概要
Safety Validation provides a critical security layer for AI-driven development by enforcing boundaries defined in .agentignore files. It automatically scans modified paths to ensure that sensitive configurations, private credentials, or core infrastructure files remain untouched by the AI agent. By halting execution upon detection of forbidden path access and maintaining a detailed audit trail in ACTION_LOG.md, it ensures reliable, safe, and compliant autonomous coding workflows.