Easy-to-manage touch screens
Free, open-source touch screens. Guided USB setup, a drag-and-drop editor and local control. Made for Home Assistant.
翻訳待ち
No updates detected yet. GitHub releases will appear here.
AI-powered memory tracking app
翻訳待ち
Loading the full README. The saved excerpt is shown below.
Larder is an AI-powered inventory tracker that turns grocery receipts into a smart kitchen. Upload a photo of any receipt and Claude AI extracts every item — name, quantity, category, price, and estimated shelf life — into a pantry view that warns you before things expire. The average US household wastes ~$1,500 of food a year; Larder helps you fix that.Read the full README ↗Larder is an AI-powered inventory tracker that turns grocery receipts into a smart kitchen. Upload a photo of any receipt and Claude AI extracts every item — name, quantity, category, price, and estimated shelf life — into a pantry view that warns you before things expire. The average US household wastes ~$1,500 of food a year; Larder helps you fix that.
A production full-stack application built with Next.js 16 + React 19 on the frontend, FastAPI + Python on the backend, Supabase (Postgres + JWT auth) for data and identity, and Anthropic Claude for AI-powered receipt parsing. Deployed across Vercel and Render with GitHub Actions CI/CD.
Screenshots are checked into docs/screenshots and captured from the local production builds with a seeded demo pantry session.
| Pantry dashboard | Expiring soon |
|---|---|
| Scan receipt | Settings |
|---|---|
| Marketing landing page |
|---|
A complete end-to-end product built solo — frontend, backend, database, auth, AI integration, DevOps, and design.
| Area | Where it lives |
|---|---|
| Frontend engineering | 3 Next.js 16 App Router apps (web + marketing + docs), 11 routes, ~3,500 LOC of TypeScript/React |
| Backend engineering | FastAPI service with 10 endpoints, Pydantic v2 validation, dependency injection, lifespan hooks |
| Authentication & security | Supabase JWT (ES256) verification with JWKS, key-rotation fallback, type-email-to-confirm flows for destructive actions |
| Database design | Postgres schema with soft deletes, status state machine, re-purchase inference, indexed queries |
| AI / ML integration | Claude vision API for receipt parsing, structured JSON extraction, shelf-life estimation |
| DevOps & deployment | Three independent deploy targets (2× Vercel, 1× Render), explicit CORS allowlist via FRONTEND_URL, env-var-driven config |
| CI/CD | GitHub Actions: ruff lint, Next.js typecheck, auto-merge on green with retry loop |
| Product & UX design | Warm earthy design system, Notion-style sidebar, Headspace-inspired marketing site, animated mockups |
| Compliance | GDPR-friendly account deletion (cascades pantry + receipts + auth user), public privacy policy |
| URL | |
|---|---|
| 🏠 Marketing site | https://larder-website.vercel.app |
| 📱 Web app | https://larder-theta.vercel.app |
| 📖 Docs | https://larder-website.vercel.app/docs |
| 🔒 Privacy policy | https://larder-website.vercel.app/privacy |
| ⚙️ Backend API | https://larder.onrender.com |
| ❤️ Health check | https://larder.onrender.com/health |
| 💻 Source | https://github.com/Chinmay1220/Larder |
/expiring page grouped into Expired / Today / Next 3 daysPASSWORD_RECOVERY event handlingrequestAnimationFrame-based count-up stats@keyframes marquee ticker stripLarder treats Claude as a first-class component, not a wrapper.
This is the beginning of the README. Read the rest at the original. Read the full README ↗