[BUG] CLI: explanatory text before AskUserQuestion completely swallowed — only bare options render

Preflight Checklist

• [x] I have searched existing issues and this hasn't been reported yet
• [x] This is a single bug report
• [x] Version checked against latest (on 2.1.220; changelog for 2.1.221/2.1.222 lists no related fix)

What's Wrong?

In the CLI, when Claude writes an explanatory message and then calls AskUserQuestion in the same turn, the explanatory text is completely swallowed. Once the turn finishes working, only the bare option picker is shown — there is no explanation, context, or question preamble to go with the choices. The user has to answer (or guess) without ever seeing the reasoning Claude produced for the decision.

This is not the last-line clipping bug: nothing of the preceding message renders at all, not just its final line.

Distinct from existing issues:

• #58207 / #30422 — popup hiding only the last line of preceding chat content; fixed in 2.1.141. This report is about the entire message being absent, on 2.1.220 (which includes that fix).
• #64776 — same class of complaint but for the Desktop app (message squeezed into a 3–5 line window); closed as not planned. This report is about the terminal CLI, where the text is not squeezed but entirely missing.

What Should Happen?

The assistant text preceding an AskUserQuestion call should render in the conversation view before/above the option picker, so the options can be understood in context ("Don't Make Me Think" — options without their explanation are frequently meaningless).

Error Messages/Logs

(none — pure TUI rendering issue, no errors emitted)

Steps to Reproduce

1. Start an interactive claude session in Windows Terminal.
2. Give the model a task where it explains its findings and then asks the user to choose between options in the same turn (any prompt that causes a text block immediately followed by an AskUserQuestion tool call).
3. Wait for the turn to finish working.
4. Observe: only the option picker renders. The explanatory text that Claude generated before the tool call never appears in the conversation view.

Claude Model

Other — Fable 5 (claude-fable-5)

Is this a regression?

Yes, this worked in a previous version

Last Working Version

Unknown — preceding text used to display before the options dialog in earlier versions; exact last-good version not recorded.

Claude Code Version

2.1.220 (Claude Code)

Platform

Anthropic API

Operating System

Windows

Terminal/Shell

Windows Terminal

Additional Information

• Windows 11 Pro (10.0.26200).
• Not yet verified whether the swallowed text is recoverable in the transcript view (Ctrl+O) or reappears after cancelling the dialog — in the normal conversation flow it is simply absent.
• Happy to provide a session recording if useful.