概要
The Verification Gate skill acts as a cognitive safety mechanism for Claude Code by mandating rigorous verification before any system-modifying action is taken. By implementing a four-stage checkpoint system—covering hypothesis formation, evidence gathering, system alignment, and pattern-matching detection—it prevents the AI from executing 'hallucinated' or generic solutions that do not match the specific project context. This skill is essential for complex debugging and refactoring where the obvious solution might not be the correct one for the specific architecture or dependency system currently in use.