Almost every rock found at the surface is altered or modified due to the reaction with liquid water abundantly present on Earth. This modification is driven by the reactivity of the Earth’s rocky shell in near-surface conditions. Modified (or altered) rocks contain information about former climates even in the deep-past (i.e., Archean) when traditional marine sedimentary proxies are challenged due to their proneness to resetting. The water-rock reactions can be traced through geological time using the isotope exchange between silicates [SiO 4- —based structures] and H 2 O, where isotopes of O have masses 16, 17 and 18 Dalton. In this quest, my colleagues and I use O-isotopes as tracers of this fluid reactivity in shallow continental and oceanic crust. Using triple O-isotope geochemistry (δ′ 18 O and Δ′ 17 O) and in-situ isotope methods, we have looked at some of the oldest examples of Yellowstone-type systems, where the heat of magma and surface waters produced distinct low-δ¹⁸O altered rocks. Such materials represent a chemically resilient isotope fingerprint of ancient atmospheric precipitation that is in turn reflective of climate in distant geological past (i.e., in absence of better proxies). In the presentation I show a recent case study on a magmatic-hydrothermal complex of Neoarchean age that recorded precipitation with δ¹⁸O as low as -18 ‰. High-precision U-Pb zircon ages constrain this record at 2673.5 ± 0.3 Ma. This strategy is used to create temporally constrained quantitative climate reconstructions that aid understanding of continental landmass elevation, relative plate orientation and atmospheric gas content. I will also present some of our recent work on marine sediments that are traditionally used for paleoclimate reconstructions but suffer from diagenetic alteration. Some recent progress on experimental bulk rock-water exchange will be highlighted. Using tectonic and triple O-isotope context, I show that not every sample can be used to reflect Earth’s surface conditions in a straightforward manner. Instead, a new set of questions can be designed to study crustal conditions with materials that are susceptible to alteration.