In volcanic zones, geophysical observations have identified the presence of deep-seated fluids around intrusions in the mid-crust. However, the fluid compositions and mechanisms of fluid migration at high temperatures in the crust are still poorly understood. In this study, we investigated plagioclase alteration in mafic schists (clinopyroxene [Cpx] schist, hornblende [Hbl] schist, and actinolite [Act] schist) at the contact with a quartz diorite intrusion on Kinkasan Island, NE Japan. The quartz diorite consists of hornblende and plagioclase, and crystallized at 670–760 °C and 0.30–0.45 GPa. The quartz diorite contains abundant pegmatitic veins, which consist of plagioclase at their margins and K-feldspar + quartz ± garnet in their centers. Some pegmatitic veins cut the schistosity of the metamorphic rocks. In the metamorphic rocks, two stages of fluid-mediated alteration during cooling were recognized. The first stage of fluid infiltration (Stage I) was characterized by the formation of a reaction zone around pegmatitic dikes in the Cpx schists. The Cpx was altered to hornblende (Hbl) and Ca-plagioclase (An83–95) was altered to Na-plagioclase (An36–67). The replacement proceeded along the grain boundaries of plagioclase and clinopyroxene. The second stage (Stage II) resulted in the formation of K-feldspar veins and replacement of plagioclase (An35–57) by K-feldspar (An0Ab1Or99) and albite (An2Ab95Or3–An12Ab87Or1) in the Act and Hbl schists, and the latter were probably derived from the Cpx schists. Assuming the pressure was the same as the crystallization of the quartz diorite intrusion, the alteration temperatures were estimated as 690–730 °C for Stage I and 400–570 °C for Stage II. Mass balance considerations with assumptions of the volume conservation during the replacements of plagioclase and clinopyroxene/amphibole grains indicate the Cpx schists gained Na and H2O and lost Ca in Stage I, and the Act schists gained K and Si and lost Al and Ca at almost constant Na in Stage II. This indicates that the fluid composition changed from Na- to K-rich during cooling. During the Stage II alteration, when ~45% of the plagioclase grains were altered, significant nano- to micro-scale pores were generated in plagioclase, which produced a porosity of ~3.0%–3.5% in the plagioclase and increased the whole-rock porosity by 1.3% ± 0.2%. These pores are now isolated, irregularly shaped, and occur preferentially along the replacement front, suggesting that the pores migrated during or after replacement. The altered zone in plagioclase developed from trans-granular microcracks rather than grain boundaries. Our results suggest that infiltration of K-rich fluids could be self-promoting, with such fluids forming their own pathways via dissolution and precipitation processes during plagioclase replacement. Given that plagioclase is the dominant mineral in various rock types in the mid-crust, reaction-induced porosity networks in plagioclase could be the dominant fluid pathway at mid-crustal depths. The nature of this porosity and fluid migration are potentially controlled by temperature and/or the composition of the fluids.