Neural interfaces to peripheral nerves are increasingly employed for basic research, neuroprosthetics, and bioelectronic medicine (‘electroceuticals’). We previously established a pipeline for hybrid electromagnetic-electrophysiology modeling of nerve stimulation on the o2S2PARC platform for collaborative, FAIR, and reproducible computational neurosciences. It leverages histological data to generate a detailed anatomical model of the nerve microstructure and to functionalize it with populations of fiber neuro-dynamics models (myelinated/unmyelinated, statistical distributions of fiber types and diameters), for the purpose of predicting and optimizing stimulation selectivity and efficiency. This pipeline has now been extended with functionality to study measurable signals originating from neural activity – namely the compound action potential (CAP) and electrical impedance tomography (EIT) – to facilitate their interpretation, help maximize their information content, and support closed-loop control applications. This required a series of innovations: i) an extended version of the reciprocity theorem to compute signals from spatio-temporally distributed neural activity in complex dielectric environments and for realistic electrode geometries, ii) a generalized activating function for the efficient prediction of fiber recruitment, iii) a semi-analytic model of CAP formation, iv) integration of EIDORS for tomographic reconstruction, and v) iterative and hybridized multi-goal genetic optimization and surrogate modeling building. The resulting pipeline has been i) validated against experimental data from porcine vagus nerve stimulation and recording, ii) used to understand subtle cancellation effects complicating the CAP signal interpretation, and iii) leveraged to control stimulation parameters to selectively activate targeted fascicles in complex nerves based on EIT.