The Jar Ships What It Promised

SEPTEMBER 30, 2026

I like following this project because it keeps giving me a reason to come back and check its homework, and this time the homework showed up faster than I expected. Two posts back I wrote about silent payments landing on the send side, with the wallet's author telling me he'd also worked out a way to recover those payments from a seed phrase — a claim I flagged as unverified, because the public repository hadn't moved since before he told me. Around the same time, in an unrelated message, he mentioned in passing that the latest build had picked up Nostr Wallet Connect too — the thing that lets you zap from a Nostr client using a wallet you already have, instead of every client needing its own. Both were "he told me" items sitting in my notes with a question mark next to them.

The question marks can come off. I pulled the public repository this morning and it had jumped forward about forty builds since the last time I looked — from page v844 to v882, engine v281 to v293, all landed over the last two days. Sitting in that jump: a new file called nwc.rs, thirteen hundred lines of it, which is the NWC support; and a new file called sp_scan.rs, eight hundred lines, plus a few hundred more added to the existing silent-payments code — which is the recovery machinery he described. Both things he told me about are now things I can point at.

What I Actually Checked

I didn't just count files and call it a day. The part I cared about most was the recovery claim, because it's the one that matters for money: if a silent payment scatters your coins across addresses nobody but you can find, the wallet had better be able to find all of them again from twelve words alone, on a phone that's never seen them before. Reading through it, the silent-payment keys the scanner uses to hunt down your coins are derived the same way every other key in this wallet is — straight from the same root key the twelve words produce, the SpKeys::from_root() call sitting right there in the code. That's the honest answer to "does this actually work from a fresh seed," and it does, by construction, the same way the rest of the wallet already does. The scanner itself walks the chain from a stored "birthday" height forward, hunting for outputs that match your key — which is exactly the kind of unglamorous, patient bookkeeping a recovery feature needs and a flashy demo doesn't. I haven't walked every line of that scanner start to finish, and I haven't watched it recover a real wallet on a real phone — that's still worth doing before the October recheck — but the shape of it matches what he told me, and the keys come from the right place.

The NWC side I read more lightly, since nothing about it touches how funds are held — it's a remote-control surface, not a custody one. It's there, it's substantial (thirteen hundred lines isn't a stub), and it does what he said: lets a Nostr client ask the wallet to pay without the client needing its own Lightning stack built in. He mentioned, almost as an aside, that he isn't fully happy with how it feels yet — the round trip over Nostr relays has some lag compared to a payment made directly in the wallet — and that he plans to keep working on that. I'd rather a builder tell me a feature works but isn't polished than tell me nothing and let me find out myself, and this is the second time in two posts he's volunteered the caveat before I asked for one.

Where This Leaves the Review

This doesn't move my verdict from the original review, and it isn't supposed to — the three things I said would turn "not yet for strangers" into "yes" were never about any single feature. They're still an LDK catch-up, an outside reader of the patch set, and a second registered LSP. But it's worth saying plainly what this pattern actually looks like from the outside: a guy tells me what he's building before it's done, I go check, and it's there, built the way he described, faster than most projects I follow manage with a whole team. Silent payments went from an idea in a text message to send-and-receive with a working recovery path in about a week. That's not nothing, and if he keeps shipping at this pace while taking his own security posture as seriously as the patch notes and the findings file suggest, the gap between "a jar I'd put my own sats in" and "a wallet I'd hand a stranger" is going to close a lot faster than most self-custodial wallets manage. I'll keep checking. So far, checking keeps being the fun part.