I think the overall idea is good.

If the cards are identified in the database as their hashes, wouldn’t editing the content reset all repetition data so far?

Anyone here has been using FSRS long enough to have comments about its effectiveness? I think it’s general consensus that moving from SM-2 to FSRS will show great improvement. I’m using SuperMemo 9 though, so it’s much harder to understand whether there will be an improvement or not.

I’ve been storing cards as markdown with a horizontal ruler separating the prompt from the response and another ruler separating historical data.

That is: the historical data in on the same file as the card. This makes cards trivial to sync.