I'm with Loyce on this one. You are basically trying to turn seed backup into a puzzle box, and puzzle boxes have a nasty habit of defeating the owner before they defeat the thief.
The biggest danger here is not someone cracking your clever setup. The bigger danger is future-you, six months from now, tired, annoyed, maybe with a dead USB stick, trying to remember exactly which pattern, which 12th word, which passphrase, which wallet, which version of the tool, and whether that persistent Tails storage was the "final final" one or the "testing final" one. That is how coins go into the great unspendable swamp.
For a hot wallet, fine, experiment a little if the balance is small and you understand the failure modes. But for a Coldcard I would not derive it from the same Border Wallet base material you used for the hot wallet. Let the Coldcard generate its own seed, or add dice rolls if you want to sleep well. Then back it up in a standard, recoverable way. Paper first, steel if meaningful funds, passphrase only if you are disciplined enough to back that up too. Boring is underrated in Bitcoin. Boring is what survives house moves, laptop deaths, panic, heirs, and "I'll totally remember this" syndrome.
Also, "the 12 words can be stored anywhere without risk" is the kind of sentence that disturbs the opsec part of me. It may not be the final seed by itself, but it is still part of the secret construction. If someone gets enough of the recipe, your clever separation turns into a treasure map with some steps missing. Maybe still hard, sure, but why create weird correlated material between hot and cold funds at all?
For cold storage, I'd keep it simple: Coldcard makes seed, you verify backup, you test restore on an offline device or spare signer before funding serious amounts, then use Sparrow/watch-only for monitoring and PSBT flow. The fancy part should be the signing hygiene, not the seed archaeology. Bitcoin already gives us enough ways to step on rakes. No need to also forge your own custom rakes in the garage, lol.
