Agents as Tools

What you will Learn

In this module you will build the foundation for everything that follows: a working mental model of how large language models think, and the practical ability to direct one as a coding agent. You will start with the internals — how LLMs generate responses, why prompts succeed or fail, and what distinguishes a coding agent from the model it contains. Then you will install Claude Code and develop cost-aware habits for model selection, weighing Fable, Opus, Sonnet, Haiku, and a free local model against the task at hand. From there you will apply the Prompt Life Cycle to a real project, using CLAUDE.md, subagents, planning mode, and the Playwright MCP server to keep Claude focused and your context window small. Finally, you will learn to save your progress: resuming and rewinding Claude Code sessions, visualizing your project in Obsidian, and using Git, GitHub, and reusable Skills to save and restore the state of your code. By the end of this module you will be ready to use Claude Code deliberately — controlling what it knows, what it can access, and how it behaves within any project you give it.

Lessons

• Lesson 1 - Working with LLMs: Build a working mental model of how large language models function, then develop the prompt engineering skills to communicate with them reliably.
• Lesson 2 - Coding Assistants: Understand the architecture of a coding agent — the orchestration loop, tools, and MCP — and install Claude Code as a cost-aware coding assistant, choosing the right model and effort level for each task.
• Lesson 3 - Claude in Action: Apply the Prompt Life Cycle, CLAUDE.md, subagents, planning mode, and the Playwright MCP server to build and review a real project while keeping your context window under control.
• Lesson 4 - Saving Your Progress: Save the state of context within your Claude Code sessions, visualize your project in Obsidian, and save the state of your project with Git and GitHub using reusable Skills.

Technologies