Agent Engineering Course

This course turns the repository into a complete learning program for building useful, inspectable, and production-aware AI agents.

One-Sentence Summary

An agent is not a chatbot with a longer prompt. An agent is a task system that combines goals, context, tools, memory, workflow, evaluation, and human approval.

Who This Course Is For

• software engineers building LLM applications
• AI engineers moving from demos to production workflows
• founders and product teams designing agentic products
• researchers who want a practical engineering map
• students who want a structured agent learning path

Prerequisites

You should know:

• basic Python
• HTTP APIs and JSON
• command-line usage
• basic LLM concepts such as prompts, tokens, context windows, and embeddings

You do not need to know every agent framework before starting. Frameworks come after the mental model.

Course Tracks

Track	Best for	Path
Beginner Builder	New to agents	Modules 00-05, Labs 00-05, Examples 01-04
Agent Engineer	Building real workflows	Modules 00-09, Labs 00-09, Examples 01-09
Domain Builder	Healthcare, finance, enterprise	Modules 00-12, domain tracks, capstone
Production Lead	Responsible for deployment	Modules 08-25, evaluation harness, observability, security, runtime, MCP governance, incident response, operating model

Course Map

Part 1: Foundations
  00 Agent Foundations
  01 Agent Architecture
  02 Tool Calling

Part 2: Context And Knowledge
  03 Memory Systems
  04 RAG And Embeddings

Part 3: Control
  05 Workflow Orchestration
  06 Graph-based Agents
  07 Multi-Agent Systems
  08 Human-in-the-loop

Part 4: Production
  09 Production Agent Systems
  12 Agent Frameworks Comparison
  13 Agent Observability
  14 Agent Security
  15 Agent Cost And Latency Engineering
  16 Durable Agent Runtime
  17 MCP Modernization
  18 Agent Memory Governance
  19 Agent Identity And Permission
  21 Agent Incident Response
  22 Advanced Agent Evaluation
  24 Agent Product UX
  25 Enterprise Agent Operating Model

Part 5: Domain Systems
  10 Healthcare Agents
  11 Finance Agents
  Capstone Agent Colony

How To Study

For each module:

1. Read the curriculum chapter.
2. Read the matching roadmap level.
3. Run the closest example.
4. Complete the lab.
5. Answer the assessment questions.
6. Run python scripts/verify_examples.py.
7. Add one improvement to the example before moving on.

Graduation Criteria

You have completed the course when you can:

• define an agent's goal, scope, inputs, outputs, and tools
• build a single-purpose agent with structured output
• add tools with validation and approval gates
• design memory with write, retrieval, deletion, and audit rules
• build a RAG pipeline and evaluate retrieval quality
• orchestrate planner, executor, reviewer, and evaluator stages
• coordinate multiple agents without losing control
• run an evaluation suite before shipping changes
• inspect agent traces and explain failures from evidence
• defend against prompt injection in retrieved content and tool results
• route agent work by cost, latency, and quality constraints
• design checkpoint and resume behavior for long-running workflows
• separate MCP tools, resources, prompts, authorization, and elicitation
• govern memory redaction, retention, merge, decay, and deletion
• assign agent identities, scopes, owners, and access reviews
• run trace-driven incident response and postmortems
• build release gates with regression, safety, adversarial, and golden trace evals
• design agent UX for approval, evidence, recovery, and trust
• operate agents with registry, owners, risk tiers, reviews, and retirement
• explain production risks such as prompt injection, data leakage, runaway cost, and unsafe tool use
• complete the capstone project

Repository Sections

Section	Purpose
curriculum/	Concept chapters
assets/	Visual diagrams and teaching images
roadmap/	Level-by-level learning milestones
examples/	Runnable minimal implementations
security/	Prompt injection and agent security labs
showcases/	First-run demos with sample outputs
labs/	Guided hands-on exercises
teaching/	Teaching audit, misconceptions, deliverables, and module blueprint
lab-solutions/	Solution guides and expected artifact shapes
lesson-plans/	Instructor-ready 90-minute teaching plans
patterns/	Reusable architecture patterns
templates/	Project templates and safety/eval documents
assessments/	Quiz bank and review questions
projects/	Capstone and portfolio projects
capstone-starter/	Runnable starter scaffold for the final project
glossary/	Agent engineering terms

Teaching Principle

Every topic should answer four questions:

• What problem appears if we do not have this?
• What is the simplest mental model?
• What mechanism makes it work?
• How do we know it is safe and useful?

If a lesson cannot answer these four questions, it is not finished yet.