ask-me-questions

Gathers requirements through structured multi-round questioning before executing ambiguous tasks.

Updated Jun 15, 2026
One-click install
npx skills add https://github.com/nsollazzo/ship-kit --skill ask-me-questions-nsollazzo
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: ask-me-questions
Source: https://github.com/nsollazzo/ship-kit/tree/main/.agents/skills/ask-me-questions
Command: npx skills add https://github.com/nsollazzo/ship-kit --skill ask-me-questions-nsollazzo

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Ambiguous requests lead agents to guess wrong and build the wrong thing. This Skill forces a structured discovery phase — targeted questions with concrete options — before any implementation begins, so the agent acts on confirmed intent rather than assumptions. ## Core Features & Use Cases - Multi-Round Questioning: Runs 1-3 rounds of focused questions (max 4 per round) scaled to task complexity, from simple config tweaks to architecture decisions. - Option-Based Questions: Every question ships with 2-4 labelled choices and a recommended default, using structured question tools where the harness supports them and plain text otherwise. - Verification-First Rule: Never asks what can be verified from code, docs, or commands — questions target only intent, preferences, and trade-offs. - Use Case: A user says "add export functionality" without specifying format, error handling, or scope. The Skill asks what happens on mid-export failure, which formats to support, and what is out of scope, then proceeds with the confirmed design. ## Quick Start Ask me questions to clarify the requirements before you start implementing the export feature.

Frequently Asked Questions about ask-me-questions

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I make an AI agent ask clarifying questions before coding?▼

Trigger the skill with phrases like "ask me questions", "interview me", or "gather requirements". The agent then runs structured rounds of questions with concrete options and recommendations before writing any code.

How many questions should an agent ask before starting a task?▼

The skill caps each round at 4 questions and scales rounds by complexity: 1 round for simple changes, 2 rounds for multi-file features, and 2-3 rounds for architecture work. Questioning stops once there is enough clarity to act.

Does this work in agents without a structured question tool?▼

Yes. Where the harness lacks a tool like Claude Code's AskUserQuestion, the skill falls back to plain-text numbered questions, each with 2-4 labelled options and a recommended default.

When should I not use requirements questioning?▼

Skip it for clear requests regardless of size, for anything verifiable from code or docs, and for unattended runs. In autonomous mode the skill states assumptions explicitly and proceeds instead of blocking.

What makes a good clarifying question for an AI agent?▼

Good questions offer meaningfully different paths, not cosmetic variations — for example "server-side vs client-side validation" rather than naming conventions. Each option needs a self-explanatory label and a short description of its trade-off.