🛡️ PrepZone — AI-Powered JECA Preparation & Learning Ecosystem

A high-performance, gamified learning platform tailored for WBJECA (MCA Entrance) aspirants. Combining adaptive AI performance diagnostics, offline-resilient practice workflows, multi-roadmap study planning, and predictive rank modeling into a distraction-free glassmorphic workspace.

🌟 Overview

PrepZone addresses the primary pain points of competitive entrance preparation: fragmented tracking, lack of actionable diagnostics, unpredictable rank projections, and loss of motivation over long preparation cycles.

Designed specifically for the West Bengal Joint Entrance for Computer Applications (WBJECA) syllabus, PrepZone structures preparation around daily quantifiable goals, real-time accuracy telemetry, automated mistake logging, and an intelligent ranking engine that models candidate performance against approximately 25,000 yearly competitors.

┌─────────────────────────────────────────────────────────────────────────┐
│                           PrepZone Ecosystem                            │
├───────────────────┬───────────────────┬─────────────────────────────────┤
│   Track & Plan    │   Test & Learn    │        Analyze & Predict        │
│ • Multi-Roadmaps  │ • LLM Import      │ • Mistral AI Diagnostic Engine  │
│ • Daily Sessions  │ • Code Snippets   │ • Pro V2 Rank Prediction (GMR)  │
│ • Mission Targets │ • Offline Dexie   │ • 365-Day Consistency Heatmap   │
│ • Spaced Revision │ • Auto Double-Sync│ • Target Institute Pinning      │
└───────────────────┴───────────────────┴─────────────────────────────────┘

✨ Core Capabilities

🎮 Dual-Tier Gamification System — Real-time XP engine with 7 competitive league ranks (Bronze to Grandmaster), dynamic level thresholds, and streak tracking with a 1-day grace period.
🧠 Hybrid AI Diagnostics — Mistral AI (mistral-small-latest) surfaces personalized study recommendations with a 24-hour / 25-MCQ token-saving cache layer and a deterministic rule-based fallback.
🎯 WBJECA Rank Predictor (Pro V2) — Non-linear mathematical model calculating Candidate Readiness (0–100%) and General Merit Rank (GMR) with proximity-weighted college admission probabilities across 14 West Bengal institutions.
⚡ MCQ Master with Zero-Data-Loss Offline Practice — Custom JSON test importer, syntax-highlighted code problem player, in-progress auto-saving to IndexedDB, and automatic background synchronization upon reconnection.
🗺️ Multi-Roadmap Study Planner — Create and switch between multiple independent preparation roadmaps with integrated AI prompt generation, timeline views, and monthly calendar heatmaps.
📊 Real-Time Analytics & Dashboards — Aggregated accuracy breakdowns across 11 core computer science subjects, weekly progress vs. custom daily goals, and GitHub-style 365-day consistency heatmaps.
📓 Mistake Bank & Active Recall Queue — Automatic and manual error indexing with root-cause correction recording and prioritized spaced-repetition revision queues.
📱 PWA & Mobile-First Glassmorphism — Installable Progressive Web App with Service Worker background caching, custom bottom navigation, and an immersive dark crimson design system.

🏗️ Technical Architecture

PrepZone is structured around a decoupled MERN architecture with an offline client layer and an asynchronous AI diagnostic pipeline.

graph TD
    subgraph Client["Frontend Client (React 19 + Vite 6)"]
        UI[Glassmorphic UI / SCSS Modules]
        Router[React Router v7]
        ReduxStore[Redux Toolkit Store<br/>7 Domain Slices]
        SW[Service Worker / Workbox]
        IDB[(IndexedDB / Dexie.js<br/>Offline Tests & Sync Outbox)]
    end

    subgraph Backend["Express 5 REST API (Node.js ESM)"]
        MW[Auth, CORS, RateLimiter, Error MW]
        RouterAPI[Express Router Hierarchy]
        Controllers[Controller Layer]
        Services[Service & Business Logic Layer]
        AIModule[AI Diagnostic Module]
    end

    subgraph External["External Services & Storage"]
        MongoDB[(MongoDB Atlas<br/>Mongoose 9.4)]
        MistralAPI[Mistral AI API<br/>mistral-small-latest]
    end

    UI --> ReduxStore
    ReduxStore --> Router
    Router --> RouterAPI
    SW <--> IDB
    UI <--> IDB

