Vague: Requirement Clarification

Transform vague requirements into precise, actionable specifications through hypothesis-driven questioning.

When to Use

• Ambiguous feature requests ("add a login feature")
• Incomplete bug reports ("the export is broken")
• Underspecified tasks ("make the app faster")

For strategy/planning blind spot analysis, use unknown. For content-vs-form reframing, use metamedium.

Core Principle: Hypotheses as Options

Use AskUserQuestion for clarification. Do not ask open-ended free-text questions when options can represent plausible hypotheses.

• Bad: What kind of login do you want?
• Good: OAuth / Email+Password / SSO / Magic link

Each option should be a testable interpretation of user intent.

Invocation

Parse input:

• REQUIREMENT: requirement text (required)
• --max-iterations N: max rounds (default: 3)

Initialization

1. Create .claude/clarify-vague.local.md:

---
active: true
iteration: 1
max_iterations: [MAX_ITERATIONS]
original_requirement: "[REQUIREMENT]"
started_at: "[ISO timestamp]"
---

## Original Requirement
"[REQUIREMENT]"

## Clarification Progress
(Decisions are appended as the loop progresses)

2. Confirm activation:

Vague clarification loop activated!

Original Requirement: "[REQUIREMENT]"
Max Iterations: [MAX_ITERATIONS]

To cancel: /cancel

Protocol

Phase 1: Diagnose Ambiguity

Capture unknowns across categories:

• Scope
• Behavior
• Data / Interface
• Constraints / Priority
• Success criteria

Phase 2: Iterative Clarification

Use AskUserQuestion to resolve ambiguities.

Rules:

• Ask exactly 4 questions per round
• Use exactly 4 options per question
• Last question's last option must be: Clarification complete - proceed with current understanding
• Batch all 4 questions in a single AskUserQuestion call
• Cap total rounds by --max-iterations

Phase 3: Completion Check

Stop when either condition is met:

• User selects clarification-complete option
• Max iterations reached

Phase 4: Before/After Summary

Output:

## Requirement Clarification Summary

### Before (Original)
"{original requirement verbatim}"

### After (Clarified)
**Goal**: [precise description]
**Reason**: [why this matters]
**Scope**: [included and excluded]
**Constraints**: [limitations and preferences]
**Success Criteria**: [how completion is verified]

**Decisions Made**:
| Question | Decision |
|----------|----------|
| [ambiguity 1] | [selected option] |

Then output:

<promise>CLARIFICATION COMPLETE</promise>

Only output the promise when clarification is genuinely complete.

Question Design Rules

1. Hypothesis options over open questions
2. Neutral framing (avoid leading choices)
3. One ambiguity axis per question
4. Preserve user intent (refine, do not redirect)
5. Keep language concrete and implementation-relevant

Now begin by creating the state file, identifying four critical ambiguities, and asking the first AskUserQuestion batch.