Exploring Advanced Features of Claude Agent SDK in TypeScript

Overview

Master advanced functionalities of the Claude Agent SDK to build robust, secure, and context-aware AI agents. This course covers implementing hooks for agent lifecycle management, utilizing source attribution and CLAUDE.md files for enhanced contextual understanding, and integrating Agent Skills to equip Claude with specialized capabilities.

Syllabus

• Unit 1: Intercepting Agent Execution with Hooks
Building a Startup Notification Hook
Blocking Unsafe Prompts with a User Prompt Hook
Logging Tool Usage with a PreToolUse Hook
Building a Bash Safety Guardrail with PreToolUse Hooks
• Unit 2: Project Configuration with Source Files
Building a Helper Agent with CLAUDE.md
Understanding CLAUDE.md Precedence in Nested Project Folders
Loading Personal Overrides from Local Settings
Overriding Project Instructions with Runtime System Prompt
• Unit 3: Extending Claude with Skills
Trigger the slack-gif-creator Skill with a Simple Prompt
Rename a Skill and Update the Prompt
Automatic Skill Discovery from Natural Language Prompts
Create a Text Case Converter Skill

---

Subjects

Artificial Intelligence