Claude Code Level 1: Coding with an Agentic AI Partner 

This introductory Claude Code course is designed for developers, not for people new to programming or the command line, so it assumes you already have experience in at least one programming language and are comfortable working in a terminal.

Throughout the course, you work directly with Claude Code, Anthropic's command-line AI tool, to take on real coding tasks through natural conversation. You practice writing functions from scratch, tracking down stubborn bugs, refactoring tangled modules, and onboarding to unfamiliar codebases with AI as a collaborative partner. The emphasis throughout is on safe habits and daily workflow fluency, the discipline that separates engineers who use Claude Code effectively from those who fight it. By practicing these habits on real repositories, you build a workflow you can carry directly into your own work.

Learners will have the opportunity to:
1. Practice safe agentic habits while working on real code repositories with Claude Code
2. Apply the audit-first habit by reviewing every proposed change before accepting it
3. Work with Plan Mode and the core slash commands to scope work before any code is written
4. Map an unfamiliar codebase using Claude's file-system tools
5. Practice recovering from drift using the three escalating recovery moves
6. Apply the agentic debugging loop to broken code across three levels of difficulty
7. Experiment with spawning ephemeral subagents for parallel exploration
8. Explore running Claude Code in headless mode for simple scripted automation
9. Practice polishing code through refactoring, documentation generation, and clean commit-message authoring
10. Work toward shipping at least one end-to-end feature using the full Level 1 toolkit
 

Topics
- Claude Code foundations: the three modes, the agentic CLI, and the risks of unsupervised "vibe-coding"
- Safe habits: the audit-first workflow, the key slash commands, and Plan Mode
- Onboarding to unfamiliar code with Read, Edit, Grep, and Glob, plus context discipline
- Plan Mode in depth: the five-tag plan, iteration moves, and recovering from drift
- The agentic loop and debugging: the verification ladder, submitting errors well, and debugging anti-patterns
- Subagents, parallel workers, and headless mode
- Refactoring through readability, performance, and maintainability lenses
- Hands-on labs and an end-to-end capstone feature build

Prerequisite: Before enrolling, you should be able to answer "yes" to the following:

- I can navigate files and directories from a command line.
- I can run terminal commands and understand their output.
- I can write and execute basic code in at least one programming language.
- I understand fundamental programming and troubleshooting concepts.
- I can install software, development tools, and packages with minimal assistance.

Prework
Complete all of the following before Session 1. Class time is not used for installation.
- Install Node.js 18 or newer, which includes npm: https://nodejs.org
- Install git: https://git-scm.com
- Install VS Code: https://code.visualstudio.com
- Install Claude Code
- Sign in once by running Claude in any terminal and following the prompts.

Includes: Access to Anthropic Claude (Pro Tier) for the duration of the training.