I just replied to you through email. However, I am posting the developer’s comments here for future reference. (Hat tip to Dr. Adrian L. Dempwolff)

Since the evaluation of CAP trajectories and the extraction of resonance
parameters (i.e., position and width) often requires very careful
fine-tuning of some associated parameters, a direct computation of the
quantities of interest after the electronic structure calculation would
not be very beneficial, since the whole calculation might have to be
repeated several times before a meaningful CAP trajectory has been
generated. For example, the CAP/EA-ADC(3) calculation of the example
took ~ 10 min, but the evaluation of a CAP trajectory usually only takes
a couple of seconds. Repeating the CAP/EA-ADC calculation over and over
again only because of changed trajectory parameters would thus just be a
waste of resources.

In the subspace-projected CAP formulation, which the CAP/ADC
implementation relies on, it is instead possible to analyze CAP
trajectories a posteriori using a separate program, which is indeed not
very complex, i.e., can be written in < 500 lines of python including a
Q-Chem output file parser and CAP trajectory plot output etc. All
needed input for sich a program is given in the Q-Chem output of the
CAP/EA-ADC(3) calculation, i.e., the (diagonal) matrix of EA-ADC(3)
eigenvalues (usually called H0, the real part of the subspace-projected
CAP/EA-ADC(3) matrix), and the CAP representation within converged
EA-ADC(3) states, which enters the CAP trajectory computation as
imaginary part (usually called W), scaled with the CAP stength eta.

Having said that, there are tools out there exactly designed for this
purpose. For example, see
[Analysis Tools — pyopencap documentation]:
The matrices as appearing in the CAP/EA-ADC(3) output can directly be
used to construct the CAPHamiltonian object.