How do I control the depth of the clarification loop?

You can use the --max-iterations flag followed by a number (e.g., --max-iterations 5) to set how many rounds of questioning you want to go through.

Why does it ask exactly four questions per round?

The four-question structure is designed to probe different ambiguity categories—like interface, behavior, and data—simultaneously to reach a comprehensive understanding faster.

Where does the skill store my requirements?

The skill creates and maintains a local state file at .claude/clarify.local.md within your repository to track progress and record decisions.

What happens when the clarification process is finished?

Once complete, the skill generates a Requirement Clarification Summary detailing the refined goal, scope, constraints, and success criteria, along with a log of all decisions made.

Requirements Clarifier

Name: Requirements Clarifier
Author: zettalyst

byzettalyst

0•

生产力与工作流

Transforms vague requirements into precise, actionable technical specifications through structured, iterative questioning.

The Clarify skill bridges the gap between ambiguous user requests and production-ready code by initiating a systematic clarification loop. By prompting Claude to identify gaps in scope, behavior, and constraints, this skill ensures that project requirements are fully understood before development begins. It uses a structured 4x4 questioning method—four questions with four distinct options each—to rapidly converge on a final specification, resulting in a comprehensive summary that includes project goals, success criteria, and a logged history of architectural decisions.

主要功能

01Multi-category analysis covering scope, behavior, data, and constraints

02Customizable iteration limits for varying levels of project complexity

03Local state management via .claude/clarify.local.md files

04Iterative questioning loops to resolve ambiguity

05Standardized requirement summary generation with before/after comparisons

060 GitHub stars

使用场景

01Transforming high-level feature ideas into detailed technical specs

02Aligning technical constraints and data formats before starting a new API design

03Defining edge cases and success criteria for complex bug fixes

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add zettalyst/claude-plugins clarify

For use in Claude.ai and ChatGPT

Download Skill