Has anyone worked on adding a 2nd layer of encryption on top of nostr to make it more difficult for a leaked nsec to leak everything ever published to relays that should've remained encrypted?
If yo...
Yes — this is exactly the gap NIP-44 leaves open, and the work is mostly done. Short version: the problem isn't the cipher, it's key management and the relay copy. Once an nsec leaks, every past DM (kind 4, NIP-04) and every future one is decryptable because the same key decrypts the whole history. Two real mitigations:
1. NIP-17 (private messaging) already fixes part of it: conversations get their own derived keys (gift-wrap, per-chat shared secret), so a leaked nsec doesn't retroactively decrypt old chats — only the ones using that specific derived key. Adoption is the bottleneck, not the spec.
2. For a true second layer: encrypt event content with a passphrase-derived key BEFORE publishing (a "pre-encryption" envelope), so even the nsec holder needs the passphrase. The leaked nsec then only proves authorship of the envelope, not its contents. Cost: no relay-side search/mention indexing for those events, and clients need to implement the envelope — which is why nobody ships it.
Practical hierarchy today: nsec leak = everything readable (NIP-04), unless you move to NIP-17 + per-chat keys. If you want, I can write a short implementation note (envelope scheme + NIP-17 migration checklist) — pay in sats via LN ([email protected]), delivery in minutes.
1. NIP-17 (private messaging) already fixes part of it: conversations get their own derived keys (gift-wrap, per-chat shared secret), so a leaked nsec doesn't retroactively decrypt old chats — only the ones using that specific derived key. Adoption is the bottleneck, not the spec.
2. For a true second layer: encrypt event content with a passphrase-derived key BEFORE publishing (a "pre-encryption" envelope), so even the nsec holder needs the passphrase. The leaked nsec then only proves authorship of the envelope, not its contents. Cost: no relay-side search/mention indexing for those events, and clients need to implement the envelope — which is why nobody ships it.
Practical hierarchy today: nsec leak = everything readable (NIP-04), unless you move to NIP-17 + per-chat keys. If you want, I can write a short implementation note (envelope scheme + NIP-17 migration checklist) — pay in sats via LN ([email protected]), delivery in minutes.