How to Use CLAUDE.md, Skills & Hooks in Claude Code | Full Course Tutorial

Claude Code keeps forgetting your context — every session starts from scratch. In Episode 2 of this Claude Code full course, you'll learn exactly how to fix that and build a Claude Code setup that actually remembers who you are, what your project is, and how you work. 🎯 What You'll Learn • How the CLAUDE.md file works as Claude Code's permanent persistent memory • Best practices for structuring CLAUDE.md so it doesn't bloat and slow Claude down • Progressive disclosure: how to reference external files instead of cramming everything into one file • The difference between CLAUDE.md, memory.md, and session context • What Skills (formerly slash commands) are and how to create your own custom ones • How to build auto-invocable vs. user-invocable skills with real examples • How Agents work in their own isolated context windows (without polluting your session) • How to configure Hooks via settings.json to trigger shell commands on pre/post tool use events • Plan Mode: what it is, when to use it, and how it protects your codebase • A tour of the Claude Plugins marketplace for discovering community-built skills 📋 Prerequisites • Watched Episode 1 of this Claude Code full course (content multiplier system build) • Claude Code installed (any plan) • Basic comfort with the terminal 🛠️ Tools & Concepts Covered • CLAUDE.md file (project-level and global) • Claude memory.md (user-level persistent memory) • Claude Code Skills (custom slash commands) • Claude Code Agents (isolated context, parallel execution) • settings.json Hooks (pre/post tool use automation) • Claude Plugins Marketplace • Plan Mode 💡 Key Insight from This Episode CLAUDE.md is deterministic — it loads the same way every time. But the instructions inside it are still interpreted by an LLM, which is probabilistic. That distinction changes how you write your rules. Must-follow constraints go at the top. Everything else uses progressive disclosure through referenced files. Treat CLAUDE.md like guidelines for a team, not a dump of everything you know. Agents are different: they run in their own context window and don't pollute your session — making them ideal for specialist tasks you can delegate and run in parallel. This is Episode 2 of the Claude Code Full Course by GenAI Unplugged. New episodes drop regularly covering real-world Claude Code setups, agentic workflows, and AI system building. 📺 CHAPTERS: 0:00 Claude Code Forgets Everything 1:00 Episode 1 Recap & Course Overview 3:05 CLAUDE.md File Deep Dive 11:19 Writing Effective CLAUDE.md Rules 15:32 Memory & Progressive Disclosure 20:51 Building Custom Skills 26:02 Claude Plugins Marketplace 28:07 Skills in Action (Live Demo) 31:12 Hooks & Settings.json Setup 36:24 Agents & What's Coming Next 📥 Helpful Links: 🛠️ Check my AI Systems store → https://store.genaiunplugged.com/ ▶️ Watch the Full n8n Zero to Hero Course Playlist → https://tinyurl.com/3rw3x6hy 🌐 Subscribe to GenAI Unplugged Substack → https://genaiunplugged.substack.com/ 🤝 Let's connect on LinkedIn → https://www.linkedin.com/in/dheerajsharma14/ 📚 Browse all my courses → https://genaiunplugged.com/ 🙌 Like this video? Here's what to do next: ✅ Hit thumbs up to support the channel ✅ Subscribe for weekly AI automation tutorials & lessons ✅ Drop a comment: What AI system would you like me to build next? #claudecode #claudecodetutorial #claude