Claude Certified Developer – Foundations prep: see where you stand before you study
Claude Certified Developer – Foundations (CCDV-F) is Anthropic's proctored certification for developers building on the Claude API: 53 questions in 120 minutes, $125 an attempt, and a pass mark of 720 on a 100–1,000 scaled score. Before you spend a week revising, find out where you stand. Alloy's diagnostic is free, needs no card, and takes about 20 minutes.
What the diagnostic is
Twenty-four questions drawn from the same question bank, under the same rules, as everything else in the course. It is not a quiz bolted on the side, so what it reports is where you stand against this exam rather than against a warm-up. The questions are the kind the exam asks: a situation, a constraint and a decision — which fix is targeted, which layer the defect lives in, which configuration fits.
Why it comes before anything else
On CCDV-F the domains are wildly unequal, and a developer's instinct about where the marks are is usually wrong. Applications and Integration is 33.1% of the exam — about 18 of the 53 items — while Claude Code (3.1%) and Eval, Testing, and Debugging (2.6%) together are about three. The exam is scored on the total scaled score only, with no separate bar for any domain, so an hour that lifts your accuracy by ten points in Applications and Integration is worth about 1.8 items, and the same hour in Eval, Testing, and Debugging is worth about 0.1. Knowing which of your weak spots sit in heavy domains is worth a week of studying the wrong thing.
The instinct is to revise a little first so the result is not embarrassing. That instinct costs you the week. A diagnostic taken after a warm-up measures the warm-up.
How it is shaped
Weighted like the exam. The areas it draws from, and how heavily it draws from each, follow the exam's published weights across its eight domains: Applications and Integration (33.1%), Model Selection and Optimization (16.8%), Agents and Workflows (14.7%), Prompt and Context Engineering (11.0%), Tools and MCPs (10.6%), Security and Safety (8.1%), Claude Code (3.1%) and Eval, Testing, and Debugging (2.6%).
Spread inside each area. Within a domain it covers different tasks rather than asking the same thing four times, so a weak result is a real signal, not one unlucky question.
Fixed length, one sitting. The course sets its length and Alloy will not let you shorten it. A diagnostic that quietly got smaller would measure less and still report the same confidence; a partial diagnostic is not a smaller one, it is a worse one.
A confidence rating on every answer. You cannot submit without saying how sure you are. Alloy watches the gap between how sure you were and how right you turned out to be. Unsure and wrong, you go back and check; certain and wrong, you never look at it again, and it walks into the exam with you.
What you get back
A starting readiness figure and, for every domain, how many questions you saw and how many you got right. Read them on the summary screen while they are in front of you.
They are counts, not percentages, deliberately. With eight domains and 24 questions, a domain gets about three. Three questions can only come out four ways, and one lucky guess would move a percentage by a third, so a per-domain percentage would claim a precision three questions cannot give. Anthropic takes the same position on the real exam, which is more than twice as long: its per-domain percentages are informational and never decide a pass. Use the counts to decide what to open first.
What to do next
- Open your weak domains heaviest first — with one exception. A domain where you are near zero, rather than merely weak, can be worth more than its weight suggests: security items are decidable once you know the failure modes and undecidable when you do not.
- Run the confident-and-wrong set. Alloy can hand back exactly the questions you were sure of and got wrong — the most valuable practice in your preparation, available only if you rated honestly.
- Follow Recommended next when you are unsure what to study. It weighs what you are about to forget, where you are weakest and the heavy material you have not opened, and puts one action in front of you.
- Read the error before the options as you practise. Decide whether a defect is schema validation, message ordering, transport or authentication, and discard every option outside that class.
Mistakes to avoid
- Warming up before the diagnostic.
- Rating every answer at the top of the scale — it throws the signal away.
- Reading three questions as a percentage.
- Spending the first week on Claude Code and evals because they are the most enjoyable topics.
- Treating the starting readiness figure as a forecast of your exam score; it is where you begin, and it moves with the work.
What is free
A free account has no card and no trial clock. It includes the full diagnostic on any of Alloy's four Claude certification courses, the first two lessons of every course with their podcasts — on CCDV-F, the exam-strategy lesson and the lesson on what a request costs, why it varies and how it reaches Claude — and a place to store and verify certifications you already hold.
Start free, take the diagnostic, then study what it tells you to study.
Alloy is exam preparation built from the official exam guides — only what the exam asks for.