Anthropic Claude Academy Guide: Free Courses and Learning Paths

Anthropic's current first-party learning hub is Claude Academy. Anthropic introduced this version in its August 20, 2026 announcement, with free courses, tutorials, and use cases for learning how to use AI safely and effectively. The live all-resources directory includes more than courses and can change as new material is added.

This guide highlights useful starting points as of October 27, 2026. For the full catalog, lesson counts, access requirements, and completion options, use the live directory and each course page.

How to use Claude Academy

The directory lets you browse all resources or filter to Courses, Tutorials, or Use cases. Open a course to see its current lessons, estimated time, prerequisites, and whether it offers a quiz or completion badge. You can start a course immediately; signing in lets Academy save your progress.

Course access and product access are separate. The Academy homepage describes the learning resources as free, but the Claude 101 course page notes that some features used in its lessons require a paid Claude plan. Developer courses can also require an API key or a cloud account, and those services can have their own usage charges.

Free courses and learning paths at a glance

These are useful entry points, not an exhaustive list:

For everyone: start with Claude and AI Fluency

Claude 101

Claude 101 is the broadest starting point. Its current curriculum moves from a first conversation and clearer prompts to projects, artifacts, skills, connectors, and role-based use cases. It is designed for anyone with a Claude account on any plan and does not assume prior AI experience. Some hands-on features in the course require a paid plan, as the course page explains.

The 4D AI Fluency framework

Anthropic's AI Fluency Framework PDF defines four connected competencies:

• Delegation: setting goals and deciding whether, when, and how to engage with AI
• Description: describing goals clearly enough to prompt useful AI behavior and output
• Discernment: assessing the usefulness of AI outputs and behaviors accurately
• Diligence: taking responsibility for what you do with AI and how you do it

The framework is meant to support interactions that are effective, efficient, ethical, and safe. It also distinguishes three ways of working with AI: Automation, where AI executes a task from human instructions; Augmentation, where people and AI collaborate; and Agency, where a person configures AI to perform future tasks on their behalf.

The Framework and foundations course applies the 4Ds to prompting, evaluation, and responsible use. It is a good first course if you want transferable judgment rather than a product tour.

Choose an audience edition

The current AI Fluency collection includes editions for:

• Builders, with acceptance tests, product decisions, and evaluating AI-generated work
• Creative work, with authorship, distinctiveness, and intentional delegation
• Educators, with course design, learning materials, and academic integrity
• pK–12 educators and a pK–12 Train the Trainer course
• Nonprofits, with research, writing, data handling, and organization-wide adoption
• Small businesses, with everyday business workflows
• Students, with learning, career planning, and maintaining agency

The collection also lists Teaching AI Fluency. Choose the audience route that matches the work you want to do. Check each page for the latest prerequisites and duration.

For developers: choose the technical route

Claude Code

Start with Claude Code 101 if you need the mental model first. It covers the agentic loop, context, tools, permissions, installation surfaces, the Explore → Plan → Code → Commit workflow, context management, CLAUDE.md, subagents, Skills, MCP, and hooks. The course asks for basic familiarity with a code editor and command line, plus a Claude account or API key.

Move to Claude Code in action after you can run single prompts. It focuses on steering long sessions, permission modes, hooks, routines, headless work, GitHub Actions, code review, and verifying unsupervised runs. Basic Git and command-line familiarity are recommended.

Afterward, follow the hands-on First Hour with Claude Code tutorial on your own project.

After the Course: Try a Handoff in Termdock

This is an optional exercise, not a requirement of an Academy course. Open a disposable project in Termdock and have an installed Claude Code make a small change and report the diff and tests. In another terminal, ask an installed CLI to review the diff against the requirement without editing. Pass the requirement and paths explicitly and verify the response yourself. For agent-driven handoffs, see the Termdock walkthrough. CLI eligibility and costs are separate from course access.

Agent Skills

Introduction to agent skills is a short, current course on reusable Markdown instructions for Claude Code. It covers creating a valid SKILL.md, progressive disclosure, tool permissions, sharing through repositories and plugins, and troubleshooting. Our Write Your First SKILL.md tutorial is a practical companion.

Claude API and MCP

Building with the Claude API is the long developer route. The course page lists Python proficiency, JSON basics, and an Anthropic API key as prerequisites. It covers authentication, multi-turn messages, structured output, prompt evaluation, tool use, RAG, extended features, prompt caching, MCP, and agent workflows.

Introduction to Model Context Protocol is a narrower hands-on route. It uses the Python SDK to build servers and clients around tools, resources, and prompts, then tests them with the MCP Server Inspector. The current page recommends Python, async/await, API concepts, and an API key for the later application exercise.

Cloud deployments

If your team already uses a cloud platform, choose the matching course rather than treating it as a generic Claude API tutorial:

• Claude with Amazon Bedrock covers Claude through AWS Bedrock and lists Python, JSON, and Bedrock access as prerequisites.
• Claude on Google Cloud covers Claude through Google Cloud's Agent Platform and lists Python, JSON, Google Cloud access, and a Claude API key for one of its projects.

Both pages include production topics such as tool use, RAG, MCP, and agent workflows. Cloud account access and model usage are separate from the Academy course itself.

Learning resources beyond Academy

• Claude API Docs are the reference for endpoints, parameters, authentication, and current API behavior.
• Claude Code Docs cover installation, permissions, configuration, and advanced workflows.
• Anthropic's courses repository contains notebook-based learning material. Its contents and model examples can change, so check the repository README before running anything. API requests may incur charges.
• The Anthropic Cookbook is a reference collection for implementation patterns such as document processing, citations, and embeddings.
• Claude tutorials add current product and workflow examples outside the course catalog.

Where should you start?

You do not need to complete everything. Pick the route closest to your goal:

"I have never used Claude." → Claude 101 → AI Fluency: Framework and foundations

"I want to understand AI limits before relying on it." → AI capabilities and limitations → AI Fluency: Framework and foundations

"I want to build software with Claude Code." → Claude Code 101 → Claude Code in action → First Hour with Claude Code

"I am building a Claude-powered product." → Building with the Claude API → Introduction to Model Context Protocol → the cloud course that matches your platform, if needed

"I am a teacher, student, nonprofit worker, or small-business owner." → Start with AI Fluency: Framework and foundations, then choose the matching edition from the AI Fluency collection.

Check the live Claude Academy directory before enrolling. It is the current source for what exists, how long a course is, and what badge or assessment it offers.