These start the same way: something was eighty per cent done, and then it stopped.
We read the code first and talk to whoever still remembers it, then say plainly what is worth keeping and what is cheaper to replace — before anyone commits to a plan.
Not a rewrite by default
Rewrites fail in ways nobody can debug under pressure, and code that survived in production usually knows things the documentation does not. Sometimes replacing it is right. It should be a decision, not a reflex.