OpenStrap backend

This is optional. The app does not need it.

OpenStrap is local-first: edge pulls the bytes off the band over Bluetooth, decodes them with protocol, computes every metric with analytics, and stores the results on your phone. None of that involves this worker, or any other server.

So what is this? A small companion worker backing a handful of narrow, mostly optional features — and nothing else. It does not receive your raw frames, it does not decode records, and it does not compute recovery, strain, sleep, or any other metric. All of that used to live here. It doesn't anymore.

What it actually does

Endpoint	What it's for	Required?
GET /app/status	OTA pointer + in-app announcements	No — the app works offline
GET /legal/terms	Terms text behind the onboarding consent gate	No
POST /auth/{request-otp,verify-otp,refresh}	Sign-in for legacy import only	Only to import
GET /profile, /strain, /sleep, /sessions	One-time read of an old OpenStrap cloud account	Only to import
POST /consent	Consent ledger entry	No
POST /telemetry	Opt-in crash/usage telemetry — device-id keyed, no account	No, off by default
POST /health/upload	Opt-in health-data contribution (full local .db)	No, off by default
DELETE /data	Right to be forgotten — erases everything keyed to you	—

The two data-collecting endpoints (/telemetry, /health/upload) are opt-in, off by default, and are compiled out entirely of App Store and Play Store builds. See edge's PRIVACY.md for the full two-tier policy.

Why it still exists

Three reasons, none of them "computing your metrics":

1. Legacy import. Early OpenStrap had a real cloud backend. People have data in it. This lets them pull it down once and then never talk to a server again.
2. An OTA/announcement pointer, so a sideloaded app can find out there's a newer build.
3. Opt-in contribution, for people who actively want to donate data to improve the sleep-staging and protocol work.

Run your own

One Cloudflare Worker on the free plan — no Queues, no Cron, no Durable Objects, by deliberate design (see the comments in wrangler.toml, and keep it that way).

npm install
wrangler d1 create openstrap-v3-slim-db
wrangler r2 bucket create openstrap-v3-health
wrangler d1 execute openstrap-v3-slim-db --file=src/db/schema.sql
wrangler secret put JWT_SECRET
wrangler deploy

Then point the app at your worker instead of mine. If you're self-hosting just for yourself, set STATIC_OTP so it doesn't need an email provider — never set that on a shared deployment.

History

This repo used to be the centre of OpenStrap: it ingested raw frames, stored them in R2, rolled them into minutes, and computed everything on a cron. The architecture moved on-device in mid-2026 and the server was cut down to what you see here. If you want the old design, it's in src/db/migrate_to_slim.sql and the git history.

---

Not affiliated with, endorsed by, or connected to WHOOP. MIT licensed.