AniUI Academy
Beginner to advanced32 lessons297 min total

Claude Code Masterclass

Start with no experience of AI coding tools and finish able to run Claude Code well on real work. What an agent actually is, how to prompt and verify it, how to configure it with CLAUDE.md, skills, hooks and MCP, and how to scale it across parallel sessions and CI without losing control of it.

Start with What Is an AI Coding Agent

Part 1 · Foundations

7 lessons · 62 min

What an agent is, how it works, and how to stay in control of one. Assumes nothing.

  1. What Is an AI Coding Agent9 minThe difference between autocomplete, a chatbot and an agent — what Claude Code can actually do to your machine, and how that changes what your job becomes.
  2. Installing Claude Code7 minThe install methods and how they differ on updates, what happens the first time you run it, and what to do when the terminal tells you it has never heard of the command.
  3. The Agentic Loop8 minGather context, take action, verify results — the three phases Claude works through, why they blend rather than run in order, and where you sit inside the loop.
  4. The Tools Claude Has9 minThe five categories of built-in tool, what each can actually do, why the tool list is the whole difference between an agent and a chat window, and how it gets extended.
  5. Permissions and Staying in Control10 minThe four permission modes and what each lets through, how auto mode's classifier decides, allowlisting the commands you trust, and the honest limit of approving prompts one by one.
  6. Checkpoints and Undo8 minHow file snapshots let you rewind a session, the difference between restoring code, conversation or both, and the two gaps that make git still necessary.
  7. Your First Real Task11 minA complete worked task, from an empty terminal through to a commit — choosing what to attempt, writing the prompt, watching the loop, and deciding when it is done.

Part 2 · The daily craft

10 lessons · 91 min

The habits that separate a session you babysit from one you can walk away from.

  1. The Context Window10 minWhy model performance degrades as context fills, what occupies it, how auto-compaction works, and why almost every good habit with Claude Code derives from this one constraint.
  2. Writing a Prompt That Works9 minFour moves that turn a prompt Claude has to guess at into one it can act on, how to hand over files, images and URLs, and when a vague prompt is the right tool.
  3. Giving Claude Something to Verify10 minClaude stops when the work looks done. Four escalating ways to give it a real check instead — in the prompt, as a session goal, as a Stop hook, and as a second opinion.
  4. Explore, Plan, Then Code9 minThe four-phase workflow — explore, plan, implement, commit — how to enter and leave plan mode, editing the plan before it runs, and when planning is just overhead.
  5. Course-Correcting9 minEsc, mid-run corrections, the rewind menu and /clear — plus the judgement call that matters most, which is when to stop correcting and start a clean session instead.
  6. Questions About an Unfamiliar Codebase8 minThe use of Claude Code that produces no diff at all — asking a strange codebase the questions you would otherwise save up for a senior colleague.
  7. Letting Claude Interview You9 minFor a feature too big to describe in a paragraph, invert the direction — have Claude question you until the decisions are made, then write the spec down.
  8. Managing Sessions9 minEvery conversation is saved to disk and tied to a directory. How to resume one, when to fork instead, and why naming sessions turns them into branches.
  9. Git, Commits and Pull Requests9 minClaude can stage, commit, branch and open pull requests. Why a CLI tool is the cheapest way to reach an outside service, and where your responsibility starts.
  10. Debugging With Claude9 minPaste the real error, say where it probably lives, ask for a failing test before the fix, and refuse a suppressed symptom as a solution.

Part 3 · Making it yours

9 lessons · 88 min

CLAUDE.md, skills, hooks, MCP and subagents — teaching Claude your project once.

  1. CLAUDE.md9 minThe file Claude reads at the start of every session — where to put it, what belongs in it, and why writing too much is the mistake almost everyone makes.
  2. Auto Memory and the .claude Directory9 minThe notes Claude keeps for itself between sessions, how much of them actually load, and a guided tour of every file in the .claude directory.
  3. Skills9 minPackaging a procedure or a body of knowledge into a file Claude loads only when it is needed, so it costs you nothing the rest of the time.
  4. Hooks9 minShell commands that run automatically at fixed points in Claude's workflow, for the things that must happen every time rather than usually.
  5. Connecting Tools With MCP9 minAn open standard for giving Claude access to systems outside your codebase — issue trackers, databases, design files — and what each connection costs you.
  6. Subagents10 minHanding work to a separate context window, so an investigation that reads twenty files comes back as one paragraph instead of twenty files of noise.
  7. Plugins10 minBundling skills, hooks, subagents and MCP servers into one installable unit, so a second repository or a new teammate gets your whole setup at once.
  8. Choosing the Right Extension12 minA decision procedure for CLAUDE.md, skills, hooks, MCP, subagents and plugins, so you stop guessing which one a given problem belongs in.
  9. Debugging Your Configuration11 minA diagnostic order for configuration that does not work — what actually loaded, which copy of it won, whether it ever fired, and how to bisect the rest.

Part 4 · Scale

6 lessons · 56 min

Beyond one session: automation, parallel agents, CI, cost and security.

  1. Non-Interactive Mode9 minRunning Claude Code with nobody at the keyboard — the print flag, output formats a script can parse, fanning out across hundreds of files, and scoping permissions for an unattended run.
  2. Parallel Sessions and Worktrees10 minWhy two Claude sessions in one directory tread on each other, how git worktrees fix it, and the ladder of options from a second terminal up to a coordinated team of agents.
  3. Claude Code in CI10 minRunning Claude inside GitHub Actions — responding to mentions, turning an issue into a pull request, automated review, and being sensible about secrets and permissions.
  4. Managing Cost9 minWhat actually drives token spend, how to check where yours is going, and the levers that reduce it — which turn out to be the same habits that make the output better.
  5. Security and Sandboxing10 minThe threat model in plain language, including instructions hidden in content Claude reads, plus the sandboxed Bash tool and the ladder of isolation up to a virtual machine.
  6. Developing Your Intuition8 minThe rules in this course are defaults rather than laws — when to break them, how to notice what actually worked and why, and where to go now that the course is finished.

Prove it: Claude Code Certification

This course teaches the whole exam syllabus, free. The exam itself is 2 hours under a clock, 60 questions, ₹199$9.99 — and a certificate with a link anyone can check if you pass.

See the exam