Finding the official learning path

When you introduce Claude Code or MCP into real work, the first friction is often finding the official learning path.

📑Table of Contents
  1. What Anthropic Academy is (Learn hub vs Skilljar)
  2. Free course comparison for developers (Claude Code / MCP / Skills)
  3. Recommended path — what to take in the first ~90 minutes
  4. Account and data boundaries (Skilljar vs Anthropic)
  5. Role-based course selection
  6. FAQ
  7. Summary — what to do next

Anthropic Academy (the Learn hub) and the free Skilljar courses give a separate track from product launch posts or usage-limit promotions: they help you decide which free course to start with and how to reach a completion certificate. Treat catalog listings as time-sensitive: launch posts and social recaps can lag behind live prerequisites and certificate wording, so locking team study time from a snapshot alone creates the risk of a mismatched syllabus.


Before you lock the schedule

Before you schedule, do the following:

  • Open the live course page
  • Confirm current prerequisites, quiz, and certificate wording
  • Commit to only the first course (usually Claude Code in Action) until that path is verified

What this article covers

This article maps:

  • what Anthropic Academy is
  • the three developer core courses (Claude Code in Action, Introduction to Model Context Protocol, and Introduction to Agent Skills)
  • a first-90-minute study path
  • Skilljar versus Anthropic account boundaries
  • role-based selection
  • how third-party free Claude Code classes differ from official Academy certificates

What Anthropic Academy is (Learn hub vs Skilljar)

Bottom line: Discovery lives at anthropic.com/learn; enrollment, progress, and certificates live on the LMS at anthropic.skilljar.com.

Learn hub and Featured courses

Anthropic brands the hub as Anthropic Academy—resources that span API development guides through enterprise deployment practices. Featured courses include Claude Code in Action and Introduction to Cowork. The hub states that new courses cover AI Fluency, API development, Model Context Protocol, and Claude Code, and that learners can earn certificates upon completion. The primary catalog entry point from the hub is Skilljar.


Why this matters for readers

For readers, that means:

  • You onboard through a curriculum, not only scattered docs
  • “I want Claude Code” and “I want Anthropic’s intended learning order” meet in one place
  • After reading release notes or weekly-limit promotions, you can return here as the practical onboarding door

Second official catalog surface

A second official catalog surface is claude.com/resources/courses, with Product filters (Claude Code, MCP, Platform, Fluency, and more) and Category filters. Use it once your role is clear.


Free course comparison for developers (Claude Code / MCP / Skills)

Bottom line: The shortest useful core is three courses: Claude Code in Action first; MCP intro when you need tool connectivity; Agent Skills when you need reusable team workflows.

Core three-course comparison table

Course Size (listed) Outcomes Certificate Prerequisites
Claude Code in Action 15 lectures / ~1h video + quiz tools, context, MCP, GitHub, hooks Certificate of completion on Skilljar Basic CLI; Claude Code + API key
Introduction to Model Context Protocol 16 lectures / ~1h video + quiz servers/clients, tools/resources/prompts, Inspector Same Python, async/await, API basics
Introduction to Agent Skills ~6 curriculum modules (~30 min on some listings) SKILL.md, sharing, troubleshooting, vs other features Follow the course page (free registration) Already using Claude Code

Official sources (as of July 2026)


Claude Code in Action

Claude Code in Action walks through core tools (files, shell, analysis), context with /init, Claude.md, and @ mentions, Plan/Thinking modes, custom commands, MCP servers, GitHub integration, and hooks.

Independent write-ups (for example on Zenn) report finishing the short lessons in about an hour and earning a completion certificate.


Introduction to MCP

Introduction to MCP focuses on building modular AI apps that connect Claude to external tools and data without hand-writing tool schemas.

  • Servers and clients
  • Resources and prompts
  • The MCP Inspector
  • Often via a document-management hands-on project (assumes the listed Python / async / API basics; without that baseline, finish CLI Claude Code first or budget extra setup time—starting MCP intro under-prepared risks stalled labs and incomplete quizzes)

Introduction to Agent Skills

Introduction to Agent Skills teaches reusable markdown Skills: create, configure, share, and troubleshoot them, and contrast them with CLAUDE.md, hooks, and subagents.

It is most useful after you already run Claude Code day to day.


Broader catalog and third-party maps

The broader catalog also lists Claude Code 101, Introduction to subagents, Claude Platform 101, and the long-form Building with the Claude API track (~8 hours). You do not need all of them first.

Independent maps (Qiita, note, Iqbal Mahmud’s free roadmap) are useful as third-party confirmation that Skilljar hosts a large free catalog. Course counts and free-track claims in those maps can lag, so treat them as discovery only—keep primary facts anchored to official course pages and re-check the live catalog before you lock a training plan.


Bottom line: Finish a path that gets Claude Code working in your repo before trying to complete the whole catalog.

Study checklist

  1. Open Anthropic Learn and go to Featured or the full Skilljar catalog
  2. Create a Skilljar account (an Anthropic account is not required just to view learning content)
  3. Complete Claude Code in Action, including the quiz, and collect the certificate
  4. If external tool wiring is next, take Introduction to MCP
  5. If you need reusable team procedures, continue with Introduction to Agent Skills
  6. Add Cowork, AI Fluency, or long API tracks only when the role matches

Immediate next actions after the course

Do not stop at the certificate. In your own environment, try one of:

  • Run /init and add a CLAUDE.md to one repository
  • Add a single hook for permissions, notifications, or formatting
  • Connect one MCP server
  • Write one Skill for a repeated task

That “official curriculum → one feature in your repo” loop is the most repeatable reader next action. Expanding every setting on day one tends to create context bloat and approval fatigue.


