| Item bank | 163 items across 6 scenarios and all 30 task statements. 36 are contrast pairs: near-identical setups where one detail changes the answer. |
| Item formats | Multiple choice and multiple response. Each item states how many responses to select. |
| Passing score | 720 on a 100 to 1,000 scale. Scored by blueprint weight rather than raw item count, so a weak high-weight domain costs what it would on the real exam. |
| Domain composition |
Each test is a fixed 60-item form built to the blueprint weights (16/11/12/12/9), timed at 120 minutes. Options are reshuffled on every attempt. Your best score per test is stored in this browser.
Any length, built to the same weights. Timed runs allow 2 minutes per item. Each run continues from where the last one stopped in the bank, so consecutive runs draw different items.
Exam guide v1.0 · section 9 Anthropic publishes 12 sample questions in the exam guide itself, to show the format and difficulty of the real thing. They are reproduced here word for word, with their official explanations, and each one is marked as coming from the guide wherever it appears.
Everything else in this bank is original, written against the guide's task statements. These are the only items that are not.
Missed items are stored in this browser. Options are reshuffled every time, so answer position is never a clue.
Untimed. The rule and the trap are shown after each answer.
The guide lists these as not tested. Some practice banks include them anyway; a question that turns on one is padding.
As on the real score report, domain percentages are informational; pass or fail is decided by the total scaled score only.
Rules behind the items you missed, most-missed first, with the guide sections to re-read.