GLE App Template

Build a health app on the ParagonDAO network in minutes.

Quick Start

git clone https://github.com/paragon-dao/gle-app-template.git my-health-app
cd my-health-app
npm install
npm run dev

Open http://localhost:5173 — you have a working GLE app.

What This Template Does

1. Captures audio from the phone microphone (swap for any sensor)
2. Encodes via BAGLE API → 128 universal coefficients (512 bytes)
3. Compares encodings to detect changes over time

Make It Your Own

Change the Sensor

Edit src/hooks/useSensor.js. The GLE encoder only needs number[]. Replace the mic capture with:

• Accelerometer: DeviceMotionEvent → [x, y, z, x, y, z, ...]
• Heart rate: Camera PPG → [peak_intervals...]
• Typing cadence: Keydown timing → [interval1, interval2, ...]
• Any time-series: Just produce an array of numbers

Change the Analysis

Edit src/hooks/useBagle.js. The API gives you 128 coefficients. You decide what to do with them:

• Similarity: Compare two recordings (authentication, change detection)
• Trend: Store daily encodings, track drift over time
• Classification: Train a simple model on coefficient patterns

Change the UI

Edit src/App.jsx and components. The template uses inline styles — add Tailwind, shadcn, or whatever you prefer.

Architecture

Phone Sensor → number[] → BAGLE API → 128 coefficients → Your App Logic
                              ↑
                    https://bagle-api.fly.dev/api/v1/encode

The encoder is signal-agnostic. It doesn't know what sensor produced the data. Every input becomes exactly 128 numbers = 512 bytes. Every app on the network speaks the same language.

API Endpoints

Endpoint	Method	Body	Response
/api/v1/encode	POST	{ features: number[] }	{ encoding: number[128], dim: 128, latency_ms }
/api/v1/similarity	POST	{ encoding_a: number[], encoding_b: number[] }	{ similarity: 0-1, distance }
/health	GET	—	{ status, model_loaded, models }

Offline Encoding

For privacy-sensitive apps, encode on-device without hitting the API:

npm install @paragondao/bagle-sdk

import { gleEncodeSignal, cosineSimilarity } from '@paragondao/bagle-sdk';

const coefficients = gleEncodeSignal(samples);  // 128 coefficients, no network call
const score = cosineSimilarity(a, b);           // compare locally

Deploy as PWA

This template includes PWA support via vite-plugin-pwa. Run npm run build and deploy the dist/ folder to any static host (Cloudflare Pages, Vercel, Netlify, GitHub Pages).

Your app works offline after first load.

License

MIT — Build freely.