    RouterAPI --> MW
    MW --> Controllers
    Controllers --> Services
    Services --> MongoDB
    Services --> AIModule
    AIModule --> MistralAPI
    IDB -.->|Background Sync via /api/practice/:id/submit| MW

Frontend

React 19 paired with Vite 6 provides the component layer and build tooling. Redux Toolkit manages centralized application state across 7 domain slices — authentication, study tracking, practice, analytics, planning, gamification, and college prediction. Sass (Modern Modules) powers the glassmorphic design system through composable SCSS tokens and mixins. Recharts renders performance charts, subject-accuracy gauges, and weekly dashboards. Dexie.js (IndexedDB wrapper) provides offline exam session persistence and a sync outbox queue. Route-level code splitting via React Router v7 with Suspense boundaries keeps the initial bundle lean.

Backend

A Node.js ESM server running Express 5 organizes all API logic through a Controller-Service architecture, enforcing separation between request handling and business logic. Each domain (auth, tracker, analytics, practice, planner, mistakes, revision) is a self-contained module with its own router, controller, and service layer. The Mistral AI SDK is integrated directly into the Analytics Service for asynchronous diagnostic generation. Winston and Morgan provide structured file/console logging and HTTP telemetry across all environments.

Database

MongoDB Atlas serves as the cloud-hosted persistence layer, accessed via Mongoose 9.4 for schema-enforced document storage. Aggregation pipelines compute real-time analytics — accuracy distributions, weekly breakdowns, 365-day heatmaps — without storing redundant pre-computed metrics. Strategic compound indexes ({ userId, date } on Tracker and { subject, question } on QuestionBank) optimize query performance for the most critical access patterns.

Project Directory Structure

PrepZone/
├── Backend/
│   ├── server.js                     # Server entrypoint & graceful shutdown handlers
│   └── src/
│       ├── app.js                    # Express app initialization, CORS, routing, error MW
│       ├── modules/
│       │   ├── analytics/            # Summary, weekly breakdown, Mistral AI suggestions
│       │   ├── auth/                 # User model, JWT auth, goals, target colleges
│       │   ├── college/              # WBJECA institutions model, seeding & controller
│       │   ├── mistakes/             # Mistake logging, correction bank & analytics
│       │   ├── planner/              # Multi-roadmap study planning & target scheduling
│       │   ├── practice/             # MCQ Tests, Question Bank & Result processing
│       │   ├── revision/             # Prioritized revision items & completion status
│       │   └── tracker/              # Daily study logging, session history, streak logic
│       └── shared/
│           ├── config/               # Database connection & environment validation
│           ├── middleware/           # auth.middleware.js, error.middleware.js, rateLimiter.js
│           └── utils/                # logger.js, responseHandler.js, constants.js
│
└── Frontend/
    └── src/
        ├── App.jsx                   # Route definition & code-split Suspense boundaries
        ├── components/               # GlassCard, Sidebar, Topbar, ReloadPrompt, InstallPrompt, Modals
        ├── hooks/                    # useAuth.jsx, useOnlineStatus.js
        ├── layouts/                  # DashboardLayout.jsx with background layers & auto-sync
        ├── modules/                  # Page Views (Dashboard, Tracker, Practice, Planner, Predictor, etc.)
        ├── services/                 # Axios API client, Dexie DB wrapper, syncService
        ├── store/                    # Redux Toolkit root store & domain slices
        ├── styles/                   # SCSS variables, mixins, global styles, page stylesheets
        └── utils/                    # predictionUtils.js, highlighter.js, chartUtils.js, dateUtils.js

🛠️ Technology Stack

Frontend

Layer	Technology	Version	Responsibility
Framework	React	`19.1.0`	Declarative UI component architecture
Build Tooling	Vite	`6.3.1`	Fast HMR and optimized production bundling
State Management	Redux Toolkit	`2.11.2`	Centralized state across 7 domain slices with async thunks
Routing	React Router DOM	`7.5.0`	Client-side routing with lazy-loaded, code-split modules
Styling	Sass (Modern Modules)	`1.86.3`	Glassmorphic design tokens and composable SCSS architecture
Data Visualization	Recharts	`2.15.3`	Performance charts, accuracy gauges, and subject graphs
Offline Storage	Dexie.js (IndexedDB)	`4.4.4`	Client-side test caching and background sync outbox queue
PWA / Service Worker	Vite Plugin PWA	`0.21.1`	PWA manifest, Workbox asset pre-caching, and update prompts
HTTP Client	Axios	`1.9.0`	REST API communication with request/response interceptors

