Detecting heavy atoms in the inflated atmospheres of giant exoplanets that orbit close to their parent stars is a key factor for understanding their bulk composition, their evolution, and the processes that drive their expansion and interaction with the impinging stellar wind. Unfortunately, very few detections have been made thus far. Here, we use archive data obtained with the Cosmic Origins Spectrograph onboard the Hubble Space Telescope to report an absorption of ∼6.4%±1.8% by neutral oxygen during the HD 189733b transit. Using published results from a simple hydrodynamic model of HD 189733b, and assuming a mean temperature of ∼(8−12) × 10 3 K for the upper atmosphere of the exoplanet, a mean vertical integrated O I density column of ∼8 × 10 15 cm −2 produces only a 3.5% attenuation transit. Much like the case of the hot-Jupiter HD 209458b, super-solar abundances and/or super-thermal broadening of the absorption lines are required to fit the deep transit drop-off observed in most far-ultraviolet lines. We also report evidence of short-time variability in the measured stellar flux, a variability that we analyze using time series derived from the time-tagged exposures, which we then compare to solar flaring activity. In that frame, we find that non-statistical uncertainties in the measured fluxes are not negligible, which calls for caution when reporting transit absorptions. Despite cumulative uncertainties that originate from variability in the stellar and sky background signals and in the instrument response, we also show a possible detection for both a transit and early-ingress absorption in the ion C II 133.5 nm lines. If confirmed, this would be the second exoplanet for which an early ingress absorption is reported. In contrast, such an early ingress signature is not detected for neutral O I. Assuming the HD 189733b magnetosphere to be at the origin of the early absorption, we use the Parker model for the stellar wind and a particle-in-cell code for the magnetosphere to show that its orientation should be deflected ∼10−30• from the planet-star line, while its nose's position should be at least ∼16.7 R p upstream of the exoplanet in order to fit the C II transit light curve. The derived stand-off distance is consistent with a surface magnetic field strength of ∼5.3 Gauss for the exoplanet, and a supersonic stellar wind impinging at ∼250 km s −1 , with a temperature of 1.2 × 10 5 K and a density ∼6.3 × 10 6 cm −3 at the planetary orbit, yet the fit is not unique.