NOSTRAS's relay, DVM, and push stack is real, production-tested software — not a weekend project. Outbox aggregation, payment-gated AI jobs, push notifications that actually work: we already built it, broke it under real load, and fixed it. NOSTRAS Node is that same stack, for your own instance.
Not hypothetical — already lived through it
This isn't a pitch for software that might work. It's the exact
stack running nostras.app today, including the real
incidents that shaped it.
Memory/connection limits found the hard way, under a real near-outage — not textbook defaults.
A real NIP-90 DVM with live Lightning invoicing, priced per job — not a mocked demo.
Fetching other users' write relays without it becoming your own bottleneck under load.
A real end-to-end job round-trip, not just an HTTP ping that stays green while a worker silently dies.
A real NIP-98 auth check, not a third-party push service that becomes another party in your data's path.
Workers that reconnect after a drop instead of silently staying down until someone notices.
Three shapes, one codebase
Pricing isn't locked yet — that's part of what the waitlist is for. These are the shapes we're actually building toward.
Good for: trying it, or a relay you're happy to run bare.
flyctl deploy, we don'tGood for: comfortable with infra, just don't want to build it.
Good for: want your own node, not your own on-call rotation.
Even Managed Node doesn't mean giving up custody. This is about who runs the server processes — your keys and your wallet connection never leave your own client, on any tier, the same non-custodial rule NOSTRAS itself has always followed.
Get on the list
We'll follow up directly once pricing and provisioning are ready. Real interest here is what decides whether this gets built out further.
Used only to follow up about NOSTRAS Node. Nothing else.
Prefer email? Reach us directly at hello@nostras.app.