Testing Skills FAQs

Question 1

How does the Red-Green-Refactor cycle work for documentation?

Accepted Answer

First, you watch the agent fail without the skill (Red). Then, you write minimal documentation to fix that specific failure (Green). Finally, you identify new excuses the AI makes and update the documentation to close those loopholes (Refactor).

Question 2

When is the best time to use this skill?

Accepted Answer

You should use this skill before deploying any new Claude Code skill that requires high discipline, such as TDD requirements, complex code review protocols, or compliance-heavy workflows where an AI might be tempted to cut corners.

Question 3

What are 'pressure scenarios' in skill testing?

Accepted Answer

Pressure scenarios are realistic test cases that include constraints like tight deadlines, sunk cost fallacies, or authority pressure. These tests verify if Claude will follow your skill's rules even when it seems easier or faster to ignore them.

Question 4

What is a rationalization table?

Accepted Answer

A rationalization table is a tool used within this skill to document the exact excuses or 'hallucinated logic' an AI uses to bypass instructions. By documenting these verbatim, you can create explicit negations to prevent future non-compliance.

Question 5

What exactly does the Testing Skills capability do?

Accepted Answer

It applies Test-Driven Development (TDD) principles to process documentation. By running 'pressure scenarios,' it forces Claude to reveal where documentation is weak, then uses a Red-Green-Refactor cycle to iterate on those instructions until they are bulletproof.

Testing Skills

Testing Skills

关于

主要功能

使用场景

关于

主要功能

使用场景