A Python toolkit for competing-risks survival analysis. Ships a competing-risks random survival forest with a scikit-learn-compatible API, Aalen-Johansen CIF and Nelson-Aalen CHF outputs, cause-specific Wolbers and Uno IPCW concordance metrics, OOB Breiman permutation variable importance, Ishwaran minimal-depth variable selection, exact cause-specific TreeSHAP attributions, and bit-identical reproducibility with randomForestSRC. As of v0.4 also ships Fine-Gray subdistribution- hazard regression (cmprsk::crr-parity β̂), a stand-alone Aalen-Johansen cumulative-incidence estimator (cmprsk::cuminc-parity variance), Gray's K-sample test (cmprsk::cuminc Tests-parity), cause-specific Cox PH (survival::coxph-parity), and a competing-risks model-evaluation harness with IPCW time-dependent AUC / Brier / iAUC / IBS and quantile-decile calibration data (riskRegression::Score / plotCalibration-parity). As of v0.5 also ships penalized Fine-Gray regression (LASSO / ridge / elastic-net / MCP / SCAD via IPCW-weighted cyclic coordinate descent, cross-validated penalty selection; crrp::crrp-parity coefficients and sandwich SEs) and a much faster forest TreeSHAP. Renamed from `crforest` in 0.3.1.