Hi Nur,
You can find the computation in
https://arxiv.org/pdf/1308.4685.pdf My experience is that rather than working with the Christoffel (which is not a tensor and for which xAct has very specific rules) it's better to define the difference between the normal and the disformal christoffel, which is a well defined tensor (cf eq 35 of the aforementioned paper). From that you can compute the Ricci, the Einstein-Hilbert action and easily isolate the total derivatives (eqs 36, 38)
Another trick you can use is to redefine the field to pi(phi) = \int_0^phi \sqrt{B(x)} dx. That way you remove all the derivatives of B, and you can always undo the transformation at the end (see eq 27 of see also
https://arxiv.org/pdf/1210.8016.pdf). With this trick the computation is simple and can be done/checked by hand.
If that doesn't work I have some old code to compute (more complicated) disformal transformations that worked reasonably well (although it might be for an older version of xAct).