Claude Code, properly

Most people use Claude Code like autocomplete. It is a loop.

Twelve levels that take you from your first session to running an agent on a codebase you did not write — graded deterministically, so a score means the same thing every time it is computed.

Start with What agentic coding actually changes (and what it doesn't)All lessons

Verified content: Every lesson names the Claude Code version it was checked against. The build refuses to publish one that does not.
Deterministic grading: The rules that score your CLAUDE.md are pure functions. Same submission, same score, with the fix named.
A simulator, not a chatbot: Practice sessions replay identically, so "you blew the context at turn 14" is a fact rather than an impression.