Claude Code for Experienced Developers
Agentic AI development: learn the mental models, workflows, and safety discipline that make Claude Code effective as an autonomous coding agent.
INFO
- Time: ~6-8 hours (self-paced, modular)
- Difficulty: Intermediate to Advanced
- Audience: Developers comfortable with terminals, git, and at least one programming language
- Updated: 2026
- Prerequisites: Install Claude Code before starting. This course teaches workflows and methodology, not feature reference — see the official docs for the complete feature reference.
What This Course Covers
This course teaches you to use Claude Code effectively as an agent — an autonomous system that reads your codebase, plans changes, writes code, runs tests, and iterates. Most experienced developers start by treating it like a chatbot. This course fixes that.
You'll learn the mental models, decision frameworks, and workflow patterns that aren't in the official documentation. The official docs tell you what Claude Code can do. This course teaches you how to think about it and when to use which approach.
Course Structure
The course is split into focused modules. Work through them in order for the full experience, or jump to specific topics as needed.
Core Modules
| Module | Topic | What You'll Learn |
|---|---|---|
| Module 0 | The Mental Model Shift | Agent vs. chatbot thinking, the agentic loop, when Claude Code shines vs. when other tools are better |
| Module 1 | Setup & First Session | Permission philosophy and building intuition for when to trust the agent |
| Module 2 | CLAUDE.md & Project Memory | Why each type of content belongs, the "onboarding docs" framing, keeping it concise |
| Module 3 | Effective Prompting | Verification-first prompting, @ file references, constraints, prompt patterns for explore/implement/test/refactor/debug |
| Module 4 | Context Management | The #1 operational skill — /clear, /compact, --continue/--resume, /btw, context decay, session strategies |
| Module 5 | Plan Mode & the PDCA Cycle | Four-phase workflow, Plan-Do-Check-Act mapping, when to plan vs. dive in |
| Module 6 | Workflows, Git Safety & Testing | Verify loop, git checkpoints, /rewind, diff review, pitfalls, production checklist |
| Module 7 | The Extensibility Stack | The 6-layer decision framework: CLAUDE.md → rules → hooks → skills → MCP → plugins |
| Module 8 | Agents & Parallel Development | Subagents, custom agents, /batch, git worktrees, Agent Teams, custom agent workflows |
| Module 9 | Automation & Multi-Surface | Headless mode, CI use cases, /loop, remote control, multi-surface workflows |
| Module 10 | Capstone Project | End-to-end feature implementation — plan, implement, test, simplify, PR |
Appendix
| Appendix | Topic |
|---|---|
| Appendix | CLI reference, troubleshooting, enterprise & compliance |
Who This Is For
This course is for experienced developers who:
- Are comfortable with terminals, git, and package managers
- Know at least one programming language well (JS/TS, Python, .NET, or similar)
- Want to use AI as an autonomous coding agent, beyond simple Q&A
- Have existing projects they want to accelerate
This is not an introduction to programming, terminals, or version control. If you need those foundations, start elsewhere.
Key Themes
Throughout the course, you'll encounter these recurring principles:
- Agent, not chatbot: give Claude Code tasks with clear outcomes, not questions with expected answers
- Verification-first: include test criteria in every prompt; let the verify loop catch issues
- Context is finite: manage your context window like you manage memory;
/clearis your best friend - Plan before you build: Plan Mode + PDCA gives you control over what gets implemented
- Review everything: if you wouldn't approve it in a PR from a colleague, don't accept it from Claude
- Layer your extensibility: use the right tool for the job: CLAUDE.md for context, hooks for enforcement, MCP for integration
Getting Started
Start with Module 0: The Mental Model Shift, even if you've used Claude Code before. The framing will change how you work with it.
If you already have Claude Code installed and just want the workflows, skip to Module 3: Effective Prompting.