Master Claude Code, level by level
Master Claude Code rung by rung: predict how it behaves, prove it in your own terminal, ship real work.
About this path
Claude Code is an agent that works in your terminal: it reads your codebase, edits files, runs commands, and verifies its own work. This course teaches it the way it deserves to be taught, hands-on. Each level pairs quick predict-then-verify challenges (commit your prediction, then prove it in your own terminal) with a boss project you complete in your real environment and submit for rubric grading. By Level 8 you are orchestrating subagents, wiring MCP servers, and building your own agent on the Agent SDK. The same levels double as hands-on preparation for the Claude Code, MCP, and agentic-architecture domains of Anthropic's CCA-F certification.
Skills you'll put on a resume
- Drive the full Claude Code workflow: explore a codebase, plan a change in plan mode, implement it, and verify it before committing
- Author CLAUDE.md hierarchies and path-scoped rules that steer Claude across a real multi-package repository
- Configure least-privilege permissions, custom slash commands, skills, and lifecycle hooks in settings.json and .claude/
- Connect Claude to external tools through MCP servers, choosing the right scope, transport, and auth model
- Orchestrate subagents with isolated contexts, and run Claude headless in scripts and CI pipelines
- Build a small working agent with the Claude Agent SDK, with scoped tools and structured output
For
Anyone who works with Claude
Prerequisites
- Comfortable in a terminal (cd, run commands, read output)
- A Claude account; a Pro/Max subscription or API key to run Claude Code locally
- Basic git for the later levels (branch, diff, commit)
Ready to start?
Pro unlocks every graded project in this course, all labs across Preporato, and every practice test. $29.99/mo, cancel anytime.