The Best Claude Code Courses in 2026 (Free and Paid)

We only list courses whose official page we could read today, and we took price, certificate, and length from those pages, not from review sites. Then we ranked them by three questions:

• Is it current? Claude Code changes fast. Courses that teach permission modes, hooks, skills, and headless runs beat ones built around last year's features.
• Does it make you build something? Watching someone else prompt doesn't teach you much. Courses with real projects ranked higher.
• Is the price justified? Anthropic gives its own courses away. A paid course has to offer something extra, usually a certificate or more structure.

We left out Udemy courses. Udemy's course pages blocked our checks today, so we couldn't confirm their contents or prices.

If you search for an "Anthropic Claude Code course", you'll find the same courses in several places. Here's how it fits together:

• Claude Academy (academy.claude.com) is Anthropic's learning hub, and the Claude Code docs link to it directly. It describes itself as "free courses on Claude and AI, from Anthropic." The Claude Code courses there are Claude Code 101, Claude Code in Action, Introduction to subagents, Introduction to agent skills, and The AI-native SDLC playbook.
• Anthropic's older Skilljar catalog (anthropic.skilljar.com) is being retired. Its courses have moved to Claude Academy, so use the Claude Academy links.
• Coursera hosts Claude Code in Action under Anthropic. On Coursera, the course materials and certificate require the paid Certificate experience, so the free Claude Academy version is the better deal unless you specifically want a Coursera certificate.
• DataCamp has Anthropic-created versions of Claude Code 101 and Claude Code in Action, adapted to its interactive format. DataCamp says both can be completed without a paid subscription, and both give you a Statement of Accomplishment.

About the Claude Code in Action course. It's the course to take once single prompts work and you want Claude Code to run longer on its own. The Claude Academy version has nine lessons: steering long sessions, a CLAUDE.md that gets followed, verification skills, permission modes, hooks, scheduled tasks and headless runs, GitHub Actions and code review, verifying unsupervised runs, and plugins. It takes about an hour and ends with a quiz and a completion badge. The DataCamp version lists Claude Code 101 as its prerequisite. Take that advice: In Action assumes you already know the basics.

All of these courses are free, but Claude Code itself is not. The quickstart lists a Claude subscription (Pro, Max, Ultra, Team, or Enterprise), a Claude Console account with API credits, or access through a supported cloud provider. Check current Claude pricing before you start.

This is where most "certification" claims fall apart, so here's what we could verify:

• There's no public Claude Code certification exam. Anthropic does run proctored certifications: Claude Certified Associate, Claude Certified Developer – Foundations, and Claude Certified Architect (Foundations and Professional), delivered by Pearson with Credly badges. But its certification FAQ says certification "is currently available only to organizations in the Claude Partner Network", and none of the four exams is described as a Claude Code exam.
• Claude Academy gives completion badges, not certificates, on Claude Code 101 and Claude Code in Action. The subagents, agent skills, and AI-native SDLC playbook courses have no badge.
• Free certificate: DataCamp's Claude Code 101 and Claude Code in Action each give a free Statement of Accomplishment.
• Paid certificates: Vanderbilt's Claude Code course and the Scrimba specialization on Coursera, plus a DeepLearning.AI accomplishment for the Claude Code short course if you have PRO.

Our take: a course certificate shows you finished a course, not that you can ship. For a job application, a public repo built with Claude Code (including its CLAUDE.md and a few merged PRs) says more than any badge.

About an hour a day. Use a real project you care about (a personal site, a script that annoys you, an open-source repo), not a toy. Every step below uses commands from the official docs.

Day 1: Install and explore. Install Claude Code (snippet below), open your project, and ask questions before changing anything: "what does this project do?", "where is the main entry point?". Watch the first half of Claude Code 101. Practice project: have Claude explain an unfamiliar open-source repo to you, then check its answers against the code.

Day 2: Plan before you code. Finish Claude Code 101. Switch to plan mode with Shift+Tab (or start with claude --permission-mode plan). Claude researches and proposes changes, but won't edit your source until you approve the plan. Practice project: plan and build a small CLI tool, such as a script that renames photos by date. Read the plan before you approve it.

Day 3: Give Claude a check to run. The single most useful habit from Anthropic's best-practices guide: give Claude something that returns pass or fail, such as tests, a build, or a linter, and ask for evidence (the command it ran and its output). Run /init to generate a starting CLAUDE.md, then cut it down to commands, conventions, and gotchas. Anthropic's memory guide recommends keeping it under 200 lines. Practice project: add tests to Day 2's tool, then ask Claude to add a feature and keep going until the tests pass.

Day 4: Steer and recover. Learn the three recovery moves. Esc stops Claude mid-action without losing work. /rewind (or Esc twice on an empty prompt) rolls code and conversation back to an earlier point. /clear starts fresh. The docs' rule: after two failed corrections, /clear and write a better prompt instead of a third correction. Practice project: attempt a risky refactor of your Day 2 tool, rewind if it goes wrong, and try a better prompt.

Day 5: Hooks and Claude Code in Action. Start Claude Code in Action. A line in CLAUDE.md is a guideline Claude usually follows. A hook is a rule Claude Code enforces every time. The Stop hook below (from the docs) won't let Claude finish while the typecheck fails. Swap in your own test command. Practice project: add a hook that blocks edits to a folder you never want touched (/hooks shows what you've configured).

Day 6: Parallel work and review. Take Introduction to subagents. Start a second task in its own git worktree with claude -w feature-name so the edits can't collide, and run /code-review to have a fresh subagent review your diff. Practice project: build two small features at once in separate worktrees, then review and merge both.

Day 7: Ship something. Pick one of the Claude Code projects below, build it end to end, and deploy it. Write the README yourself, because explaining what Claude built is how you find the parts you don't understand yet. Then try claude -p "query" to run a one-off task and exit. That's the building block for scripts and CI.

Good practice projects are small enough to finish in a day, have an obvious "it works" check, and make you use one new feature each. Ten that work well:

1. Personal link-in-bio site. Build it from a sketch or screenshot, deploy it, and add a CLAUDE.md with the deploy command.
2. CSV cleaner CLI. Takes a messy export and outputs clean data. Write the tests first, then let Claude implement.
3. Expense tracker web app. Local storage, charts, and filters. Use plan mode for the data model before any code exists.
4. README and docs generator for one of your old repos. Good for practising codebase exploration.
5. Bug-fix PR on an open-source project. Find a "good first issue", have Claude locate the cause, and write the fix with a test. Read every line before you open the PR.
6. Browser extension that strips tracking parameters from URLs. Small, real, and easy to verify.
7. Refactor an old project to a newer framework version, using a Stop hook to keep the build green.
8. Daily digest script that summarizes an RSS feed and runs with claude -p on a schedule.
9. A custom skill for something you repeat, such as your release checklist (see the agent skills course).
10. A small AI agent in Python. Claude Code is great at scaffolding one, and our free PydanticAI course shows you what the code should look like, starting with your first agent and tools.

Project 10 is where Claude Code users often get stuck: Claude writes the agent, but you can't judge whether it's good. Our course covers testing and evals and production deployment, which are the parts you need to be able to review.

• Keep the Claude Code cheat sheet open while you work through the 7-day plan.
• Not sure Claude Code is the right tool? See Claude Code alternatives for how it stacks up against other AI coding tools.
• New to coding, or more interested in building apps than learning a terminal tool? Start with the best vibe coding courses.
• Want to build AI agents, not just use one? Our free 8-lesson AI Agents with PydanticAI course goes from a first agent to a FastAPI + Docker deployment.