I am a little confused about the transmission mechanism of shocks in DSGE model. I knew it is much easier to analyze the mechanism when the model is simple, while how should I analyze the transmission of shocks when the model is much more complicated ? Are there any skills or papers concerned with this problem ?

In Gali (2015), for example, he makes many references to his log linearized equations when explaining IRFs. He kinda derives an expression for each variable he wants to explain. The variable will be on left- hand side of the equation and the expression will be on the right, so he kinda knows what is causing what and which parameters are key to the results.

But I also think knowing how your model works helps. If taxes are increasing after a shock, then go to the tax equation and see which variables are affecting it, and then you can do some tracing from there.