Could be done universally, or be only specific to the 'auto-blobing' of large values.

IndexedDB would store the blob hash in the database, and probably provide it on blob creation.

Checking the hash efficiently would probably happen on read, so there would have to be a callback to IDB (so creating a callback interface on the blob system mojom).

Another option is to just do this check on the auto-wrapped blobs, which we already have full control over and fully read into memory. This would be really easy to do in that case only.