In Standard Model, proton cannot decay in tree level, it does so only via higher dimensional non-renormalizable interactions, which are known as sphaleron effects. In theories beyond standard model such as GUTs, proton can decay via tree level diagrams. In such cases, however, the related diagram involves a very heavy gauge boson as a virtual propagator, which again suppresses this process considerably, and then theory becomes consistent with experiment.
Dynamics of proton decay process has to be seen at the quark level. GUT physics allows lepto-quark as well as di-quark couplings of heavy bosons. You can easily visualize the following mode. Exchanging a heavy X boson, a d quark transform to e+ whereas a u quark transforms to ubar. There is third u quark whose role is purely that of a spectator. Then the combination uud transforms to u ubar and e+, that is a meson and a positron. This is one of the well known decay modes of proton.
Proton decay has not been experimentally observed yet. Non-observation puts lower limits on the masses of heavy gauge bosons. In this way one can rule out the minimal SU(5) theory.
On the other hand there are theoretical mechanisms which can suppress proton decay. One such model is given by Kawamura which utilizes an orbifolding mechanism. In this scenario one embeds SU(5) GUT in a 5-D world. That is SU(5) gauge bosons are 5-D fields. Fermions however are strictly four dimensional fields, that is they stay at the boundary of the 5-D world.
you can read a write up by myself, https://www.researchgate.net/publication/1910678_Orbifold_GUT_model_with_nine_Higgs_doublets
Article Orbifold GUT model with nine Higgs doublets
I was aware of observed proton decay into neutron + positron in some unstable atomic nuclei, but had never heard of observed proton decay in to meson + positron.
Because (modifications of) conservation laws allow it, a physical mechanism within the Standard Model, consistent with such modifications, seems to exist and it's possible to identify the products of the decay as known particles, therefore to design experiments.
However it's by no means obvious that the approximations involved are, indeed, robust to corrections and, therefore, whether the modification to the conservation law that it would imply, is, indeed, a property of the Standard Model, or a property of the approximation. Indeed proton decay inevitably would involve effects beyond the Standard Model and that's what's really the issue.
That's the reply to the question.
The statement preceding the question, however, is wrong: it's not known, at present, if the proton does, in fact, decay. If it did, its lifetime would have to be longer than 1032 years, by experiments that attempt to identify the particles it could decay to with a specific proton. For the moment such searches haven't found any such events.
The proton can't decay to neutron + positron, however, due to conservation of energy-the neutron has a greater rest mass than the proton. One shouldn't confuse inverse beta decay with proton decay. The former conserves baryon number; the latter doesn't.