This paper is devoted to illustrate the capabilities and features of a new package in the R statistical computing environment named ssym. This package allows to fit and calculate diagnostic statistics (e.g., residuals, local influence measures and goodness-of-fit statistics) for an extension (addressed in this paper) of the log-symmetric regression models discussed recently in the literature, which may be used to describe data whose response variable is continuous, strictly positive, assymetric and that may have outlying observations. In the log-symmetric regression model as well as in the package ssym both median and skewness of the response variable distribution are explicitly described through semi-parametric functions of explanatory variable values, where the non-parametric functions are approximated by natural cubic splines or P-splines, and the parametric components may be linear or non-linear. In addition, the distribution of the independent and multiplicative random errors may be log-normal, log-Student- t, log-power-exponential, log-slash, log-hyperbolic, log-contaminated-normal, Birnbaum-Saunders or Birnbaum-Saunders-t. Three real data sets are analyzed in order to illustrate the flexibility of the main functions and tools of ssym.