Optical coherence tomography (OCT) is applicable to the study of cerebral microvasculature in vivo. Optimised acquisition schemes enable the generation of three-dimensional OCT angiograms, i.e., volumetric images of red blood cell flux in capillary networks, currently at a repetition rate of up to 1/10 seconds. This makes testable a new class of hypotheses that strive to bridge the gap between microscopic phenomena occurring at the spatial scale of neurons, and less invasive but crude techniques to measure macroscopic blood flow dynamics. Here we present a method for quantifying the occurrence of transient capillary stalls in OCT angiograms, i.e., events during which blood flow through a capillary branch is temporarily occluded. By making the assumption that information on such events is present predominantly in the imaging plane, we implemented a pipeline that automatically segments a network of interconnected capillaries from the maximum intensity projections (MIP) of a series of 3D angiograms. We then developed tools enabling rapid manual assessment of the binary flow status (open/stalled) of hundreds of capillary segments based on the intensity profile of each segment across time. The entire pipeline is optimized to run on a standard laptop computer, requiring no high-performance, low-availability resources, despite very large data volumes. To further reduce the threshold of adoption, and ultimately to support the development of reproducible research methods in the young field, we provide the documented code for scrutiny and re-use under a permissive open-source license.