An MVP is a learning instrument, not a stripped-down launch. It carries just enough to answer the riskiest question: will people actually use this to get a job done. Everything not needed to test that question is deliberately left out.
The discipline is defining "viable" honestly. Too thin and it teaches nothing; too broad and it stops being minimum. A well-scoped MVP produces evidence about demand and value quickly, which is why it sits close to the search for product-market fit and often follows a round of prototyping.