Draft Japanese answers for an internal FAQ.
Use only the supplied document and return bullet points.
Which prompt design is most appropriate?
Specify the role, reference context, language, output format, and behavior when the context has no answer.
Detailed explanation
The model lacks the context and expected output shape.
Input confidentiality still requires authorization and data-protection design.
Explicit goals and constraints make the response easier to evaluate.
A claim of guarantee does not provide evidence or a useful constraint.
Check official documentation
Compare the following points with the official documentation.
Compare Task 3.2 on instructions, context, and specificity with the Bedrock prompt-design examples. For an FAQ with and without supporting text, list checks for language, bullet format, and evidence boundaries.
Official references
What to confirm
Distinguish role, reference context, output format, and missing-evidence behavior, and explain why instructions alone do not guarantee factual accuracy.
Key points
- Role
- Context
- Format and constraints
Notes
- Environment: Reference review of the official AIF-C01 guide and AWS documentation; no AWS resources or inference are required.
- Include a question unsupported by the source and check the intended behavior for missing evidence.
- This is a reference and design review; no model inference result is claimed.
Foundation review
Instruction hierarchy
System, application, and user inputs may have different authority and must be separated deliberately.
Output contract
A defined format and refusal behavior make the response easier to validate downstream.