Signet

Your Personal AI Clone. Written in your hand.

Upload a chat export — Signet learns exactly how you talk. It analyzes your tone, phrase length, vocabulary, and emoji habits to create an authentic replica of your texting style.

---

Overview

Never leave a friend on read. Modern communication is constant and overwhelming. When driving, attending meetings, sleeping, or working, we often leave people hanging, leading to missed opportunities or social friction. Traditional autoreply bots are robotic, rigid, and impersonal — they kill conversational flow instantly.

Signet is a full-stack AI cloning platform built for HackSprint 2k26. Feed it a WhatsApp chat export (or any messaging history), and it builds a vector embedding of your communication style. The result? An AI clone that talks, thinks, and responds exactly like you.

You can then chat with your AI clone, tune its creativity, explore public personas, and dive into conversation analytics — all wrapped in a glassmorphism-designed UI.

Why Signet?

Problem	Signet's Solution
Autoreply bots sound robotic	RAG-powered tone cloning — learns your exact vocabulary, emoji habits, and sentence rhythm
Generic AI doesn't know you	Semantic few-shot injection — retrieves your most similar past messages as context
Setup is too complex	One-click upload — export your WhatsApp chat and you're done
Privacy matters	Local-first architecture — ChromaDB + Ollama support for fully offline operation

The project spans three platforms:

Platform	Stack	Location
Web App	React 19, Vite 8, Tailwind v4	web/
Mobile App	Expo 54, React Native 0.81, Redux	mobile/
Backend API	Express, MongoDB, ChromaDB, Ollama / OpenAI	backend/

Features

Core AI

Feature	Description
Chat Upload & Parsing	Upload .txt WhatsApp exports; the regex-based parser intelligently extracts conversation pairs with timestamps, contact names, emoji counts, word counts, and message threading — even handling group chats and media messages
RAG-Powered AI Clone	Retrieval-augmented generation combines ChromaDB vector similarity search with your computed tone profile to produce replies that authentically sound like you — not a generic chatbot
Algorithmic Tone Profiling	Quantitative analysis of your messaging style: average reply length, emoji frequency, capitalization ratio (formal vs. casual), punctuation habits (ellipses, exclamation marks), and common slang patterns
Persona Library	Chat with predefined personas (Steve Jobs, Naruto, Einstein, Sherlock Holmes, and more) or explore community-created clones — each with curated tone profiles
Continuous Learning	Every chat with your clone adds new message pairs back into ChromaDB, making it smarter and more accurate over time — it literally learns from every interaction
Temperature Control	Tune the creativity slider from precise/factual (low temp ~0.3) to wild/creative (high temp ~1.5) — find your perfect balance
Deep Analytics	View rich statistics on your conversation patterns: message length distribution, emoji usage frequency, average response time, peak activity hours, and vocabulary richness
WhatsApp Auto-Pilot	Real-time WhatsApp Web client integration via whatsapp-web.js — whitelist contacts, set auto-reply timeouts, and let your clone handle responses when you're busy

Platform-Specific

Platform	Highlights	Details
Web	Glassmorphism UI · Light/dark theme · Responsive design · Interactive demo	web/ — React 19 SPA with Vite 8, Zustand state management, React Router v7, and a polished glassmorphism design system
Mobile	Native Expo app · Bottom-tab navigation · Offline mode · Biometric auth	mobile/ — React Native 0.81 with Expo 54, Redux Persist for offline state, React Navigation 7, FlashList for smooth scrolling
Backend	RESTful API · JWT auth · Rate limiting · Helmet headers · Graceful fallbacks	backend/ — Express 4 server with in-memory MongoDB fallback for dev, multi-provider LLM support (Ollama, OpenAI, Groq), ChromaDB vector store

