The rapid expansion of artificial intelligence and the internet of things demands an unprecedentedly high level of efficiency in edge computing. Processing-in-memory sensors, based on memristors, play a pivotal role in addressing these challenges, facilitating real-time decision-making, data optimization, and energy efficiency. Despite widespread interest, there's a dearth of innovation in memristor materials. Metal halide perovskites offer distinct advantages over mainstream memristive materials due to their lower formation energy and strong interaction with light, positioning them as an integrated optoelectronic platform for in-memory computing sensors. In this work, we review metal halide perovskite memristors emphasizing materials, devices, and applications. We discuss their mixed ionic and electronic properties and optoelectronically induced conduction channel formation. Furthermore, we examine device structures, memristive mechanisms, and synapse-like behaviors, together with an exploration of the potential for edge computing applications. Finally, we present our vision for their future development from both hardware and software perspectives.