Backend

Layer	Technology	Version	Responsibility
Server Framework	Express	`5.2.1`	REST API routing and middleware pipeline
Database & ODM	MongoDB / Mongoose	`9.4.1`	Schematized document storage, aggregation pipelines
AI Integration	Mistral AI Official SDK	`2.2.0`	LLM-powered diagnostics via `mistral-small-latest`
Authentication	JSON Web Tokens (JWT)	`9.0.3`	Stateless authentication delivered via httpOnly cookies
Password Security	bcryptjs	`3.0.3`	Salted credential hashing (12 rounds)
Validation	express-validator	`7.3.2`	Input sanitization and payload schema enforcement
Rate Limiting	express-rate-limit	`8.3.2`	Brute-force and DoS attack prevention
Logging	Winston & Morgan	`3.19.0` / `1.10.1`	Structured file/console logging and HTTP telemetry

⚙️ Engineering Highlights

Controller-Service Architecture — The Express backend enforces a strict two-layer separation: controllers handle HTTP concerns (request parsing, response formatting) while services encapsulate all business logic and database interaction. This pattern enables independent testability and maintainability across 8 domain modules.
7-Slice Redux State Architecture — Frontend state is partitioned into domain-specific slices (auth, tracker, practice, analytics, planner, gamification, prediction), each with scoped async thunks. This prevents cross-domain coupling and keeps the component layer stateless.
Zero-Data-Loss Offline Resilience — MCQ exam sessions auto-persist to IndexedDB via Dexie. On reconnection, a background sync service drains the outbox queue and reconciles results with the server — including XP and streak recalculation — without user intervention.
AI Response Caching Strategy — Mistral API calls are guarded by a dual-condition cache (24-hour TTL and fewer than 25 new MCQs solved), stored directly on the User document. This eliminates redundant LLM calls while keeping recommendations fresh relative to actual study velocity.
Deterministic AI Fallback — If the Mistral API key is absent or the service is unreachable, the analytics pipeline transparently serves curated pedagogical suggestions from a rule-based constant engine. The frontend receives a consistent response shape regardless of the suggestion source.
Adaptive Rank Modeling — The Pro V2 Rank Predictor uses cubic accuracy scaling combined with volume, consistency, and weak-subject penalty coefficients to model candidate readiness non-linearly, avoiding the overconfidence of linear rank projections.
Global Question Bank Aggregation — Each imported MCQ test performs a compound-indexed upsert ({ subject, question }) into a platform-wide QuestionBank collection, eliminating duplicates while continuously expanding shared question coverage.
Graceful Process Management — The server handles SIGTERM, SIGINT, unhandledRejection, and uncaughtException signals, cleanly closing database connections before exit — production-safe behavior by design.

🧠 AI Integration — Mistral Diagnostic Engine

The Mistral AI integration is a core functional component of PrepZone, not a cosmetic feature. The diagnostic pipeline continuously monitors a candidate's accuracy across 11 JECA subjects and generates targeted, subject-specific study recommendations using mistral-small-latest.

Diagnostic Logic Flow:

On each analytics request, the service evaluates the candidate's per-subject accuracy from aggregated Tracker data.
Subjects below a 60% accuracy threshold are identified as weak topics and forwarded to the Mistral prompt payload — keeping token usage minimal by excluding already-mastered subjects.
The model is invoked with temperature: 0.2 and maxTokens: 350, enforcing structured JSON output and predictable pedagogical responses.
Generated suggestions are cached on the User document (aiCache) with a timestamp and MCQ count at generation time. Subsequent requests within the cache validity window are served from MongoDB — no repeated API calls.
If the Mistral service is unavailable, the system falls back to a deterministic rule engine with curated subject-specific guidance.

sequenceDiagram
    autonumber
    actor User as Candidate
    participant UI as Analytics UI
    participant Service as Analytics Service
    participant Cache as User.aiCache (MongoDB)
    participant Mistral as Mistral AI API

