Foundation
Output: Diagnostic + first concept mastered
Take the 10-Q diagnostic to find your weakest domain. Master one concept end-to-end so you know what "mastered" means here (dwell 2 min + 3 correct practice Qs).
Every Claude certification course on this site is free, with no paid tier and no account needed to start. Six tracks: a beginner course, the month-by-month CCA-F exam path, 71 concept lessons grouped by exam domain, Anthropic Academy's own lessons mirrored, and a scored mock to prove it. New here? Take the free 10-minute diagnostic first. Comparing your options first? Every Claude training route, and what each costs. Targeting a different Claude credential? Compare all four at the certifications hub.

No paid tier, and no account needed to start. The only thing that costs money on the whole path is the exam itself.
They run in this order, but nothing is locked. Start wherever you are.
No prior Claude experience assumed. What Claude is, what an agentic system actually does, and the shortest honest path from nothing to your first mock exam.
Start Claude 101 →The blueprint-weighted study path below. One focus per month, one output, and the pillar links where the work happens. Diagnostic first - everything routes from there.
See the plan →Terminal sessions, durable CLAUDE.md context, hooks that make rules mechanical, and subagents that hand off state cleanly. Free to read; Claude Code itself needs a paid plan.
Start the Claude Code course →Each technical primitive on its own page - what it does, why the exam cares, and the anti-pattern that catches people. Grouped by the five weighted domains.
Open the concepts →Anthropic's own Skilljar lessons condensed to their exam-critical points, with citations back to the source. Read these to see how Anthropic teaches a topic.
Open the mirrors →A 10-question diagnostic that needs no account, a weakest-domain warm-up, and a 30-question mock scored to 1000 against the real 720 pass mark.
Open practice →The exam isn't evenly distributed. Domain 1 alone is 27%. Spend your prep proportionally.
Each month has one focus, one primary output, and the pillar links to use. Diagnostic first - everything else routes from there.
Output: Diagnostic + first concept mastered
Take the 10-Q diagnostic to find your weakest domain. Master one concept end-to-end so you know what "mastered" means here (dwell 2 min + 3 correct practice Qs).
Output: Agentic loops, MCP, tool_choice
Domains 1 + 2 carry 45% of exam weight combined. Master subagent context isolation, tool description quality, and the 18-tool degradation cliff first.
Output: Component + failure-mode reading
Scenarios make abstract concepts concrete. Walk through 4-6 of the 25 - especially conversational-ai-patterns and agent-skills-for-enterprise-km.
Output: Anti-patterns + cheatsheets
The Reference pillar consolidates the 8 anti-patterns, the 2 distractor patterns, and 3 downloadable posters. Print them; reread 24 hours before the exam.
Output: Weekly mocks + targeted drills
Take the 30-Q mock. Score below 720? The result page builds an adaptive plan — two, four or eight weeks — keyed on your specific misconceptions, not a generic syllabus.
Output: Adaptive plan closes weak areas
Re-run mocks weekly and watch your mock score climb toward the 720 bar. The Exam Guide pillar covers logistics, registration, and day-of distractor patterns.
Every Q is phrased as a real Google search query. Answers cite the same evidence-tagged sources used elsewhere on the site.