Security & Privacy

Measure	Implementation
HTTP Security	Helmet middleware sets secure headers (CSP, X-Frame-Options, HSTS, etc.)
Password Hashing	bcrypt with 12 salt rounds — industry-standard password storage
Authentication	JWT-based with configurable expiry, refresh support
Rate Limiting	Dedicated rate limiters on /upload and /chat endpoints to prevent abuse
Session Expiry	TTL-based auto-cleanup — inactive sessions are automatically purged
Data Retention	Chat data is never stored permanently — automatically wiped after session expiry
Local-First Option	Run fully offline with Ollama + local ChromaDB — zero data leaves your machine

Tech Stack

Backend (backend/)

Category	Technology	Purpose
Runtime	Node.js 24, Express 4	HTTP server & API routing
Primary Database	MongoDB 7 with Mongoose 9	User accounts, sessions, personas, chat metadata
Vector Database	ChromaDB 1.9	384-dimensional embedding storage & similarity search
Embeddings	HuggingFace Xenova/all-MiniLM-L6-v2 (ONNX)	Local embedding generation via Transformers.js
Authentication	JWT + bcryptjs	Token-based auth with 12-round salted hashing
LLM Providers	Ollama (local), OpenAI API, Groq (Llama-3.3-70b)	Multi-provider LLM abstraction layer
File Upload	Multer + Sharp	Multipart file parsing & image processing
WhatsApp	whatsapp-web.js + Puppeteer	WebSocket-based WhatsApp Web automation
Testing	Vitest + Supertest + mongodb-memory-server	Unit, integration, and benchmark testing
Infrastructure	Docker Compose	Orchestrates ChromaDB + MongoDB + Ollama containers
Package Manager	pnpm (workspace)	Fast, disk-efficient monorepo management

Web Frontend (web/)

Category	Technology	Purpose
Framework	React 19 + Vite 8	Modern SPA with lightning-fast HMR
Styling	Tailwind CSS v4 + Glassmorphism tokens	Utility-first CSS with custom design system
State Management	Zustand 5	Lightweight, hook-based reactive state
Routing	React Router v7	Client-side routing with lazy-loaded pages
HTTP Client	Axios 1	Promise-based HTTP with interceptors
UI Components	Lucide React, react-dropzone, react-hot-toast, QR code	Icon library, file drag-drop, toast notifications, QR generation
Linting	Oxlint 1	Rust-based blazing-fast linting
Deployment	Netlify	SPA deployment with SPA redirect rules

Mobile (mobile/)

Category	Technology	Purpose
Framework	Expo 54 + React Native 0.81	Cross-platform native mobile runtime
Navigation	React Navigation 7 (native stack + bottom tabs)	Type-safe screen transitions
State Management	Redux Toolkit + Redux Persist	Predictable state with offline persistence
HTTP Client	Axios 1	Shared API client with web frontend pattern
UI Enhancements	Expo Blur, Linear Gradient, FlashList	Glassmorphism effects, gradients, high-perf lists
Build & Deploy	EAS Build (dev/preview/production)	Cloud-based iOS & Android builds
Language	TypeScript 5.9	Full type safety across the codebase

