Skill: Ask Questions If Underspecified

SkillMedia

Use this skill when scope ambiguity blocks progress in any SOP step (PRD, TDD, design, task decomposition). Produces a tight set of clarifying questions (max 5) that, once answered, unblock the next step. Avoids the failure mode where Agents proceed on assumption and produce wrong work. Chinese trig

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the Skill: Ask Questions If Underspecified skill

What this skill tells your AI

The instructions your AI receives, as published by charliehzm/medharness in .claude/skills/ask-questions-if-underspecified/SKILL.md and read by ahel’s review.

v2.88 Key Changes (MODEL-AGNOSTIC)

  • Model-agnostic: Uses model configured in ~/.claude/settings.json or CLI/env vars
  • No flags required: Works with the configured default model
  • Flexible: Model-agnostic — runs on whatever model the session runs
  • Settings-driven: Model selection via ANTHROPIC_DEFAULT_*_MODEL env vars

ultrathink - Take a deep breath. We're not here to write code. We're here to make a dent in the universe.

The Vision

Clarity is the foundation of inevitable solutions. Every question should narrow the path to truth.

Your Work, Step by Step

  1. Analyze requirements: Identify missing inputs and ambiguities.
  2. Separate MUST vs NICE: Block on essentials, assume the rest.
  3. Ask precisely: Short, concrete questions with defaults.
  4. Confirm understanding: Summarize before proceeding.

Ultrathink Principles in Practice

  • Think Different: Question hidden assumptions.
  • Obsess Over Details: Align questions with real constraints.
  • Plan Like Da Vinci: Build the question set before asking.
  • Craft, Don't Code: Precision over volume.
  • Iterate Relentlessly: Refine questions as context evolves.
  • Simplify Ruthlessly: Ask only what matters.

Purpose

Ensure task clarity BEFORE implementation by identifying ambiguities.

When to Use

  • ANY new task or feature request
  • Complex modifications
  • Unclear requirements

Process

1. Analyze Requirements

Identify:

  • Missing technical details
  • Unclear scope boundaries
  • Ambiguous terminology
  • Unstated assumptions

2. Categorize Questions

MUST_HAVE (Blocking)

Questions that BLOCK implementation until answered:

  • Critical architecture decisions
  • Security requirements
  • Data model choices
  • Integration points
NICE_TO_HAVE (Assumptions)

Questions where you can make reasonable assumptions:

  • UI/UX preferences
  • Performance targets
  • Edge case handling

3. Output Format

## 🔍 Clarification Needed

### MUST_HAVE (Please answer before I proceed):
1. [Critical question 1]
2. [Critical question 2]

### NICE_TO_HAVE (I'll assume these if not specified):
- [Optional question] → I'll assume: [default value]
- [Optional question] → I'll assume: [default value]

### My Understanding:
[Summarize what you understand so far]

4. Wait for Answers

DO NOT proceed with implementation until MUST_HAVE questions are answered.

Examples

Good Clarification

MUST_HAVE:
1. Should auth support both email/password AND OAuth providers?
2. What's the session timeout requirement?

NICE_TO_HAVE:
- Rate limiting? → I'll assume: 100 req/min
- Password complexity? → I'll assume: min 8 chars, 1 number, 1 special

Bad (Too Vague)

What do you want?
Can you give more details?

Signals

GitHub stars
86
Forks
8
Last commit
Jun 2026
Advanced
Catalog kind
skill
Gateway key
ask-questions-if-underspecified
Source
github.com/charliehzm/medharness