Wouldn’t this be like asking an LLM to generate and svg image? Yes, technically it’s xml, but will it make sense when it’s rendered? Only if the model was specifically trained to understand svg. I never heard of kicad before, i assume it’s even more niche.