Learn Claude Code is an educational repository about harness engineering for AI coding agents. Its central idea is that the model provides the underlying reasoning capability while the surrounding harness determines how that model interacts with tools, context, files, memory, and the development environment, so the course focuses on building the infrastructure that turns a model into a practical agent product.

The repository presents a progressive lesson track that explores agent loops, tools, context handling, coding workflows, and increasingly capable harness patterns using Claude Code as a reference point. It is useful for developers who want to understand how modern coding agents work internally and how to build their own agent CLI, SDK, or persistent assistant rather than only learning how to prompt an existing product.