概要
The Confidence Check skill is a proactive validation framework for Claude Code designed to eliminate 'wrong-direction' execution and technical debt. By enforcing a strict 90% confidence threshold across five critical dimensions—codebase duplication, architectural alignment, official documentation verification, open-source references, and root cause analysis—it ensures the AI agent only proceeds when a task is fully understood. This 'measure twice, cut once' approach significantly reduces token waste and prevents the introduction of redundant functions, ensuring all code remains compliant with your project's specific tech stack and standards.
