Kaizen-UX← Home
An Examination

Claude Code Proficiency

A scenario-based placement test, from never used to expert who could teach it.

This is a closed-book, scenario-based test that places you on the spectrum from never having used Claude Code to being an expert who could train a team. The objective questions are graded for you, and your result maps to a proficiency band with guidance on how to progress.

The five parts run in order and grow harder; Part A covers foundations, Part E covers teaching and judgment. Within each part, the questions and their answer options are shuffled, so no two attempts look alike.

If a part feels too hard, skip the rest and jump to your score; everything you skip counts as missed. A blank Part E is normal for a strong daily driver, so don't let an early plateau discourage you.

Before you begin

  • Closed-book and solo: no Claude Code, docs, or search open. The point is to measure what you know.
  • Questions and options are randomised each attempt; the parts run A to E.
  • Stuck on a part? Use “Skip to my score”. The rest counts as missed, and you save time.

The five bands

  1. L0Never used0-10Has heard of Claude Code but cannot operate it.
  2. L1Novice11-30Can run it for assisted edits; shaky on modes, context, and recovery.
  3. L2Competent31-60Confident daily driver: permissions, context, git, model choice, recovery.
  4. L3Advanced61-85Configures the environment: plan mode, subagents, skills, MCP, memory, hooks.
  5. L4Expert / Teacher86-100Automates and extends Claude Code, and can onboard and calibrate a team.

For a shareable result · optional

Add your name, organization, and email to get a personal, shareable result link. Your name and organization show on the shared page; we keep your email private and never show it there. Your organization also lets your team compare results later.

46 questions · 5 parts · closed-book

Facts current as of 2026-06-03, verified against the official Claude Code documentation. Claude Code ships often, so a few version-pinned answers (permission-mode order, flag names, hook events) may drift over time.