Architecture

┌─────────────────────────────────────────────────────┐
│                    Web (Vite)                        │
│   React 19 · Tailwind · Zustand · React Router      │
│   └─→ api/client.ts ←── VITE_API_BASE_URL           │
└──────────────────────┬──────────────────────────────┘
                       │ HTTPS / REST
┌──────────────────────▼──────────────────────────────┐
│              Backend API (Express)                   │
│   ┌──────────┐  ┌──────────┐  ┌──────────────────┐  │
│   │  Routes   │  │  Brain   │  │  LLM Providers   │  │
│   │  /upload  │  │  Embedder│  │  Ollama          │  │
│   │  /chat    │  │  Retriever│ │  OpenAI          │  │
│   │  /auth    │  │  Reranker│  │  Groq            │  │
│   │  /session │  │  Prompts │  │                  │  │
│   │  /persona │  │  Personas│  └──────────────────┘  │
│   │  /whatsapp│  └────┬─────┘                         │
│   └──────────┘         │                              │
│        ┌───────────────┼──────────────┐               │
│   ┌────▼─────┐  ┌──────▼──────┐  ┌───▼───────┐      │
│   │  MongoDB  │  │  ChromaDB   │  │  Memory   │      │
│   │ (Mongoose)│  │ (Vectors)   │  │  Cache    │      │
│   └──────────┘  └─────────────┘  └───────────┘      │
│        ┌──────────────────────────────────┐          │
│        │  WhatsApp Web Client             │          │
│        │  (whatsapp-web.js + Puppeteer)   │          │
│        └──────────────────────────────────┘          │
└──────────────────────┬──────────────────────────────┘
                       │ HTTPS / REST
┌──────────────────────▼──────────────────────────────┐
│              Mobile (Expo)                           │
│   React Native · Redux · React Navigation           │
│   └─→ api/client.ts ←── EXPO_PUBLIC_API_URL         │
└─────────────────────────────────────────────────────┘

Mermaid Diagram

graph TD
    A[WhatsApp Chat Export .txt] -->|Upload| B(Express Backend Parser)
    B -->|Extract Msg Pairs| C(Xenova Local Embedder)
    C -->|Vector Embeddings| D[(ChromaDB Vector Store)]
    B -->|Tone Statistics| E[(MongoDB Session Metadata)]

    F[Incoming WhatsApp Message] -->|whatsapp-web.js Listener| G{Is Whitelisted?}
    G -->|No| H[Ignore / Idle]
    G -->|Yes| I{User Replies in X seconds?}
    I -->|Yes| J[Cancel Auto-Pilot]
    I -->|No| K[Embed Incoming Message]

    K -->|Semantic Query| D
    D -->|Retrieve Top Pairs| L(Prompt Builder)
    E -->|Fetch Tone profile| L
    L -->|Compiled Prompt| M[LLM Engine: Groq/Ollama]
    M -->|Generated Reply| N[whatsapp-web.js Client]
    N -->|Auto-send Msg| O[Friend's Phone]

Loading

Data Flow

1. Upload → User uploads a WhatsApp .txt export → parser extracts conversation pairs (incoming + reply) → each pair is embedded via HuggingFace Transformers → vectors stored in ChromaDB → tone profile computed → session created in MongoDB.
2. Chat → User sends message → retriever queries ChromaDB for similar conversation pairs → RAG pipeline builds a system prompt + user prompt with retrieved examples → LLM generates a reply in the user's style → new pair is added back to the vector store (continuous learning).
3. Personas → Predefined persona pairs (Steve Jobs, Naruto, Einstein, etc.) are seeded at startup → users can browse, bookmark, and chat with them just like their own clones.
4. WhatsApp Live → WhatsApp Web client connects via QR code → real-time message ingestion → continuous learning from live conversations.

Getting Started

Prerequisites

Requirement	Version	Notes
Node.js	>= 22	Required for all runtime environments
pnpm	Latest	Recommended package manager (npm also works)
Docker	Latest	Required for local ChromaDB + MongoDB + Ollama
Git	Latest	For version control

1. Clone & Install

git clone https://github.com/your-org/pixelpwnz.git
cd pixelpwnz

# Install all dependencies (uses pnpm workspaces)
pnpm install

# Or install individually with npm:
# cd backend && npm install && cd ..
# cd web && npm install && cd ..
# cd mobile && npm install && cd ..

Troubleshooting: If you encounter ERR_PNPM_NO_PACKAGE_MANIFEST, ensure you're in the project root. If native dependencies fail (Sharp, Puppeteer), make sure your system has build tools installed (build-essential on Linux, Xcode CLI on macOS).

