Frequently Asked Questions
Quick answers about the exam, our site, analytics, and study strategies.
What is the Claude Certified Architect (Foundations) exam?
Anthropic's foundational certification for engineers who build with Claude. It tests practical knowledge across five domains: agentic architecture, MCP tooling, Claude Code, prompt engineering, and context management. It's scenario-driven, asking you to pick the best move in real-world production situations.
Is anthropicexams.com official or affiliated with Anthropic?
No. anthropicexams.com is an independent community resource built by and for AI engineers. We are not affiliated with, endorsed by, or sponsored by Anthropic. The official source of truth for the exam and its objectives is Anthropic itself.
How much does this prep site cost?
It is 100% free. Every mock exam, flashcard deck, and interview simulator is free. You don't even need to create an account to start using it.
Where is my progress saved?
By default, everything is stored locally in your browser's localStorage. This means your exam history, readiness scores, and study streak will persist on your device. However, if you want to sync your progress across multiple devices or ensure you never lose it, you can Sign In with Google. This securely syncs your local data to our Firebase cloud backend.
How does the Readiness Score in Analytics work?
Your readiness score is a weighted average of your recent mock exam scores across all the different modules (Developer Foundations, Certified Architect, etc). We track your most recent attempts and the number of questions you answer to build a statistical confidence rating of how prepared you are.
How is the mock exam scored?
Our mock exams mirror the real exam's weighting and pass threshold so you can properly calibrate before sitting the real test. Most exams are scored on a 1000-point scale where a score of 720+ is a passing grade.
Are there prerequisites for the real exam?
There are no formal prerequisites. However, you'll perform significantly better if you've shipped at least one project utilizing the Claude API, written a custom MCP tool, or used Claude Code on a real codebase.
Which Claude models does the exam cover?
As of 2026, the current production lineup (e.g. Opus 4.7, Sonnet 4.6, and Haiku 4.5) is featured heavily. The exam tests your ability to choose the right model for the right task (when to step up to Opus, when to step down to Haiku, when to mix them).