Feature request: interactive UI elements for structured user choices (multi-select, action menus)

Problem

When Claude Code presents structured options to the user (e.g., a list of next steps after completing a task, or a set of files to review), the only way to collect a response is via AskUserQuestion with free-text input. The user must type something like "задачи 1,3, начать 2" — which works but feels clunky compared to what a terminal UI could offer.

Proposed solution

Add interactive UI primitives that Claude can use when presenting structured choices:

1. Multi-select / checkboxes — present a list of options, user toggles with arrow keys + space, confirms with Enter. Similar to inquirer.js checkbox prompt or gum choose --no-limit.

1. Single-select / radio — pick one from a list (already partially exists in permission prompts).

1. Action menu — present labeled actions (e.g., "Create tasks", "Start executing", "Skip") that the user can select.

Use case

After completing a task, the agent outputs a summary with "next steps" — a numbered table of follow-up actions with priorities. The agent then needs to ask: "Which of these should I create as tasks? Which should I start now? Or skip all?"

Currently this is a text prompt. With interactive elements, it would be a multi-select where the user checks items and picks an action — much faster and less error-prone.

Example interaction

👉 Next steps:

  [x] 1. 🔴 Fix auth token expiry handling
  [ ] 2. 🟡 Add integration test for webhook
  [x] 3. 🟢 Update support KB

  Action: [Create tasks] [Start executing] [Skip]
  
  > Press Space to toggle, Enter to confirm

Context

This would be especially valuable for agentic workflows where the agent regularly presents structured choices: task selection, file review, deployment options, etc. The terminal already supports rich interactions (as seen in permission prompts) — extending this to agent-initiated prompts would significantly improve UX.