What is a pressure scenario in skill testing?

A pressure scenario is a test case that combines realistic constraints like tight deadlines, authority overrides, or sunk costs to see if the AI will break its own rules or rationalize a failure.

Why do I need to run a 'RED' phase for a skill?

The RED phase involves watching the agent fail without the skill active. This allows you to document the exact excuses and rationales the AI uses, so you can specifically counter them in the documentation.

How do I know if my Claude Code skill is 'bulletproof'?

A skill is considered bulletproof when the agent chooses the correct, disciplined path even under maximum simulated pressure and explicitly cites skill sections to justify its refusal to take shortcuts.

Does this require knowledge of Test-Driven Development (TDD)?

Yes, this skill is designed as an extension of the TDD mindset. It requires an understanding of the RED-GREEN-REFACTOR cycle, specifically the 'superpowers:test-driven-development' skill.

Skill Testing with Subagents

Name: Skill Testing with Subagents
Author: Krosebrook

byKrosebrook

•

Seguridad y Pruebas

Implements a Test-Driven Development cycle for Claude Code skills to ensure AI agents follow process documentation even under high-pressure scenarios.

Acerca de

This skill applies the RED-GREEN-REFACTOR methodology to the creation and refinement of AI behavioral documentation. By forcing agents through simulated pressure scenarios—where they are tempted to bypass rules due to time constraints, sunk costs, or perceived pragmatic benefits—users can identify and plug logical loopholes and rationalizations. It ensures that your custom skills move beyond academic suggestions to become bulletproof instructions that Claude will follow consistently in real-world development environments.

Características Principales

Multi-pressure scenario generation (Time, Authority, Sunk Cost)
Rationalization capture for identifying AI loopholes
RED-GREEN-REFACTOR cycle for process documentation
1 GitHub stars
Meta-testing framework to identify documentation ambiguity
Verification checklists for bulletproofing AI behavior

Casos de Uso

Hardening TDD skills to prevent agents from skipping tests during deadlines
Testing compliance documentation for security-critical deployment workflows
Refining internal team standards to ensure high-fidelity AI adherence

Acerca de

Características Principales

Multi-pressure scenario generation (Time, Authority, Sunk Cost)
Rationalization capture for identifying AI loopholes
RED-GREEN-REFACTOR cycle for process documentation
1 GitHub stars
Meta-testing framework to identify documentation ambiguity
Verification checklists for bulletproofing AI behavior

Casos de Uso

Hardening TDD skills to prevent agents from skipping tests during deadlines
Testing compliance documentation for security-critical deployment workflows
Refining internal team standards to ensure high-fidelity AI adherence