Ask Clarify Question

• Already have a clear bounded question.
• Need the final answer directly.
• Need repository analysis, implementation, or review.

Required Inputs

• RAW_TOPIC (string; required): Messy topic, rough thought, or problem statement draft.
• AUDIENCE (string; optional): Who the future answer or decision is for.
• CONSTRAINTS (list; optional; shape: {CONSTRAINT}): Known time, resource, domain, or evidence constraints.

Input Contract Notes

• RAW_TOPIC can be messy; the skill should reduce ambiguity instead of demanding a polished first draft.
• If the user does not provide explicit constraints, infer only the single most limiting constraint from the prompt and mark it as inferred.
• Ask what should be avoided before expanding ideal outcomes.

Structured Outputs

• FOG_KEYS (list; required; shape: {KEYWORD}): Three concrete anchor words that represent the core issue.
• NEGATIVE_SPACE (list; required; shape: {AVOID}): What the future question should avoid or not become.
• PRIMARY_CONSTRAINT (string; required): Single dominant constraint that most limits the question.
• PROBLEM_STATEMENT (string; required): One clear problem-definition sentence.
• NEXT_RECOMMENDED_SKILL (string; required): Smallest useful next question-design step.

Output Contract Notes

• FOG_KEYS should stay short and concrete, not abstract themes.
• PROBLEM_STATEMENT should express one bounded question direction, not a list of possible asks.
• NEXT_RECOMMENDED_SKILL should usually point to ask-break-it-down unless the question is already bounded enough to ask.

Primary Lens

• primary_lens: inversion-focus
• why: Question clarification should lower cognitive load, define the negative boundary first, and collapse ambiguity into one usable problem statement.

Artifacts

• artifacts_in: none
• artifacts_out: ask-clarify-question.v1

Neutrality Rules

• Do not answer the topic while clarifying the question.
• Separate what the user wants from what the user wants to avoid.
• If the prompt stays ambiguous, narrow to one problem statement instead of inventing extra goals.

Execution Constraints

• Reduce the thought to three anchors, one dominant constraint, and one problem statement.
• Prefer the shortest clear wording that helps the user take control of the question.
• Do not expand into solution advice or answer content.

Response Format

Think and operate in English, but deliver the final response in Korean. 쉽고 간결한 한국어로 답하라. 전문 용어 금지. 핵심만 간단하게.

Output the PROBLEM_STATEMENT as a single sentence up front.

Then show:

• Anchors: [word1], [word2], [word3]
• Avoid: [what the question should not become]
• Dominant constraint: [one sentence]

Follow immediately with the NEXT_RECOMMENDED_SKILL.

Mandatory Rules

• Always emit exactly one PROBLEM_STATEMENT.
• Keep FOG_KEYS to three items unless the prompt gives fewer usable anchors.