This is an example likelihood package that can be externally installed and then referenced from Cobaya .yaml files via the qualified module name. It just reproduces the Planck 2018 lensing likelihoods already installed with Cobaya.
The likelihood classes can also be instantiated and called directly:
from plancklens import PlanckLensing
like = PlanckLensing()
print('chi2 = ',-2*like.log_likelihood(cl_dict, A_planck=1.0))
This example packages data internally. For larger data the data should probably be stored externally and installed separately.