Causal Safety Engine is a safety-first validation layer for AI agents. It analyzes agent actions using causal signals (not just correlations) to detect unsafe, non-identifiable, or unstable decisions before execution. Designed for high-risk and autonomous systems, it favors “causal silence” over false positives and integrates with existing AI pipelines as a governance and safety control.