Module 1
Agent Loop, Harness & System Design
How agents reason and act, what a harness does, the fundamentals of AI system design and architecture, and the Claude Code ecosystem.
- 4 topics so far
- About 48 min
- Agent loopSystem designClaude Code
In this companion
What turns a model into an agent, what Claude Code wraps around it, and the four ways you shape it: CLAUDE.md, Skills, MCP and Subagents.
The path through this module
| Topic | What you do | Time | |
|---|---|---|---|
| 1.1 | What turns an LLM into an agent | Context, Tools, Loop. What turns a model that writes text into an agent that does things. | 15 min |
| 1.2 | The harness and Claude Code | What wraps the loop (tools, context, permissions, memory), then Claude Code as your build partner. | 15 min |
| 1.3 | CLAUDE.md and Skills | Memory you write down, and one word for a whole workflow. | 10 min |
| 1.4 | MCP and Subagents | New tools from outside, and specialists that keep your context clean. | 8 min |
Start here: 1.1 What turns an LLM into an agent