What is the Iron Law of this skill?

The Iron Law is 'No skill without a failing test first.' This ensures that every skill is grounded in a real-world problem and provides a measurable improvement in agent behavior.

What is 'hallucinated utility' in agent skills?

Hallucinated utility occurs when a skill provides generic, non-actionable advice that looks comprehensive but doesn't actually solve a specific failure mode or change the LLM's output in a meaningful way.

When should I avoid using this skill?

This skill is not intended for simple task execution that doesn't require persistent instructions, nor is it meant for general conversational responses that don't need a structured logic flow.

How does TDD apply to AI skill authoring?

It follows a five-step process: defining a user problem, writing a failing test case (Red phase), establishing acceptance criteria, authoring the content, and then verifying the skill against the test (Green phase).

Skill Authoring TDD

Name: Skill Authoring TDD
Author: JoelLewis

byJoelLewis

0•

开发者工具

Enforces a test-driven development approach to creating and refining Claude Code skills to ensure functional clarity and grounded advice.

This meta-skill provides a rigorous framework for developing high-quality Claude Code agent skills by implementing a strict Test-Driven Development (TDD) cycle. It prevents the creation of generic or 'hallucinated' utility by requiring a documented failing test case and specific acceptance criteria before any skill content is authored. This process ensures that every instruction added to a skill directly addresses a verified failure mode, making it an essential tool for developers and prompt engineers who need to build production-ready agent instructions that consistently solve specific user problems.

主要功能

010 GitHub stars

02Integrated diagnostic checklists for quality assurance

03Defines clear acceptance criteria for skill validation

04Strict TDD workflow for agent skill development

05Prevents 'hallucinated utility' through pre-authoring test cases

06Provides a structured state machine for step-by-step authoring

使用场景

01Developing new agent skills from scratch with specific performance goals

02Refactoring existing skills that produce poor-quality or generic outputs

03Ensuring consistency and rigor across a library of domain-specific agent skills

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add joellewis/skill-library writing-skills

For use in Claude.ai and ChatGPT

Download Skill