    User->>UI: Navigate to AI Insights
    UI->>Service: GET /api/analytics/suggestions
    Service->>Cache: Check aiCache (generatedAt & totalMCQsAtGen)

    alt Cache is Valid (< 24h & < 25 new MCQs)
        Cache-->>Service: Return cached suggestions
        Service-->>UI: { suggestions, source: 'mistral-ai', isCached: true }
    else Cache Expired or Invalid
        Service->>Service: Filter weak subjects (< 60% accuracy)
        alt Mistral API Key Available
            Service->>Mistral: Chat completion (mistral-small-latest)
            Mistral-->>Service: Structured JSON suggestion array
            Service->>Cache: Persist suggestions & update timestamp
            Service-->>UI: { suggestions, source: 'mistral-ai', isCached: false }
        else No API Key / Service Error
            Service->>Service: Generate rule-based suggestions from constants
            Service-->>UI: { suggestions, source: 'rule-based' }
        end
    end

📊 Data & Analytics

PrepZone's analytics layer goes significantly beyond basic CRUD dashboards. All metrics are derived from MongoDB aggregation pipelines computed at query time, ensuring accuracy without denormalized or duplicated data.

What the system tracks and surfaces:

Subject-level accuracy distributions across 11 CS subjects — identifying weak areas with statistical precision rather than self-reported guesses.
Daily MCQ goal adherence — each Tracker document records granular study sessions within the day, enabling both intra-day breakdowns and weekly trend analysis.
365-day consistency heatmap — maps annual preparation density on a 5-tier intensity scale (0 MCQs → <20 → <50 → <100 → 100+), rendered in a GitHub-style contribution grid via Recharts.
Mistake frequency analytics — aggregates error distribution by subject and topic, surfacing repeat mistakes with repeatCount tracking to prioritize the revision queue intelligently.
Study session telemetry — each session records totalMCQs, accuracy, timeSpent, and weakTopics, creating a longitudinal record of preparation velocity over the full exam cycle.
XP and progression metrics — the gamification layer contributes engagement data (streaks, level progression, XP velocity) that correlates study consistency with academic performance indicators.

🔒 Security & Reliability

Security decisions in PrepZone reflect production-oriented engineering, not checkbox compliance.

JWT via httpOnly Cookies — Authentication tokens are issued with httpOnly: true, sameSite: 'strict', and secure: true in production. This eliminates the attack surface for XSS-based token theft that affects localStorage-based auth schemes.
Strict CORS Policy — Requests are validated against an explicit origin whitelist (CLIENT_URL). Cross-origin requests from unauthorized origins are rejected at the middleware layer before reaching any route handler.
Dual-Layer Rate Limiting:
- Auth routes (/register, /login): max 20 requests per 15 minutes per IP.
- General API routes (/api/*): max 200 requests per 15 minutes per IP.
Input Validation & Sanitization — All request bodies are validated via express-validator schema definitions before reaching the service layer, rejecting malformed or malicious payloads early.
Graceful Process Shutdown — The server registers handlers for SIGTERM, SIGINT, unhandledRejection, and uncaughtException, ensuring database connections close cleanly and no requests are dropped mid-flight.
Structured Logging — Winston provides multi-transport structured logging (file + console) across all environments; Morgan logs HTTP telemetry per request. This enables production observability without external tooling.

🔬 Core Subsystems — Technical Detail

1. The Scholar's Journey — Gamification & XP Engine

PrepZone converts preparation into an RPG-like progression model. Every study session, practice drill, and mock test rewards user effort through a well-defined XP formula:

Base XP = (MCQs Completed × 10) + (Minutes Studied × 1)
Accuracy Multiplier: 1.2× if session accuracy ≥ 80%; 1.0× otherwise
XP Earned = round(Base XP × Multiplier)
Level Progression: User advances to level L+1 when cumulative XP reaches L × 1000

Level Range	Rank Designation	Visual Accent
Levels 1–5	Bronze Novice	`#cd7f32`
Levels 6–15	Silver Practitioner	`#c0c0c0`
Levels 16–30	Gold Specialist	`#ffd700`
Levels 31–45	Platinum Expert	`#e5e4e2`
Levels 46–60	Diamond Master	`#b9f2ff`
Levels 61–80	Master Scholar	`#ff00ff`
Levels 81+	Grandmaster	`#ff4500`

Streak Calculation: Daily activity timestamps are evaluated server-side. If the elapsed time since lastActiveDate falls within the configured grace window (1 day), the streak increments; otherwise it resets to 1.

