What problem does it solve?
Agents often ask unclear, poorly timed, or overwhelming questions that waste user time, lead to incorrect assumptions, and derail development workflows. This skill standardizes interactive questioning to ensure agents collect the right information efficiently without frustrating users.
Core Features & Use Cases
- Runtime Tool Selection: Guides agents to use the correct runtime-specific input tool (e.g., Claude's ask_followup_question, Codex's request_user_input) or fall back to plain chat for interactive sessions.
- Batching & Serial Logic: Groups independent questions to reduce round-trips, and enforces one-question-at-a-time for branching decisions that depend on prior answers.
- Quality Guardrails: Prevents asking discoverable questions, eliminates low-impact implementation detail queries, and surfaces agent-owned assumptions at approval gates.
- Use Case: When a planning skill needs to clarify feature requirements, use this skill to batch independent questions about validation rules and optional fields, then ask serial approval questions one at a time to avoid overwhelming the user.
Quick Start
Use the ask-questions skill to collect the missing requirements for the new user onboarding flow from the user.
