ask-user-question_skill

Overview

This skill lets your agent present modal questions to the user and capture their responses via the UI. Use it whenever the agent needs clarification, preferences, or explicit confirmation before proceeding. The user cannot see any console or CLI output — this tool is the only way to show interactive prompts.

How this skill works

The skill opens a UI modal with one or more question objects that include a prompt, an optional short header, and optional selectable options. Options can include an "Other" choice to capture free-text input; multiSelect can be enabled to allow multiple selections. The tool returns a clear string indicating the user selection, typed response, or dismissal so the agent can branch logic accordingly.

When to use it

• Clarify ambiguous instructions before taking action
• Ask user preferences (naming, organization, formats, etc.)
• Confirm destructive or irreversible operations
• Get approval for sensitive actions (financial, messaging, deletion)
• Collect details required to continue a multi-step task

Best practices

• Keep questions short and specific; use a 12-character header to label context
• Provide 2–4 clear option choices and include an "Other" option for free-text when needed
• Use multiSelect only when the user genuinely needs to pick multiple items
• Always wait for and handle the tool response — do not assume a default
• Treat a dismissed modal as an explicit non-response and prompt again or abort safely

Example use cases

• Ask how to organize a Downloads folder: by type, date, or project
• Confirm deletion of duplicate files with options: Delete all, Review first, Cancel
• Request send confirmation for an email or message
• Ask for naming conventions when creating shared project folders
• Collect preferred output format and level of detail before generating reports