What is the EARS syntax used in this skill?

EARS (Easy Approach to Requirements Syntax) provides a set of patterns (Universal, Event-driven, State-driven, Unwanted Behavior, Optional) to ensure requirements are clear and structurally consistent.

What happens if the generated requirements are incomplete?

The skill enters a loop where it generates a critique and may ask up to 20 'mean questions' to extract missing details from the user before finalizing the document.

Does this skill help with testing?

Yes, every requirement generated includes a Given-When-Then (GWT) acceptance test, making it easy to translate the documentation into automated test suites.

How does the automated scoring work?

The skill runs a Python script that evaluates your requirements against a rubric (C.U.T.E.), checking for ambiguity, testability, and adherence to the template, outputting a score out of 100.

Perfect Requirements Generator (sdd-req100)

Name: Perfect Requirements Generator (sdd-req100)
Author: taiyousan15

bytaiyousan15

0•

学習とドキュメンテーション

Generates EARS-compliant, testable software requirements with automated scoring and iterative refinement loops.

The sdd-req100 skill is a specialized agentic workflow designed to eliminate ambiguity in software documentation by producing high-quality requirements.md files. It adheres strictly to the EARS (Easy Approach to Requirements Syntax) patterns and incorporates Given-When-Then (GWT) acceptance tests for every requirement. The skill features a built-in scoring engine (C.U.T.E.) that provides a quantitative quality score; if the score falls below the 98% target, the agent automatically enters a critique-and-refine loop, asking deep-dive questions to uncover hidden constraints and ensuring every requirement is precise, testable, and actionable.

主な機能

01Automated C.U.T.E. scoring system with score.json output.

02Edge-case discovery via 20 targeted 'mean questions' for developers.

03Standardized Given-When-Then (GWT) acceptance criteria for all requirements.

04Iterative improvement loops that refine docs based on automated critiques.

05EARS-compliant requirements generation for zero ambiguity.

060 GitHub stars

ユースケース

01Drafting high-fidelity technical specifications for complex system features.

02Auditing existing project documentation to identify and fix ambiguous language.

03Transforming vague project ideas into structured, machine-scorable requirements.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add taiyousan15/taisun_agent sdd-req100

For use in Claude.ai and ChatGPT

主な機能

01Automated C.U.T.E. scoring system with score.json output.

02Edge-case discovery via 20 targeted 'mean questions' for developers.

03Standardized Given-When-Then (GWT) acceptance criteria for all requirements.

04Iterative improvement loops that refine docs based on automated critiques.

05EARS-compliant requirements generation for zero ambiguity.

060 GitHub stars

ユースケース

01Drafting high-fidelity technical specifications for complex system features.

02Auditing existing project documentation to identify and fix ambiguous language.

03Transforming vague project ideas into structured, machine-scorable requirements.