Computing a density forecast and point forecast

hallo Leute

I have a question related to forecast with dynare
In my model, I do forecast by the command forecast inside the command estimation as following

mh_replic= 250000,
mh_nblocks = 1,
mh_drop = 0.45,
mh_jscale = 0.25,
mode_check_neighbourhood_size =1000,
forecast=4) data_pid data_wp data_c data_i x_hat data_R E_t data_y data_x data_m;

If I do this way, then Dynare will:
First of all, estimating model. then do doing forecast as next-two step following

Step 1: a denstiy forecast is computed
Step 2: then, a point forecast is calculated as the mean of the desity forecast.

So my understanding is correct?
If such, how dynare computes a density forecast at step 1?

As documented in Pfeifer (2014): An Introduction to Graphs in Dynare the mean forecast is generated by setting future shocks to 0 and then drawing parameters from the posterior distribution. The point forecasts in contrast sample for both the shocks and parameters.