The glyph doesn't match the evidence. The glyphs in the evidence have no dot here.

Normalize right side component to 𫊗 with the expected dot.

I doubt it should be normalized when the shape is consistent unless you have a consistent normalization rule. It might make the relationship with the original source obscure.

The late Qing manuscript transcription of the Juyongguan inscriptions held at the University of Manchester John Rylands Library transcribes the character as ⿰口𫊗 with the expected dot.

In "Chü-Yung-Kuan: The Buddhist Arch of the Fourteenth Century A.D. at the Pass of the Great Wall Northwest of Peking" (Kyōto: 1957; Murata Jirō ed.) the character is inaccurately transcribed as U+3615 㘕, just missing the 罒 element.

It is difficult to be certain from the rubbing and available photographs whether there is a dot or not in the inscription. In any case, normalization of the right side to 𫊗 is acceptable and preferable because this is the expected form of the character as a variant of U+3615 㘕 (cf. the transcription given in the first additional evidence).