A bidirectional， multi-level threat-assessment framework is proposed for engagements against mixed swarms of aerodynamic and ballistic targets， addressing the inability of current methods to reconcile target-type heterogeneity and to quantify swarm scale and cooperative effects. The framework establishes an index system that integrates adversarial threat attributes， key-area defense effectiveness indices， and global correction factors. Intrinsically static attributes are evaluated via Pythagorean fuzzy entropy combined with TOPSIS， while dynamic attributes are fused through grey relational analysis and TOPSIS. A global correction factor explicitly quantifies swarm scale， cooperative gains， critical target types， and higher-echelon intelligence to recalibrate threat attributes. The relative relationship between adversarial threat attributes and defense effectiveness indices finally yields an integrated threat value for each target. Simulations demonstrate marked improvements in accuracy and operational utility for mixed-swarm threat assessment， providing a reliable basis for optimal resource allocation and efficient interception decision-making.