A guided path from "what's a terminal?" to shipping AI agents, plus quizzes and timed mock exams for the Claude Certified Architect and Claude Developer certifications.
The two worries we hear most are that AI code is insecure and messy. Set up well, Claude Code is neither. Here is how.
Myth
Busted
Claude Code is as secure as the setup around it. Deny rules keep it away from your secrets, the sandbox fences in its commands, /security-review checks every change, and nothing ships without a person reviewing it.
Myth
Busted
It follows the standards you give it. A few lines in CLAUDE.md, small tasks and a quick look at each change keep its work lean and tidy, and it can clean up its own code when you ask.
See how to keep it lean →Myth
Busted
You decide what it can do. Plan mode shows you the plan before anything changes, it asks before risky actions, and checkpoints and git let you undo any step.
Stay in control →Each level builds on the last. Short lessons, everyday analogies, and a quick quiz at the end of every one.
What Claude Code is, how to install it and how to ask for what you want.
🚲Level 2Editing real files safely, steering with modes and undoing mistakes.
🚗Level 3Memory, context and custom commands that make Claude feel like a teammate.
🏎️Level 4Skills, subagents, hooks and MCP: extend Claude Code to fit your workflow.
🚀Level 5Automation, the Agent SDK and architecting production-grade agentic systems.
Loading today's question…
Scenario-based questions with one right answer and three tempting traps, like the real thing. Every answer comes with an explanation.
Use Claude well at work: prompting, checking outputs, choosing tools, workflows, governance and troubleshooting.
Build on Claude: the API, agents, model choice, prompts and context, tools and MCP, security.
Design production systems with Claude: agents, tools, MCP, Claude Code, prompts and reliability.
Lead Claude solutions end to end: design, models and context, integration, evaluation, governance and stakeholders.
Hands-on Claude Code: the CLI, configuration, skills, subagents, hooks, MCP, CI and the API basics.
Instant feedback and a clear explanation after every answer. Drill a single domain until it clicks.
Questions weighted by domain, a countdown, flag-for-review, and a score out of 1000 with a domain breakdown.
Flip through scenarios on your phone on the train. The fastest way to lock in the key decision rules.
Every meme hides a real lesson. If you get the joke, you're learning. If you don't yet, the link underneath explains it.
Writing "please ALWAYS run the tests" in CLAUDE.md and praying
A PostToolUse hook that runs them every single time
Copy-pasting code from a chatbot
Asking Claude Code to fix it
Plan mode first, then fix it
Subagents reviewing each other while you make a cup of tea
Running --dangerously-skip-permissions on the production server
This is fine.
Claude Code is Anthropic's agentic coding tool. You describe what you want in everyday language and it reads your project, edits files, runs commands and tests, and works through multi-step tasks with you. It runs in the terminal, IDEs, a desktop app and the browser.
It can be, and it is as secure as the setup around it. Deny rules keep Claude away from secrets such as .env files, the sandbox limits what its commands can reach, /security-review checks each change for common vulnerabilities, and a person reviews anything that touches logins, payments or personal data.
Not when you set standards. Claude follows the rules in your CLAUDE.md, so a few lines asking for small changes, no comments that repeat the code and no new files unless needed keep its work lean. Small tasks and a quick review of each change do the rest, and it can tidy its own work when you ask.
No. Level 1 assumes zero coding experience and explains every term as it comes up. Many people start with Claude Code to build internal tools and automations, then grow into more technical work.
The 25 lessons take about 3.4 hours to read. Most people work through a level a week alongside real projects, so expect four to six weeks to feel confident.
Yes, it is designed to. The Architect practice bank covers all five exam domains with scenario-based questions, a timed mock exam weighted by domain, and explanations for every answer. Use it alongside Anthropic's official materials.
Everything is free and there is no sign-up. Open any lesson, quiz or mock exam and start straight away.
The whole course on one tape. Side A gets you going, Side B makes you dangerous. No pencil needed to rewind.
Learn Claude Code is made by Tenzing Digital. We build custom AI agents and software for Irish businesses. If you'd rather have experts alongside you than learn it all alone, we'd love to help.
TenzingDigitaltenzing.ie →
No account, no credit card, no jargon. Dive in.