2. Start Infrastructure

cd backend
docker compose up -d

This spins up three services:

Service	Port	Purpose
ChromaDB	8000	Vector embedding storage & similarity search
MongoDB	27017	Primary data store (users, sessions, personas)
Ollama	11434	Local LLM inference (optional — falls back to Groq)

To verify everything is running:

docker compose ps

3. Configure Environment

# Backend — defaults work with Docker, but copy and tweak as needed
cp backend/.env.example backend/.env

# Web — point to your local backend
cp web/.env.example web/.env
# Edit web/.env: VITE_API_BASE_URL=http://localhost:5000/api

# Mobile — point to your backend
# Set EXPO_PUBLIC_API_URL in your shell or .env

Key environment variables to customize:

Variable	Default	Description
JWT_SECRET	(auto-generated)	Secret key for JWT token signing
MONGODB_URI	mongodb://localhost:27017/signet	MongoDB connection string
CHROMA_URL	http://localhost:8000	ChromaDB server URL
LLM_PROVIDER	groq	LLM provider: ollama, openai, or groq
GROQ_API_KEY	—	Required if using Groq as LLM provider
OPENAI_API_KEY	—	Required if using OpenAI as LLM provider

4. Seed Personas (optional)

Populate the database with predefined personas (Steve Jobs, Einstein, Naruto, etc.):

cd backend
node upload-personas.js

You should see output confirming each persona was uploaded. Run this once after initial setup.

5. Run the Application

You'll need three terminal windows to run all services simultaneously:

# Terminal 1: Backend API (http://localhost:5000)
cd backend && npm run dev

# Terminal 2: Web app (http://localhost:5173)
cd web && npm run dev

# Terminal 3: Mobile (Expo Go / simulator)
cd mobile && npm start

Verify it's working:

• Backend health check: curl http://localhost:5000/api/health → { "status": "ok" }
• Web app: Open http://localhost:5173 in your browser
• Mobile: Scan the Expo QR code with Expo Go on your phone

