The purpose of the study was to introduce and evaluate a high‐resolution diode array for patient‐specific quality assurance (PSQA) of CyberKnife brain stereotactic radiosurgery (SRS) and stereotactic radiotherapy (SRT). Thirty‐three intracranial plans were retrospectively delivered on the SRS MapCHECK using fixed cone, Iris, and multileaf collimator (MLC). The plans were selected to cover a range of sites from large tumor bed, single/multiple small brain metastases (METs) to trigeminal neuralgia. Fiducial tracking using the four fiducials embedded around the detector plane was used as image guidance. Results were analyzed before and after registration based on absolute dose gamma criterion of 1 mm distance‐to‐agreement and 0.5%–3% dose‐difference. Overall, the gamma passing rates (1 mm and 3% criterion) before registration for all the patients were above 90% for all three treatment modalities (96.8 ± 3.5%, the lowest passing rate of 90.4%), and were improved after registration (99.3 ± 1.5%). When tighter criteria (1 mm and 2%) were applied, the gamma passing rates after registration for all the cases dropped to 97.3 ± 3.2%. For trigeminal neuralgia cases, we applied 1 mm and 0.5% criterion and the passing rates dropped from 100 ± 0.0% to 98.5 ± 2.0%. The mean delivery time was 33.4 ± 11.7 min, 24.0 ± 4.9 min, and 17.1 ± 2.6 min for the fixed cone, Iris, and MLC, respectively. With superior gamma passing rates and reasonable quality assurance (QA) time, we believe the SRS MapCHECK could be a good option for routine PSQA for CyberKnife SRS/SRT.