Although benthic microbial community offers crucial insights into ecosystem services, they are underestimated for coastal sediment monitoring. Sepetiba Bay (SB) in Rio de Janeiro, Brazil, holds long-term metal pollution. Currently, SB pollution is majorly driven by domestic effluents discharge. Here, functional prediction analysis inferred from 16S rRNA gene metabarcoding data reveals the energy metabolism profiles of benthic microbial assemblages along the metal pollution gradient. Methanogenesis, denitrification, and N2 fixation emerge as dominant pathways in the eutrophic/polluted internal sector (Spearman; p < 0.05). These metabolisms act in the natural attenuation of sedimentary pollutants. The methane (CH4) emission (mcr genes) potential was found more abundant in the internal sector, while the external sector exhibited higher CH4 consumption (pmo + mmo genes) potential. Methanofastidiosales and Exiguobacterium, possibly involved in CH4 emission and associated with CH4 consumers respectively, are the main taxa detected in SB. Furthermore, SB exhibits higher nitrous oxide (N2O) emission potential since the norB/C gene proportions surpass nosZ up to 4 times. Blastopirellula was identified as the main responsible for N2O emissions. This study reveals fundamental contributions of the prokaryotic community to functions involved in greenhouse gas emissions, unveiling their possible use as sentinels for ecosystem monitoring.