To address the limitations of existing kill-path recommendation methods in multi-attribute trade-off, F2T2EA process constraints, real-time solution, and network resilience evaluation, an optimal kill-net path solution method based on improved multi-attribute utility theory (MAUT) is proposed. A directed kill-net model is constructed according to the Find-Fix-Track-Target-Engage-Assess process, and an evaluation system including execution time, success probability, resource consumption, risk level, and path resilience is established. Normalized utilities, nonlinear preference functions, and resilience utility are introduced to characterize decision preferences under different scenarios. Depth-first search is used to enumerate candidate paths satisfying process constraints and reliability constraints, and their comprehensive utilities are evaluated one by one. Meanwhile, Top-K ranking pruning is adopted as an approximate acceleration strategy for large-scale scenarios. In an air-defense scenario, 72 feasible paths are generated, and the optimal utility under equal four-attribute weights is 0.7734. After resilience utility is introduced, the optimal path changes, and disturbance substitution capability increases from 0.7531 to 0.7545. Node and link disturbance experiments show that C2_Center_7 is a key single-point bottleneck. The estimation error of success probability is 0.00042 with 10,000 Monte Carlo simulations. Results show that the proposed method supports multi-attribute path selection and provides quantitative evidence for identifying weak links and configuring redundant links.