Wildland fire behavior is shaped by interactions among fuels, terrain, weather, and the lower atmosphere. Winds and fuel conditions control fire spread, while heat and moisture released by the fire can modify atmospheric stability and local winds, feeding back onto subsequent fire behavior. Here we present ERF-LSFIRE v1.0, a coupled fire–atmosphere modeling framework that integrates a C++/AMReX level-set surface-fire spread model based on the Rothermel formulation with the Energy Research and Forecasting model (ERF), providing a modular and scalable computational foundation for high-resolution coupled fire–atmosphere simulations. LSFIRE can be run standalone with prescribed winds or coupled to ERF in one-way and two-way configurations, with fire-generated sensible heat and moisture returned to the atmosphere in the two-way configuration. We verify the level-set implementation using analytical spread tests over flat and uniformly sloping terrain and demonstrate the coupled framework using simulations of the 2025 Palisades Fire in California and 2025 Cram Fire in Oregon. Standalone LSFIRE closely reproduces the analytical elliptical spread solutions. In the strongly wind-driven Palisades case, two-way coupling with ERF produces localized atmospheric perturbations but only modest changes in perimeter-scale spread. In the Cram case, where background winds are weaker and more variable, fire feedback produces larger near-fire perturbations and substantially alters fire progression relative to the one-way coupled simulation. These results demonstrate that ERF-LSFIRE provides a flexible framework for examining fire–atmosphere feedbacks and suggest that the influence of two-way coupling depends on the strength of fire-induced perturbations relative to the ambient flow. The framework also provides a basis for future applications involving wildfire impacts on energy-system operations and resilience.