The prompt forces the model to confess when it lacks a complete solution—no guessing allowed. This keeps answers honest and prevents hallucinations, giving users a clear signal of confidence and highlighting exactly where extra help or verification is needed.