Full simulation · free · no sign-up
CCDV-F timed mock exam
A full-length Developer Foundations practice exam under real conditions: 53 questions in 120 minutes, drawn in the same domain proportions as the published CCDV-F blueprint, with nothing revealed until you submit. It is scored on the same 100–1000 scale the real exam uses, and the report tells you how far you landed from the 720 cut.
Loading…
What the CCDV-F mock exam covers
The form is drawn in the proportions the published CCDV-F guide gives each domain, which means it is far from even: Applications and Integration takes 17 of the 53 items and Eval, Testing, and Debugging takes 1. Within a domain the draw is stratified by objective, so every objective contributes an item before any contributes a second.
| Domain | Blueprint weight | Items drawn |
|---|---|---|
| D1 · Agents and Workflows | 14.7% | 8 |
| D2 · Applications and Integration | 33.1% | 17 |
| D3 · Claude Code | 3.1% | 2 |
| D4 · Eval, Testing, and Debugging | 2.6% | 1 |
| D5 · Model Selection and Optimization | 16.8% | 9 |
| D6 · Prompt and Context Engineering | 11% | 6 |
| D7 · Security and Safety | 8.1% | 4 |
| D8 · Tools and MCPs | 10.6% | 6 |
A consequence worth knowing before you read a mock report: with 25 objectives and 53 items, a single form cannot reach all of them. The thin domains rotate, so two consecutive mocks will test different corners of them. Read a zero on a one-item objective as no information rather than as a gap.
How the mock is scored
Your raw percentage is converted onto the 100 to 1000 scale the real exam uses, and the report shows the margin between where you landed and the 720 cut. The marking rules are the strict ones rather than a friendlier version.
- Single-answer items are marked on the exact option. Nothing is given for being close.
- Multiple-response items need the whole set and only that set. No partial credit.
- Matching items are all-or-nothing across every pair in the item.
- Anything left unanswered counts as wrong. A guess and a blank score the same when the guess is wrong, and better when it is not.
Where this differs from the real exam
These questions are original, written against the published exam guide and Anthropic's own developer documentation. None of them is real exam content, and a practice set claiming otherwise would be telling you it had broken an exam agreement. A good score here is evidence that you know the objectives rather than a forecast with a confidence interval.
The material ages differently too, and this exam feels it more than the architecture ones. Model generations change which parameters a request may carry, defaults flip, and a configuration that returns a 400 today was the documented approach a year ago. These items were written to turn on mechanisms rather than on spellings, which holds up better and does not make them immune.
How long should I spend per question?
120 minutes across 53 items is a little over two minutes each, which is more room than the architecture papers give. The diagnostic items are where it goes: a stem describing a symptom takes real reading before the options mean anything. Flag what runs long and come back to it.
Does the clock keep running if I close the tab?
Yes. The countdown runs against the wall clock rather than against time spent on the page. The sitting is saved and can be resumed, but it comes back with the elapsed time already deducted.
Can I sit the CCDV-F mock more than once?
As many times as you like. The form is redrawn from the 200-item pool each time and the options are reshuffled, so a repeat is not the same paper. Repeats do get easier for the wrong reason once you have seen most of the pool, which is worth remembering before reading much into a rising score.
Sitting a mock cold is a diagnostic. The drills, the mistakes deck and the 106-item run are what you do between attempts.
All CCDV-F practice modes