Some authentication flows use different PromptModel instances, which causes prompt text to vary default text in some places, app-customized text in others. This makes user messaging inconsistent and hard to maintain across platforms.
Expected behavior All flows should use the app-configured PromptModel, and prompt text should always be derived from:
Reason -> toHumanReadable

Why this matters

• Consistent user-facing prompt copy
• Single source of truth for prompt text
• Prevents drift between runtime behavior and docs