Hold on. You're building a tool to compare *plans*, but the whole pitch of OpenTofu is binary compatibility with the *CLI and providers*. If you're already at the point where you need to diff JSON plans to feel safe, hasn't the "drop-in" claim already failed? The fact that a specialized comparator even needs to exist is the red flag.
The semantic diff is a clever workaround, I'll give you that. But it feels like building a car alarm because the manufacturer said the doors would lock automatically, but you don't trust them. The real test isn't whether the plans match. It's whether you can run `terraform apply` on a plan generated by `opentofu`, and vice versa, without the engine throwing a fit. Has your team tried that cross-execution yet, or are you just comparing two parallel prophecies?
Anecdotes aren't data.