Project Structure

pixelpwnz/
├── backend/                        # Express API server
│   ├── src/
│   │   ├── brain/                  # AI core
│   │   │   ├── chromaClient.js     # ChromaDB vector store client
│   │   │   ├── embedder.js         # HuggingFace embedding pipeline
│   │   │   ├── index.js            # Brain module orchestrator
│   │   │   ├── personas.js         # Persona definitions & seed data
│   │   │   ├── promptBuilder.js    # RAG prompt construction
│   │   │   ├── reranker.js         # Result re-ranking logic
│   │   │   ├── retriever.js        # Vector similarity search
│   │   │   └── PROMPTS.md          # Prompt engineering docs
│   │   ├── llm/
│   │   │   └── provider.js         # LLM abstraction (Ollama, OpenAI, Groq)
│   │   ├── middleware/
│   │   │   ├── auth.js             # JWT authentication
│   │   │   ├── errorHandler.js     # Global error handling
│   │   │   └── upload.js           # Multer file upload config
│   │   ├── models/
│   │   │   ├── ChatMessage.js      # Chat message schema
│   │   │   ├── Persona.js          # Persona schema
│   │   │   └── User.js             # User account schema
│   │   ├── parser/
│   │   │   ├── index.js            # WhatsApp chat parser
│   │   │   └── regex.js            # Regex patterns for chat parsing
│   │   ├── routes/
│   │   │   ├── auth.js             # Register, login, profile
│   │   │   ├── chat.js             # Chat with clone
│   │   │   ├── config.js           # Public config endpoint
│   │   │   ├── persona.js          # Persona CRUD & bookmarks
│   │   │   ├── session.js          # Session CRUD
│   │   │   ├── sessions.js         # List user sessions
│   │   │   ├── stats.js            # Conversation analytics
│   │   │   ├── upload.js           # Chat file upload
│   │   │   └── whatsapp.js         # WhatsApp integration routes
│   │   ├── store/
│   │   │   ├── Session.js          # Session model
│   │   │   └── sessionStore.js     # In-memory + MongoDB session store
│   │   ├── whatsapp/
│   │   │   └── client.js           # WhatsApp Web client (whatsapp-web.js)
│   │   ├── config.js               # Environment configuration
│   │   ├── db.js                   # MongoDB connection (with in-memory fallback)
│   │   └── index.js                # Express app entry point
│   ├── tests/
│   │   ├── fixtures/               # Test chat exports
│   │   │   ├── simple-chat.txt
│   │   │   ├── group-chat.txt
│   │   │   └── media-heavy.txt
│   │   ├── auth.integration.test.js
│   │   ├── chat.integration.test.js
│   │   ├── latency.test.js
│   │   ├── parser.test.js
│   │   ├── promptBuilder.test.js
│   │   ├── retriever.test.js
│   │   ├── sessions.integration.test.js
│   │   └── upload.integration.test.js
│   ├── docker-compose.yml          # ChromaDB + MongoDB + Ollama
│   ├── upload-personas.js          # Persona seed script
│   ├── vitest.config.js            # Test configuration
│   ├── pnpm-workspace.yaml         # pnpm workspace config
│   └── .env.example                # Environment variables template
│
├── web/                            # React web application
│   ├── src/
│   │   ├── api/
│   │   │   └── client.js           # Axios HTTP client
│   │   ├── components/
│   │   │   ├── ui/                 # Primitive UI components
│   │   │   │   ├── card.jsx
│   │   │   │   └── skeleton.jsx
│   │   │   ├── DashboardLayout.jsx
│   │   │   ├── DeleteModal.jsx
│   │   │   ├── Footer.jsx
│   │   │   ├── InsightsModal.jsx
│   │   │   ├── InteractiveDotGrid.jsx
│   │   │   ├── MessageBubble.jsx
│   │   │   ├── MessageList.jsx
│   │   │   ├── Navbar.jsx
│   │   │   ├── PremiumLoader.jsx
│   │   │   ├── PrivacyModal.jsx
│   │   │   ├── ProtectedRoute.jsx
│   │   │   ├── StatsPanel.jsx
│   │   │   ├── ThemeToggle.jsx
│   │   │   └── ToastProvider.jsx
│   │   ├── pages/
