Claude Skill Authoring & TDD

关于

This skill provides a rigorous Red-Green-Refactor framework for developing specialized Claude Code capabilities. By treating process documentation like production code, it mandates that every skill must be preceded by a documented baseline failure using pressure-tested subagent scenarios. This methodology ensures that skills aren't just theoretical guidance, but are battle-tested against common agent rationalizations like time pressure, sunk cost bias, and authority influence, resulting in highly resilient and predictable AI behavior.

主要功能

• Provides a standardized XML structure for skill interoperability
• 13 GitHub stars
• Identifies and closes common agent rationalizations and excuses
• Optimizes skill token efficiency and searchability via metadata rules
• Pressure-tests AI behavior using subagent simulation scenarios
• Enforces a strict Red-Green-Refactor cycle for documentation

使用场景

• Hardening existing skills that agents are currently bypassing or ignoring
• Standardizing team-wide technical documentation and architectural patterns
• Creating new, domain-specific Claude Code skills for specialized workflows