Test doubles enable developers to isolate software components, simulate complex scenarios, and validate interactions during testing. Despite their critical role, there is a lack of systematically created and structured datasets for studying, devising, and evaluating techniques that aim to automate various aspects of the creation, usage, and maintenance of test doubles. To address this limitation, this paper introduces a novel dataset of 1,070 projects including test doubles. The dataset also includes carefully extracted metadata describing the use of the test doubles in the projects. Finally, the paper discusses the potential applications of the dataset in future research and directions for extending the dataset.