Account and data boundaries (Skilljar vs Anthropic)

Bottom line: The learning LMS and Claude product accounts are separate systems. Mixing them causes “I have a certificate but cannot run Claude” or “I thought a third-party class was the official cert” mistakes.

Boundary table from the official FAQ

Per the official FAQ on Claude Code in Action:

Topic Detail
Learning access Skilljar account only; Anthropic account not required for course content
Using Claude Separate claude.ai / Claude Code / API account or key
Learning data Progress, lesson completion, quiz scores, time spent (for certificates)
Hosting Skilljar hosts educational content, separate from Console / Claude AI accounts
Deletion requests academy-support@anthropic.com
Compliance note Skilljar described as SOC 2 compliant on the course privacy FAQ

Re-verify before policy work

Source: Claude Code in Action privacy FAQ (July 2026).

Re-verify hosting and compliance claims on the live page before using them in policy work—FAQ text can change after that snapshot.

If the live wording differs, update your internal policy draft from the current page rather than this article’s dated summary.


Third-party free classes vs Academy

DataCamp’s Claude Code 101 can be a strong independent onboarding path covering install through Plan Mode, subagents, skills, MCP, and hooks.

It is not an Anthropic Academy / Skilljar certificate. If you need an official curriculum signal for HR or internal training records, use Skilljar’s Certificate of completion.


Social discovery signals vs official checks

⚠️ Note: Social posts that list “13+ free courses” or job-market framing are discovery signals only.

Before you enroll, verify the following on the official Skilljar/Learn pages:

  • lecture counts (use the official page, not social recaps; social numbers can be stale)
  • quizzes and stated prerequisites (confirm before booking hands-on time)
  • certificate badges and whether free completion is still offered (do not assume a prior snapshot still applies)

Role-based course selection

Bottom line: Rank courses by role, not by catalog completeness.

Priority table by role

Role / goal Prioritize Defer Decision cue
CLI agent adoption Claude Code in Action → Agent Skills → subagents Full API 8h track, all Fluency Get tools/context/hooks working in one repo; confirm API key and local permissions before scheduling class time
Tool / data connectivity MCP intro → MCP Advanced Topics Cowork intro Understand servers/clients and tools vs resources first; do not start here if Python/async basics are missing
App / platform building Platform 101 / Building with the Claude API Deep Claude Code catalog API is the main workplace; Claude Code tracks alone will not cover platform depth

Org literacy and team standardization

Role / goal Prioritize Defer Decision cue
Org AI literacy AI Fluency tracks, Introduction to Cowork MCP Advanced Non-engineers need a shared vocabulary first; do not equate literacy tracks with CLI agent readiness
Team standardization Agent Skills + sharing settings Fluency-only paths Reuse and permission boundaries matter more than personal tips; Skills after Claude Code is already in use

Individual engineers vs team leads

For an individual engineer, reduce choice to two branches after Claude Code in Action:

  • Need MCP?
  • Need Skills?

For team leads, design how Skills will be shared (repo, plugins, managed settings) after a personal pass so later adoption does not fragment.


FAQ

Q1. Are the courses really free through the certificate?

A. The Learn hub advertises certificates. Claude Code in Action and Introduction to MCP show Certificate of completion on Skilljar. Enrollment and progress live on Skilljar. Always re-check the live page before you claim a policy to stakeholders.


Q2. Which course should I start with for Claude Code?

A. Start with Claude Code in Action (tools, context, MCP, GitHub, hooks). Add Claude Code 101 or subagents only after you feel a concrete gap.


Q3. Do I need an Anthropic account?

A. Not for watching Academy learning content on Skilljar. You do need a separate Anthropic/claude.ai account or API key to run Claude Code in production-like work.


Q4. How do third-party free Claude Code courses differ?

A. Academy/Skilljar is Anthropic’s official curriculum. Independent courses (for example DataCamp) are useful supplements, not the same certificate.


Q5. Should I take the long API course first?

A. Usually no. Long tracks pay off when API implementation is already your main job. The first win is safely running one Claude Code capability in your environment.


Summary — what to do next

Anthropic Academy is the official curriculum door for Claude Code, MCP, and Agent Skills. Keep it separate from launch news and limit promotions—treating those as the syllabus can mis-set prerequisites and certificate expectations—and follow this order after a live-page re-check:

Next-step checklist

  1. Register via the Learn hub into Skilljar
  2. Finish Claude Code in Action and collect the certificate
  3. Continue with MCP intro or Agent Skills only if needed
  4. Immediately try one of hooks, MCP, or Skills in your own repo
  5. Treat third-party classes as optional practice, not official proof

That sequence uses the free catalog without drowning in it, and turns Academy into a practical onboarding path rather than a bookmark list.

Related articles:

Related new article:

krona23

Author

krona23

Over 20 years in the IT industry, serving as Division Head and CTO at multiple companies running large-scale web services in Japan. Experienced across Windows, iOS, Android, and web development. Currently focused on AI-native transformation. At DevGENT, sharing practical guides on AI code editors, automation tools, and LLMs in three languages.

DevGENT about →

Leave a Reply

Trending

Discover more from DevGENT

Subscribe now to keep reading and get access to the full archive.

Continue reading