Nerve Documentation
Nerve is notification infrastructure. One API call fans a message out across SMS, email, push, WhatsApp and voice, with provider routing, failover and compliance handled for you — built around the providers and constraints that actually matter for Africa-first products.
Start here
| If you want to… | Go to |
|---|---|
| Send your first notification | Quickstart |
| Understand how requests are authenticated | Authentication |
| Know which providers back each channel | Channels & providers |
| Make retries safe | Idempotency |
| Read the endpoint-by-endpoint reference | API Reference |
How Nerve fits together
The gateway is the only thing you talk to. It authenticates the call,
enforces rate limits and idempotency, and hands the event to the broker — then
returns. Everything after that is asynchronous: a 202 means Nerve has
durably accepted responsibility for the notification, not that it has arrived.
The workers pick channels, apply subscriber preferences and DND rules, and dispatch to providers with per-provider circuit breakers. When a provider goes down, traffic moves to its failover partner without you doing anything.
Delivery outcomes come back as delivery receipts.
Why this documentation can be trusted
The API Reference is not maintained by hand. It is generated from the gateway's Rust source — the same types the server compiles against — and the build fails if the two disagree. So the reference cannot describe a field that does not exist, or omit one that does.
If you find the reference wrong, that is a bug in our pipeline, not a stale page. Please report it.