│   │   │   ├── app-dashboard/      # Dashboard sub-pages
│   │   │   │   ├── BookmarksPage.jsx
│   │   │   │   ├── NewDashboardPage.jsx
│   │   │   │   ├── NewProfilePage.jsx
│   │   │   │   ├── NotificationsPage.jsx
│   │   │   │   └── WhatsAppPage.jsx
│   │   │   ├── ChatPage.jsx
│   │   │   ├── CreateNewPage.jsx
│   │   │   ├── DashboardPage.jsx
│   │   │   ├── DemoPage.jsx
│   │   │   ├── DocsPage.jsx
│   │   │   ├── ExplorePage.jsx
│   │   │   ├── LandingPage.jsx
│   │   │   ├── LoginPage.jsx
│   │   │   ├── NotFoundPage.jsx
│   │   │   ├── ProfilePage.jsx
│   │   │   ├── SecurityPage.jsx
│   │   │   └── UploadPage.jsx
│   │   ├── sections/               # Landing page sections
│   │   │   ├── AboutSection.jsx
│   │   │   ├── CtaSection.jsx
│   │   │   ├── DemoSection.jsx
│   │   │   ├── DocsSection.jsx
│   │   │   ├── FaqSection.jsx
│   │   │   ├── FeaturesPreviewSection.jsx
│   │   │   ├── HeroSection.jsx
│   │   │   ├── HowItWorksSection.jsx
│   │   │   ├── PrivacySection.jsx
│   │   │   ├── SecuritySection.jsx
│   │   │   └── TrustedSection.jsx
│   │   ├── store/
│   │   │   ├── authStore.js        # Auth state (Zustand)
│   │   │   ├── chatStore.js        # Chat state (Zustand)
│   │   │   └── uiStore.js          # UI/theme state (Zustand)
│   │   ├── assets/                 # Static assets
│   │   │   ├── hero.png
│   │   │   ├── react.svg
│   │   │   └── vite.svg
│   │   ├── App.jsx                 # Root component with router
│   │   ├── main.jsx                # Vite entry point
│   │   └── index.css               # Global styles + design tokens
│   ├── public/                     # Public assets
│   │   ├── favicon.png
│   │   ├── favicon.svg
│   │   ├── icons.svg
│   │   └── logo.png
│   ├── .oxlintrc.json              # Oxlint configuration
│   ├── index.html
│   ├── vite.config.js
│   ├── netlify.toml                # Netlify deployment config
│   └── .env.example
│
├── mobile/                         # React Native / Expo app
│   ├── api/
│   │   └── client.ts               # Axios HTTP client
│   ├── components/
│   │   ├── StatsHeader.tsx
│   │   └── ThinkingDots.tsx
│   ├── constants/
│   │   └── theme.ts                # Design tokens (colors, typography, spacing)
│   ├── navigation/
│   │   ├── AppNavigator.tsx        # Root navigator
│   │   ├── AuthNavigator.tsx       # Auth flow navigator
│   │   └── MainTabNavigator.tsx    # Bottom tab navigator
│   ├── screens/
│   │   ├── BookmarksScreen.tsx
│   │   ├── ChatScreen.tsx
│   │   ├── DiscoverScreen.tsx
│   │   ├── HomeScreen.tsx
│   │   ├── LandingScreen.tsx
│   │   ├── LoginScreen.tsx
│   │   ├── PrivacyModal.tsx
│   │   ├── ProfileScreen.tsx
│   │   ├── RegisterScreen.tsx
│   │   └── UploadScreen.tsx
│   ├── store/
│   │   ├── authSlice.ts            # Auth state (Redux)
│   │   ├── bookmarksSlice.ts       # Bookmarks state (Redux)
│   │   ├── chatSlice.ts            # Chat state (Redux)
│   │   ├── sessionSlice.ts         # Session state (Redux)
│   │   ├── hooks.ts                # Typed Redux hooks
│   │   └── index.ts                # Redux store configuration
│   ├── assets/                     # App icons & splash screen
│   │   ├── icon.png
│   │   ├── splash-icon.png
│   │   ├── favicon.png
│   │   ├── android-icon-background.png
│   │   ├── android-icon-foreground.png
│   │   └── android-icon-monochrome.png
│   ├── App.tsx                     # Root component
│   ├── index.ts                    # Expo entry point
│   ├── app.config.js               # Expo configuration
│   ├── eas.json                    # EAS Build profiles
│   ├── babel.config.js
│   ├── tsconfig.json
│   └── .env.example
│
├── render.yaml                     # Render.com deployment manifest
├── package-lock.json               # Root lockfile
└── README.md                       # This file

