CCAO-F - Troubleshooting and Optimization (10% of the exam) - Section 7.1

Identify, diagnose, and resolve issues with underperforming prompts or poor outputs.

Working out why a result is poor, whether from missing context, an unclear request, the wrong feature or model, outdated Project knowledge, or an overlong conversation, and applying the fix that addresses that cause.

missing contextunclear instructionswrong feature or modeloutdated knowledge

Practice question for this objective

Free sampleTroubleshooting and Optimizationmedium

An operations analyst at a logistics company set up a Project months ago so depot managers could get quick answers, with the project instruction 'Keep every reply under 100 words.' She now uses the same Project to draft a quarterly performance review of eight depots. Each draft comes back as a brief overview that skips most depots, even when her request asks for full detail on every one. What is the most likely cause, and what should she do?

  • AThe conversation has grown too long, so start a new chat in the same Project and paste in a summary.
  • BThe depot data is too much for one reply, so ask Claude to cover two depots at a time and combine them.
  • CThe word limit applies to every chat in the Project, so draft the review in a separate Project without it. Correct
  • DClaude needs firmer wording, so put 'FULL DETAIL REQUIRED' in capitals at the start of each request.
A standing project instruction applies to every chat in the Project, so check it when outputs conflict with a request. Project instructions are applied to every conversation in the Project, which is what makes them useful for recurring work and also why an old rule can quietly undermine a new kind of task. The short replies match the 100-word instruction, not a fault in the request. Giving the review its own Project with suitable instructions removes the conflict without breaking the depot managers' setup.

Why A is wrong: Restarting with a summary is the right fix for drift in a long chat. A new chat in the same Project still carries the 100-word instruction, so the review would come back just as brief.

Why B is wrong: Splitting the work looks practical. Each reply is still held to the standing word limit, so every pair of depots gets a thin overview, and the real cause is left in place.

Why C is correct: Project instructions shape every chat in that Project, so the 100-word rule overrides the request for detail. Doing the review in a separate Project set up for long reports removes the conflict and keeps quick answers working for depot managers.

Why D is wrong: Stronger wording is a common reaction when Claude seems to ignore a request. It leaves two instructions in conflict, and the standing limit set for the whole Project keeps shaping the reply.

See more CCAO-F practice questions, answers explained.

Exam traps in Troubleshooting and Optimization

Answers that look right on this material and are not. Each one is a distractor from a different question in the CCAO-F bank for this domain.

  • Move the tone and length lines to the top of the instructions and keep the catalogue and framework below them.

    Why it is wrong: Putting the guidance first is a reasonable instinct, because order does matter in a prompt. It still leaves forty pages of reference material in the place meant for short behavioural guidance, so the two lines remain buried in a long block of text.

  • Add a project instruction telling Claude to prefer the newer document whenever the two disagree.

    Why it is wrong: An instruction looks like a quick way to settle the conflict without deleting anything. It is wrong because the contradicting file stays in knowledge, so Claude can still draw on the old threshold, and the officer is relying on an instruction to manage a problem that removing the file would eliminate.

  • Regenerate the same request several times and keep the draft that comes closest to the real offer.

    Why it is wrong: Regenerating feels quick and sometimes produces a better draft by chance. It is wrong because the request still lacks the offer and product details, so every version is a guess and the coordinator would be choosing the least wrong one rather than fixing the cause.

Examworthy is not affiliated with or endorsed by Anthropic. Original, blueprint-aligned practice material only.