Wildfires represent a major natural driver of the land carbon cycle. Although their instantaneous combustion emissions are increasingly well captured by Earth system models (ESMs), the associated legacy carbon dynamics and post-fire ecosystem recovery remain largely absent from reduced-complexity ESMs and hard to isolate in fully fledged ones. Here we present OSCAR-fire v1.0, a wildfire carbon dynamics module developed within the OSCAR compact ESM. The module extends OSCAR's land carbon bookkeeping framework by introducing two new prognostic pools for coarse woody debris (CWD) and pyrogenic carbon (PyC) produced by incomplete combustion, alongside the existing living biomass and soil pools. A fire event induces a three-phase carbon cycle response in the OSCAR-fire module: instantaneous combustion emissions accompanied by dead biomass transfer, legacy dead carbon release, and gradual post-fire ecosystem recovery. The module's performance is assessed using prescribed burned area from GFED4.1s and GFED5.1 datasets, and parametric uncertainty is characterised using a Monte Carlo (MC) ensemble of 2000 configurations. Simulated global instantaneous fire emissions are in good agreement with both reference products, differing from reference totals by about 6 % under GFED4.1s and 3 % under GFED5.1. Legacy CWD emissions in boreal regions under GFED4.1s are broadly consistent with references but substantially higher under GFED5.1, and simulated PyC production under GFED4.1s falls below published estimates and exceeds them under GFED5.1. Besides, simulated post-fire recovery timescales reproduce the latitudinal gradient in ecosystem productivity. OSCAR-fire v1.0 provides a process-based and computationally efficient framework for quantifying wildfire contributions to the land carbon budget within a probabilistic modelling context, with broad applicability to historical carbon cycle attribution studies and assessments of fire-climate feedbacks under future climate scenarios.