API

The backend exposes a RESTful API at /api. All endpoints return JSON responses. Authenticated endpoints require a Bearer <token> header.

Authentication

POST /api/auth/register

Create a new user account.

// Request
{ "name": "Alex", "email": "alex@example.com", "password": "secure123" }
// Response 201
{ "success": true, "token": "eyJhbG...", "user": { "id": "...", "name": "Alex", "email": "alex@example.com" } }

POST /api/auth/login

Log in with existing credentials.

// Request
{ "email": "alex@example.com", "password": "secure123" }
// Response 200
{ "success": true, "token": "eyJhbG...", "user": { "id": "...", "name": "Alex", "email": "alex@example.com" } }

GET /api/auth/me (Auth required)

Get the currently authenticated user's profile.

// Response 200
{ "success": true, "user": { "id": "...", "name": "Alex", "email": "alex@example.com", "createdAt": "2026-07-11T..." } }

Sessions & Chat

POST /api/upload (Auth required)

Upload a WhatsApp chat export (.txt file, multipart form-data).

Field	Type	Description
file	File	WhatsApp .txt export
contactName	string	Name of the contact for pair extraction

// Response 201
{ "success": true, "sessionId": "...", "pairsExtracted": 142, "toneProfile": { "avgLength": 8.3, "emojiFreq": 0.42, "capsRatio": 0.65 } }

POST /api/chat (Auth required)

Send a message to your AI clone.

// Request
{ "sessionId": "...", "message": "What do you think about the new design?" }
// Response 200
{ "success": true, "reply": "honestly? it's clean but the nav feels a bit crowded, maybe collapse the sidebar?", "sources": 3 }

GET /api/session/:id (Auth required)

Get session details including tone profile and message history.

DELETE /api/session/:id (Auth required)

Delete a session and its associated vector embeddings.

// Response 200
{ "success": true, "message": "Session deleted" }

GET /api/sessions (Auth required)

List all sessions for the authenticated user.

// Response 200
{ "success": true, "sessions": [{ "id": "...", "contactName": "Mom", "messageCount": 89, "createdAt": "..." }] }

GET /api/stats/:id (Auth required)

Get detailed analytics for a session.

// Response 200
{ "success": true, "stats": { "totalMessages": 142, "avgResponseTime": "2.4s", "emojiUsage": 42, "topWords": ["lol", "yeah", "okay"], "lengthDistribution": { "1-5": 23, "6-10": 45, "11-20": 38, "20+": 14 } } }

Personas

GET /api/persona/bookmarks (Auth required)

Get the current user's bookmarked personas.

POST /api/persona/:id/bookmark (Auth required)

Toggle a bookmark on a persona (add or remove).

// Response 200
{ "success": true, "bookmarked": true, "personaId": "..." }

System

GET /api/config

Get public configuration (LLM provider, features, etc.).

// Response 200
{ "llmProvider": "groq", "maxUploadSize": "10MB", "features": { "whatsapp": true, "analytics": true } }

GET /api/health

Health check endpoint.

// Response 200
{ "status": "ok", "uptime": 123456, "version": "1.0.0" }

Quick Reference

Method	Endpoint	Auth	Purpose
POST	/api/auth/register	No	Create account
POST	/api/auth/login	No	Sign in
GET	/api/auth/me	Yes	Current user
POST	/api/upload	Yes	Upload chat export
POST	/api/chat	Yes	Chat with clone
GET	/api/session/:id	Yes	Session details
DELETE	/api/session/:id	Yes	Delete session
GET	/api/sessions	Yes	List sessions
GET	/api/stats/:id	Yes	Session analytics
GET	/api/persona/bookmarks	Yes	Bookmarked personas
POST	/api/persona/:id/bookmark	Yes	Toggle bookmark
GET	/api/config	No	Public config
GET	/api/health	No	Health check

See the full API documentation for detailed request/response schemas.

Deployment

Backend (Render)

The backend is deployed via render.yaml which defines two services:

Service	Type	Description
signet-backend	Node.js Web Service	Express API with JWT auth, ChromaDB integration, and LLM routing
signet-chromadb	Docker Service	ChromaDB vector store for production embedding queries

Key production environment variables:

Variable	Description
NODE_ENV=production	Enables production middleware (compression, CORS hardening)
JWT_SECRET	Strong random secret for token signing
MONGODB_URI	Production MongoDB Atlas connection string
CHROMA_URL	Production ChromaDB service URL
LLM_PROVIDER	groq or openai (not Ollama in production)
CORS_ORIGIN	Frontend domain for CORS whitelist
RATE_LIMIT_WINDOW	Rate limit time window in ms
RATE_LIMIT_MAX	Max requests per window

Frontend (Netlify)

The web frontend is configured for Netlify deployment via web/netlify.toml:

[build]
  command = "npm run build"
  publish = "dist"

