You can explain to your boss that them wanting to check for undecidable properties is impossible.

It's important to know what isn't possible, to not throw infinite resources at the problem.

Sometimes probabilistic answers are okay, though.

But usually a good idea to understand that that's what you're aiming at producing.

Maybe a company could decide that developing a tool that, given two programs as imput, tells if those two programs are equivalent would be a great tool to help teams migrate their code to a different programming language (or equivalently migrate to a new version of the same language, like it happened from Python 2.x to Python 3.x). But you can reduce this problem to the halting problem and show that this is not possible. More generally, the halting problem is a great tool to prove that other more practical things are unfortunately impossible to achieve, at least in the general case. But as another commenter pointed out, given a specific problem you could come up with heuristics that work most of the times (but sometimes would not be able to give a conclusive answer)

Only preventative measures! One would know enough, for example, not to try to construct a static analysis tool to helpfully tell the user if their code will inadvertently loop infinitely. Imagine running the tool on itself.