A better way to look at MMT is to distinguish axioms from theorems. There is only one axiom in MMT: Inflation (i.e., loss of confidence) is the only threat to the currency. EVERYTHING else can be deduced from that simple assumption.

Thus, the government can print money until it prints so much that people lose confidence in their ability to spend it. That's why government spending must be useful. The more useful it is, the less likely it will cause a loss of faith in the currency. From the MMT perspective, Government spending "pays for itself" if the result of the spending is greater confidence that the money in circulation will be spendable. Lots of things the government spends money on have that effect. But some don't.

MMT becomes relevant when potential supply exceeds actual demand. That can happen in a recession, but it can also happen after a surge in supply, as in the case of some major innovation in production or availability of goods and services (Think container ships, the internet, and robots.) If MMT were around in the late nineteenth century, there's no telling what we could have done by allowing silver money, much less fiat money. But MMT can only be relied upon if inflation can be measured in real time, and it's not clear we were ready for it back then.

As for the slippery slope, the main purpose of taxes may be to quell demand, but the main way to quell demand is not taxation. Raising interest rates works more efficiently, because most of our money is created by banks and not the government. Reduce creditworthiness by raising the cost of credit, and less money is created. That's better than confiscating it. More here. https://remarklj.medium.com/why-mmt-matters-now-949f0c9ad84a