Claude Code Capabilities

• Subagents with a tightly scoped toolset. You define agents that work in their own context window, and build a doer-reviewer pair you can verifiably trust.
• Skills that actually get loaded. You write trigger descriptions that reliably fire, and build your first skill using progressive disclosure.
• Hooks instead of polite requests. Lifecycle hooks in settings.json block dangerous actions deterministically. You also learn how to spot a guardrail that fails silently.
• Dynamic workflows outside your context window. You have Claude write a workflow script that orchestrates many subagents, save it as your own command, and verify the results with git diff and your own gates.
• Goal mode with a real stop. You write completion conditions a separate evaluator model can actually verify from the transcript, and cap every autonomous run before it gets away from you.
• Automation without a chat window. Cloud routines that keep working on a schedule without your laptop, and headless calls with claude -p that you wire safely into scripts, CI, and the Agent SDK.

Structure: 7 modules, 21 lessons

• Module 1: Subagents and delegation3 lessons
• Module 2: Skills that travel3 lessons
• Module 3: Hooks as guardrails3 lessons
• Module 4: Dynamic workflows3 lessons
• Module 5: Goal mode3 lessons
• Module 6: Routines3 lessons
• Module 7: Headless and the Agent SDK3 lessons

Every module comes from real setups: the places where subagents, skills, or hooks did not behave as expected on the first try, and the fixes that came out of them. No syntax tables copied from the docs, only practice and verification.

Your access is unlocked immediately after payment: all lessons directly in the browser, in your account. Because this is digital content that starts right away, you confirm immediate access during checkout. Details are in the withdrawal information.