Changes in version 0.3.0 (2026-11-18) First public release. - The kiqr() family fits Knowledge Integration Quantile Regression (KIQR) models with SCAD or LASSO penalties, using fixed tuning values, cross-validation (CV), or a high-dimensional Bayesian information criterion (HBIC). - Quantile tuning supports automatic lambda and zeta grids, a choice of automatic lambda path (lambda_scale), fold-specific prior coefficients, and parallel CV. - The kir() family fits sparse mean-regression models: the prior LASSO (pLASSO) and a prior SCAD (pSCAD) estimator, computed as SCAD- or LASSO-penalized least squares through ncvreg, with fixed fits, CV, or HBIC, automatic lambda and zeta grids, and fold-specific priors. - Fitted and tuned models provide print(), summary(), coef(), and predict() methods. Summary printing is controlled by kir.max_print and the max_print, show, and digits arguments.