You can obtain the HR as follows:

predictions <- predict(cox, newdata = data, type = "lp", se.fit = T)
HR <- exp(predictions$fit)

se.fit will get the fit (log HR) and the standard errors.