No, I think the question is how to actually create these plots and the plots in the post just serve as an example/illustration of what @Lin_Caijin wants to create. In a simple Cox model each variable is associated with a fixed hazard ratio, so what I think is visualized here is the interaction effects between the continuous (x-axis) variable and the categorical (y-axis) variable.

Some googling shows that these figures come from the following poster:

https://sabcs18.posterview.com/nosl/p/P3-11-02

The authors describe here indeed that they fit a multivariable Cox model that includes treatment (y-axis variable), risk score (x-axis variable) and a interaction term between the both of them. If you want to create a similar plot for the analyses you are performing, you should start with fitting a similar model. You can then calculate the HR for the treatment at each value of the continous variable and plot them similarly to the examples above.