Monitoring intracranial pressure (ICP) dynamics is critical for the management of traumatic brain injury, stroke, other neurosurgical conditions, and cerebral blood flow autoregulation; however, invasive ICP monitoring carries risks such as infection, hemorrhage, and sensor zero drift. Increasing evidence suggests that ICP waveform morphology provides clinically relevant information beyond mean ICP value alone. In this first-in-human prospective comparative clinical study, we evaluated the feasibility and accuracy of a novel, fully passive, non-invasive ICP pulse waveform monitoring system (Archimedes 02) based on the detection of eyeball mechanical movement. Fifteen intensive care unit patients (6 males, 9 females; mean age 57.1 ± 18.8 years) with clinically indicated invasive ICP monitoring or external ventricular drainage were enrolled. Three-minute monitoring sessions were performed to simultaneously acquire non-invasive ICP pulse waveforms, invasive ICP waveforms, and invasive radial artery blood pressure (ABP) waveforms. Averaged waveforms were derived for each patient and compared graphically and using correlation analysis. Non-invasive ICP pulse waves recorded with Archimedes 02 showed a strong correlation with invasive ICP waveforms ($\overline{\mathrm{R}}$ = 0.965). In contrast, correlations between non-invasive ICP and ABP waveforms ($\overline{\mathrm{R}}$ = 0.699), as well as between invasive ICP and ABP waveforms ($\overline{\mathrm{R}}$ = 0.749), were lower. These findings indicate that the non-invasive signal primarily reflects ICP dynamics rather than arterial blood pressure. This novel non-invasive ICP monitoring approach has the potential to enhance neurocritical care, particularly in settings where invasive monitoring is impractical or unavailable. Further validation in larger and more diverse patient populations is warranted.