[[redirects]]
  from = "/*"
  to = "/index.html"
  status = 200

Deployment steps:

1. Push the web/ directory to a Netlify-linked Git repository
2. Set VITE_API_BASE_URL to your production backend URL (e.g., https://signet-api.onrender.com/api)
3. Deploy — Netlify automatically runs npm run build and serves the dist/ folder
4. All routes fall through to index.html for the React Router SPA

Mobile (EAS)

The mobile app uses Expo EAS Build with three profiles in mobile/eas.json:

Profile	Command	Use Case
development	eas build --profile development	Internal dev client with Expo Go-like debugging
preview	eas build --profile preview	TestFlight / Internal testing track
production	eas build --profile production	App Store / Google Play submission

# Build for production
cd mobile
eas build --platform all --profile production

Testing

Backend Tests (Vitest)

cd backend

# Run the full test suite
npm test

# Watch mode — re-runs on file changes
npm run test:watch

# Run benchmarks (response time measurements)
npm run test:bench

# Run a specific test file
npx vitest run tests/parser.test.js

Test Suites

Suite	File	Coverage	Fixtures Used
Auth	tests/auth.integration.test.js	Register, login, token validation, duplicate emails, invalid credentials	—
Chat	tests/chat.integration.test.js	Message flow, RAG pipeline, session context, temperature effects	simple-chat.txt
Upload	tests/upload.integration.test.js	File parsing, ingestion pipeline, tone profile computation	simple-chat.txt, group-chat.txt, media-heavy.txt
Sessions	tests/sessions.integration.test.js	CRUD operations, TTL expiry, edge cases	—
Retriever	tests/retriever.test.js	Vector search relevance, result ranking, similarity thresholds	simple-chat.txt
Parser	tests/parser.test.js	WhatsApp format parsing, group chats, media messages, edge cases	simple-chat.txt, group-chat.txt, media-heavy.txt
Prompt Builder	tests/promptBuilder.test.js	Prompt construction, system message assembly, few-shot formatting	—
Latency	tests/latency.test.js	Response time benchmarks, p50/p95/p99 metrics	—

Test Fixtures

Located in backend/tests/fixtures/:

Fixture	Description
simple-chat.txt	Standard 1-on-1 WhatsApp conversation (~50 messages)
group-chat.txt	Group chat with multiple participants (~120 messages)
media-heavy.txt	Chat with images, videos, and voice notes included

Web Linting

cd web

# Run Oxlint (Rust-based linter, extremely fast)
npm run lint

# Check specific rules
npx oxlint --deny-warnings

CI Pipeline

Tests run automatically on push via the CI pipeline. The pipeline:

1. Installs dependencies with pnpm install
2. Starts ChromaDB + MongoDB via Docker
3. Runs all Vitest test suites
4. Runs Oxlint on the web frontend
5. Reports results and coverage

Built With

Frontend & Mobile

Library	Version	Purpose
React	19.2	UI framework
Vite	8.1	Build tool & dev server
Tailwind CSS	4.3	Utility-first CSS
Zustand	5.0	Lightweight state management
React Router	7.18	Client-side routing
Expo	54	React Native framework
React Native	0.81	Mobile runtime
Redux Toolkit	2.12	Predictable state container
React Navigation	7	Mobile navigation

Backend & AI

Library	Version	Purpose
Express	4.19	HTTP server framework
MongoDB	7	Document database
Mongoose	9	MongoDB ODM
ChromaDB	1.9	Vector database
HuggingFace Transformers	—	Local embedding generation
Ollama	—	Local LLM inference
OpenAI	—	Cloud LLM provider
Groq	—	Fast inference LLM provider

WhatsApp & Automation

Library	Purpose
whatsapp-web.js	WhatsApp Web client automation
Puppeteer	Headless browser control
QR Code Terminal	QR display for WhatsApp auth

---

Built for HackSprint 2k26  ·  Made by PixelPwnz