CCAO-F Prompting and Context Engineering Practice Question

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✓

  Use a prefilled assistant turn that begins the JSON, so Claude continues from a known starting point.

  Why this is correct

  Prefilling the assistant turn with the opening of the JSON, such as an opening brace or the first key, constrains Claude to continue in that format rather than starting with prose. It is a well-known technique for enforcing output shape in the Claude Messages API. Combined with an explicit schema, it makes the response reliably parseable.

• ✗

  Set temperature to its maximum so Claude considers many possible JSON layouts.

  Why it's wrong here

  Maximum temperature increases variability, which is the opposite of what structured extraction needs. It raises the chance of inconsistent keys, extra fields, or malformed JSON. Reliability comes from constraining the output shape, not from encouraging creative variation. For fixed-schema extraction, low or zero temperature is generally preferable.

• ✓

  Provide an explicit schema or example JSON in the prompt and instruct Claude to return only JSON matching it.

  Why this is correct

  Showing the exact target shape, whether as a schema description or a concrete example, gives Claude an unambiguous contract for field names, types, and nesting. Combined with an instruction to emit only that JSON, it sharply reduces stray prose and key drift. This is a core technique for reliable structured extraction in Claude prompts and directly supports downstream parsing.

• ✗

  Precede the JSON with a short natural-language explanation so a human can verify the result.

  Why it's wrong here

  Adding prose before the JSON makes the response harder to parse automatically and can encourage Claude to include commentary inside or around the object. If human review is needed, it is better handled outside the machine-parsed payload. For reliable structured output, the goal is to minimize non-JSON content, not to invite it.

• ✗

  Ask Claude to decide at runtime which fields are most relevant and include only those.

  Why it's wrong here

  Allowing Claude to choose fields breaks the fixed schema the downstream code expects. The scenario requires exactly "vendor", "amount", and "currency" every time. Making field selection dynamic introduces missing or renamed keys and defeats the purpose of a stable contract. The schema should be fixed by the prompt, not negotiated by the model per call.