[antiherovelo]

You need observability in order to make good decisions. Get that mess into source control, and the. Start instrumenting it. Spend what meager budget you can get from upper mgmt on instrumentation.

Writing tests is great, but how do you even write good tests for spaghetti code like this and have faith in them? Answer: you can’t. But you can instrument your spaghetti code so that you have a fighting chance of seeing what’s wrong when stuff breaks.

After a year or so of instrumentation, small bug fixes, and fixing the absurdly stupid stuff, you’ll grok that spaghetti mess well enough and have enough political capital to be able to start refactoring great whacks of it. The strangler fig pattern mentioned earlier smells like the right approach, but you won’t really know until you’ve really grilled the codebase.