[FEATURE] Add `confirm` prompt type to AskUserQuestion (Tab-to-amend support)

Resolved 💬 2 comments Opened Feb 2, 2026 by trenaryja Closed Mar 3, 2026

Preflight Checklist

  • [x] I have searched existing requests and this feature hasn't been requested yet
  • [x] This is a single feature request (not multiple features)

Problem Statement

Claude Code's native permission prompts support a powerful interaction pattern:

  • Arrow keys to highlight an option
  • Tab to add optional context ("Tab to amend")
  • Enter to submit
  • Returns both selection AND typed context

This "Tab to amend" pattern is not available to skills using AskUserQuestion. While AskUserQuestion supports several prompt variations (single select, multi-select, multiple questions—see Additional Context), none allow users to both select an option AND add contextual notes in a single interaction. The "Other" option is always separate.

Proposed Solution

Add a type parameter to AskUserQuestion:

AskUserQuestion({
  type: "confirm",  // "select" (default) | "checkbox" | "confirm"
  question: "Delete this file?",
  header: "1/34",
  options: [
    { label: "← Delete", description: "Remove this file" },
    { label: "→ Keep", description: "Still relevant" }
  ]
})

Prompt Types

| Type | Behavior | Status |
|------|----------|--------|
| select | Numbered list, single choice, separate "Other" option | ✅ Current default |
| checkbox | Numbered list, spacebar multi-select | ✅ Exists (via multiSelect) |
| confirm | Inline options, Tab to add context, single choice | ❌ Missing |

Response Shape for confirm

{
  selected: "← Delete",
  context: "outdated, already completed"  // Optional, from Tab-to-amend
}

Alternative Solutions

Workaround 1: Use "Other" for notes
Users can type decisions + notes in the "Other" field (e.g., "delete - it's outdated"), but this loses the structured selection and requires parsing free text.

Workaround 2: Two-step flow
Ask for decision first, then ask "Any notes?" as follow-up. This doubles the interactions and breaks rapid workflow patterns.

Workaround 3: Build custom CLI
Ship a separate CLI binary with the skill that uses @inquirer/prompts. But this breaks Claude Code's UX—context doesn't flow naturally and two-way interaction is clunky.

Priority

Medium - Would be very helpful

Feature Category

Interactive mode (TUI)

Use Case Example

Building a /tinder skill for rapid swipe-style decisions (file cleanup, task triage, review queues):

  1. Skill reads 69 plan files from ~/.claude/plans/
  2. For each file, present a preview and ask Delete/Keep
  3. User arrows to "Keep", presses Tab, types "still working on this", presses Enter
  4. Repeat rapidly through all files
  5. Skill summarizes: "12 to delete, 22 to keep (3 with notes). Want me to..."
  6. User confirms, skill executes

The current select type adds cognitive overhead—scanning numbered lists and selecting a separate "Other" option to add notes breaks the flow. The native permission prompt UX ("Tab to amend") is exactly what this workflow needs.

Additional Context

Inspiration

This sorta mirrors @inquirer/prompts, the Node.js CLI prompt library:

| Inquirer | AskUserQuestion Equivalent |
|----------|---------------------------|
| select() | type: "select" ✅ exists |
| checkbox() | type: "checkbox" ✅ exists (via multiSelect) |
| confirm() | type: "confirm"missing |

Screenshots

Native permission prompt (desired UX) - Note "Tab to amend" at the bottom:

<img width="965" height="376" alt="Image" src="https://github.com/user-attachments/assets/84a6e2a4-2acc-4573-b0d9-e73f0bc41361" />

Current AskUserQuestion variations - None support Tab-to-amend:

<img width="1055" height="346" alt="Image" src="https://github.com/user-attachments/assets/00d371b9-c122-4d2a-b9fe-a0f5cde35185" />

<img width="1055" height="515" alt="Image" src="https://github.com/user-attachments/assets/ce693fa2-48e3-445f-817a-6a2fdb1a7e96" />

<img width="1055" height="625" alt="Image" src="https://github.com/user-attachments/assets/b59bcee9-c2a7-46a3-9080-554886262ff6" />

<img width="1055" height="660" alt="Image" src="https://github.com/user-attachments/assets/41f3074b-4728-401f-a96a-89895303b9d7" />

<img width="1055" height="680" alt="Image" src="https://github.com/user-attachments/assets/824a4acc-df57-4e39-904b-bb35851dcc9c" />

<img width="1055" height="680" alt="Image" src="https://github.com/user-attachments/assets/f78a03a6-7392-481d-9298-32560e9a9757" />

View original on GitHub ↗

This issue has 2 comments on GitHub. Read the full discussion on GitHub ↗