ask-me-questions
Systematic requirements discovery through multi-round questioning before taking action. Use when the user explicitly asks — "ask me questions", "ask me first", "interview me", "gather requirements", "
SKILL.mdskills/ask-me-questions/SKILL.md
GitHubAsk Me Questions
Understand before you act. Ask targeted questions in focused multi-round conversations before proceeding with any task.
Asking mechanism (portability). Where your harness has a structured question tool (e.g.
Claude Code's AskUserQuestion), use it — the multiple-choice form gets far better answers
than open prose. Where it doesn't, ask in plain text: numbered questions, each with 2-4
labelled options and a recommended default. Everything below applies either way.
Rules
- Never jump to solutions. When this skill triggers, questioning comes FIRST.
- Never ask what you can verify. Read the code, run the command, check the docs first. Ask only about intent, preferences, priorities, and trade-offs the code can't reveal.
- Offer options, not open prose. Every question comes with concrete choices the user can pick from, plus your recommendation — never a bare "what do you want?".
- Max 4 questions per round. Make each one count.
- Adapt rounds to complexity:
- Simple (single-file change, config tweak): 1 round (2-4 questions)
- Medium (feature, multi-file change): 2 rounds (4-8 questions)
- Complex (architecture, new system): 2-3 rounds (6-12 questions)
- Stop questioning when: you have enough clarity to act confidently. Don't over-ask.
- Unattended runs don't block. If running autonomously (background, cron, headless) where no one can answer, skip questioning: state your assumptions explicitly and proceed.
Questioning Workflow
Round 1: Scope & Intent
Establish what the user actually wants and why. These questions cut through ambiguity fast.
Focus areas:
- What: What specific outcome do they want? What does "done" look like?
- Why: What problem does this solve? What triggered this request?
- Constraints: Are there hard requirements (tech stack, timeline, compatibility)?
- Scope: What's explicitly OUT of scope?
Question design tips:
- Lead with the most disambiguating question — the one whose answer changes everything
- Use options with descriptions to surface hidden assumptions
- Put the recommended/most-common option first
- Use
multiSelect: truewhen choices aren't mutually exclusive
Example:
question: "What should happen when the export fails midway?"
header: "Errors"
options:
- label: "Retry from checkpoint"
description: "Resume where it left off. Requires tracking progress state."
- label: "Retry from scratch"
description: "Simple but may be slow for large exports."
- label: "Fail with partial output"
description: "Return whatever was exported so far."
Round 2+: Targeted Follow-ups
Based on Round 1 answers, ask about:
- Edge cases revealed by their choices
- Trade-offs between approaches they've implied
- Specifics that their answers left open (e.g., they said "fast" — how fast?)
- Pattern conflicts — follow existing codebase patterns by default without asking. Only ask when the codebase shows two conflicting patterns AND the choice materially affects the outcome.
Skip Round 2 if Round 1 answers were clear and complete.
After Questioning: Proceed
Once you have sufficient clarity:
- Briefly summarize what you understood (2-3 sentences in plain text, NOT a file)
- Proceed directly with the task — no permission-asking, no "shall I proceed?"
Writing Good Questions
Option Design
Each option should represent a genuinely different path, not slight variations:
Good options (meaningfully different outcomes):
- "Server-side validation" vs "Client-side validation" vs "Both"
- "Add to existing table" vs "Create new table" vs "Use a JSON field"
Bad options (cosmetic differences):
- "Use camelCase" vs "Use snake_case" — too trivial for a question
- "Option A" vs "Option B" — labels must be self-explanatory
Header Tips
Headers appear as chips/tags. Keep under 12 chars:
- Good:
"Scope","Auth","Data model","Error flow","API style" - Bad:
"Authentication method","How should we handle errors"
When to Use multiSelect
Use multiSelect: true for:
- Feature selection ("Which of these should we support?")
- Capability flags ("What constraints apply?")
- Inclusive lists ("Which environments need this?")
Keep multiSelect: false (default) for:
- Mutually exclusive approaches
- Architecture decisions
- Single-choice trade-offs