Palestra

Level 1: The Loop

Understand what Claude Code actually is and how the agentic loop works.

Welcome to Level 1

Most people open Claude Code and start typing like they are in a chat window. That works, but it misses the point. Claude Code is not a chatbot. It is an autonomous agent — a system that gathers information, takes action, checks the result, and repeats.

This distinction changes everything about how you use it.

The goal of Level 1 is to shift your mental model from Claude Code as a chatbot to Claude Code as a loop you manage.

What you will learn

ModuleTopicCore Idea
1.1You're Not ChattingThe gather→act→verify loop that drives everything
1.2Getting StartedInstallation, authentication, your first session
1.3What You See on ScreenReading permission prompts, tool calls, and the status line
1.4The Context WindowClaude's only memory — and why it matters
1.5Your First Real CollaborationA guided hands-on exercise
1.6CheckpointProve your understanding without help

The mental model: The Loop

Every interaction with Claude Code follows the same pattern:

  1. Gather — Claude reads files, searches code, checks status
  2. Act — Claude edits a file, runs a command, writes something new
  3. Verify — Claude checks the result, runs tests, reads output
  4. Repeat — If the result is not right, Claude gathers more information and tries again

You are not having a conversation. You are managing a loop. Your job is to set the goal clearly, provide the right context, and intervene when the loop goes off track.

Key Takeaway

Claude Code is not a chatbot — it is an autonomous agent that runs a gather→act→verify loop. Your role is to manage this loop: set clear goals, provide context, and steer when needed.

Connect

As you work through Level 1, pay attention to the tool calls Claude makes. Can you identify which ones are 'gather' (reading, searching), which are 'act' (editing, writing), and which are 'verify' (running tests, checking output)?

1.1 You're Not Chatting