Rohin Shah comments on Another take on agent foundations: formalizing zero-shot reasoning