Redundant test cases, often covering overlapping functional requirements, can significantly increase overall testing time and resource usage. To address this challenging optimization problem effectively, we introduce a novel Two-Phase Adaptive Hybrid SCA-PSO (AH-SCA-PSO) framework specifically aimed at enhancing and improving multi-task test redundancy reduction. The first phase employs the Sine Cosine Algorithm to ensure broad exploration and maintain solution diversity, while the second phase applies Particle Swarm Optimization to fine-tune solutions through adaptive local search. A task-aware knowledge sharing strategy is embedded in both phases to boost convergence. Experimental results in benchmark datasets show that AH-SCAPSO outperforms standalone algorithms and the Hybrid-GNA-SA method, achieving better redundancy reduction with low computational cost. Additionally, the framework exhibits scalability across a range of dataset sizes, which makes it useful in real world software testing settings where effectiveness is crucial. These results highlight the robustness and suitability of the framework for complex software testing tasks and optimization.