AI Skill TDD Framework

概要

This skill provides a rigorous framework for developing and validating AI instructions by applying Test-Driven Development (TDD) principles directly to process documentation. By simulating realistic 'pressure scenarios'—where an agent might be tempted to cut corners due to time constraints, sunk costs, or authority—it helps developers identify loopholes in their Claude Code skills. Users can systematically document agent failures, write targeted skill improvements, and iterate until instructions are bulletproof and compliance is guaranteed even in high-stakes environments.

主な機能

• Red-Green-Refactor cycle adapted for process documentation
• Verbatim rationalization tracking to identify common AI excuses
• Skill bulletproofing via explicit negation and rationalization tables
• Meta-testing protocols to verify instruction clarity and agent compliance
• 0 GitHub stars
• Pressure scenario generation using time, authority, and economic constraints

ユースケース

• Verifying that high-stakes compliance skills are followed during emergencies
• Refining internal team guidelines to prevent 'spirit vs letter' rationalizations
• Hardening TDD requirements for agents prone to skipping tests