In this journal article, which was published in open access in the John Wiley’s Journal of Software Testing, Verification and Reliability (STVR) on the 3rd October 2024, we address the critical challenge of ensuring the optimality of plans generated by AI planners in autonomous systems, especially those employed in safety-critical and resource-constrained applications such as space exploration, elderly assistance, and industrial control. While previous research has delved into verifying the soundness and completeness of AI planners, the verification of optimality remains a largely unexplored area due to the inherent complexity and scalability issues associated with determining optimal solutions. Our contribution is a novel framework called MorphinPlan that employs mutation-guided metamorphic testing to detect non-optimal plans. Metamorphic Testing (MT) is a technique that uses metamorphic relations (MRs) to generate follow-up test cases from initial test cases, thus alleviating the need for a test oracle. We extend this approach by integrating mutation testing to guide the selection of follow-up test cases, focusing on states within the state space that are most likely to reveal non-optimal plans.