CCAO-F Prompting and Context Engineering Practice Question
A fintech developer builds an assistant that answers questions about account activity. The system prompt currently says: 'You are a helpful banking assistant. Use the provided account data to answer questions.' Testing shows Claude sometimes answers general banking questions from its own knowledge rather than from the supplied data, and occasionally states figures that are not in the data at all. Which revision to the system prompt best addresses both problems?
⚠ Common exam trap
The trap here is believing that a general instruction to be accurate is equivalent to a grounding constraint, when grounding requires naming the permitted source and defining behavior for missing information.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Add: 'Answer only from the account data in <data> tags. If the answer is not present, say you cannot find it in the provided data. Do not answer general banking questions.'
The effective revision names the permitted source, provides a fallback response when the data does not contain the answer, and constrains the assistant's scope. Source restriction stops substitution of outside knowledge, the fallback gives the model a safe alternative to guessing, and the scope limit removes the general-question path that produced unsupported figures.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Add: 'Always be accurate and never make up information.'
Why it's wrong here
The instruction names the desired outcome without specifying the mechanism. Claude still has no rule telling it to prefer supplied data over internal knowledge, nor a defined response when the data is silent. Vague accuracy directives are known to have weak effect compared with concrete constraints on source and fallback behavior.
- ✗
Move the account data into the system prompt instead of the user turn.
Why it's wrong here
Relocating the data changes where it appears but not the instruction governing its use. Without a rule that the data is the sole source, the model may still blend in outside knowledge or invent figures. Placement is a secondary concern here; the primary defect is the absence of a grounding and refusal policy.
- ✓
Add: 'Answer only from the account data in <data> tags. If the answer is not present, say you cannot find it in the provided data. Do not answer general banking questions.'
Why this is correct
This revision closes both gaps at once. Restricting answers to the delimited data prevents the model from substituting its own knowledge, and the explicit refusal instruction supplies a sanctioned response when the data lacks the answer. Declining general banking questions keeps the assistant inside its intended scope and removes the main path to unsupported figures.
- ✗
Add five examples of correct answers to common account questions.
Why it's wrong here
Examples of correct answers may improve tone and format, but they do not establish that the supplied data is the only permitted source. The model can still answer general banking questions from pretraining knowledge and can still produce figures absent from the data. The missing control is a source restriction, which examples alone do not supply.
About these practice questions
One of 259 original CCAO-F practice questions on Courseiva, each with a full explanation and wrong-answer analysis — not exam dumps or protected exam content. Learn why practice questions differ from exam dumps →
JA
Written and reviewed by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
Last reviewed September 2026 · checked against the official Anthropic exam blueprint
This CCAO-F practice question is part of Courseiva's free Anthropic certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the CCAO-F exam.