead

1. Key Takeaways: The Three-Layer Stack
2. Where Multi-Agent Systems Are Headed
3. Your Roadmap for Continued Learning
4. Final Thoughts on Building Responsible AI Agents

References