The barbell strength log you actually own.
Open-source · offline-first · self-hosted · MIT

---

Rampset is a training log for people who lift a barbell. Guided 5×5 and Madcow programs with automatic progression, freeform routines, per-side plate math, ramped warm-ups, and personal-record tracking — as an installable PWA you run yourself.

It's yours: reshape any program, rewrite the progression engines or add your own, keep every rep on your device, and export it all whenever. No subscription, no account to lose, no roadmap that abandons you.

Website & guide → https://isaacrowntree.github.io/rampset

What's in it

• Program mode — StrongLifts-style 5×5 (A/B, Plus, Lite, Mini, Ultra, Ultra Max). Tap-cycle plate circles, warm-up ramps, per-side plate math, automatic +weight on success and deload after repeated fails.
• Madcow mode — weekly ramped 5×5: heavy / light / intensity days, a Friday PR set that sets next week's top, back-off sets.
• Routine mode — freeform Strong-style set rows, every set prefilled from your last session.
• Progress — estimated 1RM, top set, volume and reps per lift, with a personal-record ★ the moment you beat an all-time best; body-weight trend.
• History — a calendar of every session plus recaps.
• Import — StrongLifts and Strong CSV exports (auto-detected, idempotent). Export gives everything back.
• Offline-first — IndexedDB on the device is the source of truth; the whole app works with no network (gym dead zones, airplane mode), and screen wake-lock keeps it on during a workout.

Run it locally

git clone https://github.com/isaacrowntree/rampset
cd rampset
npm install
cp .env.example .env.local     # then edit: your lifters, units, starting weights
npm run dev                    # → http://localhost:3000
npm test                       # vitest — the engines, importers, store, components

Self-host on Cloudflare

Rampset deploys to Cloudflare Workers via the OpenNext adapter, and uses Cloudflare Access (a free zero-trust login) as its auth — so only the emails you allow can get in, with no passwords for you to manage.

npm run deploy                 # opennextjs-cloudflare build && deploy

Then, once deployed:

1. Set your Cloudflare account id in wrangler.jsonc (replace the placeholder) or export CLOUDFLARE_ACCOUNT_ID.
2. In the Cloudflare dashboard, add an Access application in front of your Worker's URL and allow your lifters' emails (email one-time-code or Google). That's your login — each profile is keyed to the authenticated Access email.
3. Create the R2 bucket the BACKUPS binding points at (for cloud backups).

What it costs: Workers and Cloudflare Access are free for a household (Access is free up to 50 users), and there's no KV/cache to configure. The one paid gate is R2 — the app's backups sit comfortably inside R2's 10 GB free tier, but Cloudflare requires a payment method on file to enable R2 (you won't be charged under the free limits). Prefer no card at all? Remove the BACKUPS R2 binding from wrangler.jsonc — the app is fully offline-first on local IndexedDB and Settings → Export gives you a complete CSV backup whenever you want.

Configuration — no personal data in the repo

Lifters are defined entirely by an env var (.env.local, gitignored). See .env.example for the shape: id, name, email, accent (blue | green), unit (kg | lb), program template, and optional starting working weights. The committed code carries only generic program templates. Switch lifters with the avatar in the header (or Settings → Account).

Architecture

• Next.js 16 App Router, all screens client components.
• IndexedDB (Dexie) is the on-device source of truth — fully functional offline. Optional cross-device sync (a per-user SQLite Durable Object that orders each user's ops) and backups (R2 snapshots) via Cloudflare Workers.
• Engines are pure, unit-tested functions in src/lib/: linear progression
  • deload, Madcow ramps, routine prefill, plate-per-side math, Epley e1RM, PR detection. Editing them is the point — start there.
• PWA: manifest + service worker (network-first shell, offline fallback), installable to the home screen.
• Design: OLED black + IWF plate palette, per-user light/dark skins.

Contributing & making it yours

Fork it and go. The progression logic lives in src/lib/ (each engine has a .test.ts next to it — write a failing test, make it pass). New programs are data in src/lib/programTemplates.ts. It's MIT licensed: build the training app you want, closed or open, and keep your data.

License

MIT. Rampset is an independent open-source project. "StrongLifts" and "Madcow" refer to well-known training methods and are not affiliated with or endorsed by this project.