Skip to the main content

Exam technique

Claude Certified Architect wrong-answer patterns

Wrong options on the Claude Certified Architect (CCAR-F) exam recur in six shapes. Name the shape and you can cross an option out without weighing it.

SpacedPublished 2 min read

A recurring wrong answer on the CCAR-F is the option "use a larger context window". It sounds responsible, and it fails because the question describes a quality problem. Wrong options on this exam come in a small number of recurring shapes, and once you can name the shape you can cross the option out without weighing it.

1. The capacity fix

What it looks like: a larger context window, a higher-tier model, more retries, a bigger iteration cap.

Why it fails: the problem is quality. Doubling a budget that is being spent badly leaves it spent badly. Spotting this pattern is close to a free elimination.

2. Wrong side of the line

What it looks like: a check that runs after the action (a PostToolUse hook, an audit log, an alert) when the question demands that something never happens.

Why it fails: it fires after the harm, so it detects what the question wants prevented. On enforcement questions it sounds like control, so it draws a careful reader.

Rule · remember thisLook for the guarantee in the questionWords like never, every or cannot be bypassed mean only something that runs before the action can be right. A PreToolUse hook prevents; a PostToolUse hook records.

3. The over-engineered fix

What it looks like: a classifier, a custom server, a retraining loop, where rewriting a tool description would have done.

Why it fails: it is out of proportion to the cause, and it may not enforce what it claims to. The exam favours the smallest change that reaches the cause.

4. The premature fix

What it looks like: something correct that is not first. Consolidating tools, splitting a pull request, answering a question nobody asked.

Why it fails: the action is right and the order is wrong. The pattern fits when the ask is the best first step, which is why root-cause and first-step questions about the same scenario have different right answers.

5. The symptom tuner

What it looks like: a change to a retry count or a threshold that leaves the thing producing the failure untouched.

Why it fails: it treats the symptom and leaves the cause untouched. If a tool keeps returning an error the model cannot act on, retrying it more often only fails more often. Fix the shape of the response.

6. The invented mechanism

What it looks like: a flag, an environment variable or a configuration value that sounds plausible and does not exist.

Why it fails: the option names something that is not in the product. If you have prepared, options that name a setting or feature that does not exist are free marks.

Trap · looks right, isn’tMatch each option to a pattern firstEvery pattern here sounds reasonable on its own. Before you weigh an option on its merits, ask which pattern it is; if it is one of these, it is out.

Some options are wrong in every scenario. The post Never the answer on the Claude Architect exam lists them. For the order to apply all of this in, see how to answer CCAR-F questions.

All posts

Ready to learn? Let’s get started.

Three questions and a short baseline quiz, and you have a plan. Whether you are booked weeks out or panicking days out, you can start tonight, with or without an exam date.

Start learning

Domain 1 is free. No card needed.