2. WBJECA Rank Predictor — Pro V2

The Pro V2 Predictor simulates candidate standing across 25,000 yearly WBJECA candidates using a non-linear composite readiness model:

Accuracy Score = (Avg Accuracy / 100)³ × 70 — Cubic scaling penalizes low-accuracy candidates disproportionately.
Volume Score = min(Total MCQs / 5000, 1) × 20
Consistency Score = min(Current Streak / 15, 1) × 10
Weak Subject Penalty = max(0.75, 1 − (Weak Subject Count × 0.05))
Readiness (%) = round((Accuracy + Volume + Consistency) × Weak Penalty)
Predicted GMR = max(1, min(25000, round(25000 × (1 − Readiness / 100)^2.2)))

The predicted GMR is evaluated against 14 pre-seeded West Bengal MCA institutions across 4 tiers:

🥇 Top Tier — Jadavpur University (JU), MAKAUT Campus, Calcutta University (CU)
🥈 Good Govt — Kalyani University, KGEC, JGEC
🥉 Private (High ROI) — IEM, UEM, Heritage (HIT), RCCIIT
🧪 Backup Institutions — Techno Main Salt Lake, Haldia, NSEC, Techno Hooghly

If the candidate's GMR falls outside a college's standard cutoff, a quadratic proximity model computes marginal admission chances (2%–25%) rather than returning a flat failure result, and advises the precise rank improvement required.

3. MCQ Master & Offline Practice Engine

Global Question Bank Aggregation — On test import, questions are upserted via bulkWrite into a shared QuestionBank collection using a compound unique index { subject, question }. This eliminates duplicates and continuously expands platform-wide coverage from every user's imported content.
Custom Code Syntax Highlighting — A zero-dependency custom highlighter parses C/C++, Java, and Python code snippets inline within MCQ cards, with line numbers.
Double-Storage Sync — Test submission atomically updates the PracticeResult record and upserts the user's daily Tracker log — awarding XP and updating streaks — without requiring separate manual entry.
Offline Practice Resilience — In-progress sessions are auto-saved to IndexedDB. On submission, if the network is unavailable, results are queued in a syncOutbox. The background sync service drains this queue upon reconnection and reconciles all data with the server.

Offline Sync Workflow:

sequenceDiagram
    autonumber
    actor User as Candidate
    participant Player as Practice Player (React)
    participant IDB as IndexedDB (Dexie)
    participant SyncSvc as syncService.js
    participant API as Express API
    participant DB as MongoDB Atlas

    User->>Player: Start MCQ Test
    Player->>IDB: Cache active progress & timer
    User->>Player: Answers Questions & Clicks Submit

    alt Network Online
        Player->>API: POST /api/practice/:id/submit
        API->>DB: Save PracticeResult & Upsert Tracker Entry (XP + Streak)
        API-->>Player: Return 200 OK & Updated Scores
        Player->>User: Render Results
    else Network Offline
        Player->>IDB: Push result to syncOutbox queue
        Player->>User: Render Results + "Offline: Saved Locally"
        Note over Player,SyncSvc: When device regains connection:
        SyncSvc->>IDB: Read pending syncOutbox items
        SyncSvc->>API: POST /api/practice/:id/submit
        API->>DB: Sync results & update User XP
        SyncSvc->>IDB: Remove item from syncOutbox
        SyncSvc->>Player: Trigger refetch & update sync badge
    end

4. Roadmap Architect & Multi-Plan Study Planner

PrepZone supports multiple concurrent, independent study roadmaps (e.g., Core Syllabus, 30-Day Revision Crash Course, Target JU Sprint).

AI Prompt Architect — Generates structured JSON schema prompts pre-populated with start dates, target subjects, syllabus focus areas, and daily MCQ quotas, ready to paste into any LLM.
Dual View Mode — Toggle between a sequential timeline view and an interactive monthly calendar heatmap per roadmap.
Dashboard Integration — The active primary roadmap injects "Today's Mission" onto the main dashboard with a live progress bar comparing daily solved MCQs against the day's target.

5. Mistake Bank — Active Recall & Revision Queue

Mistake Bank — Stores detailed problem context, user misconceptions, and corrected explanations with repeatCount tracking and subject-level error frequency aggregation.
Revision Queue — A prioritized checklist with pending / completed status toggles and priority weights (high, medium, low), enabling targeted review sessions around exam dates.

6. Analytics & Consistency Heatmap

MongoDB Aggregation Pipelines — Compute real-time totals (MCQs, study hours, accuracy, subject distributions) without pre-stored or duplicated metrics.
365-Day Activity Heatmap — Annual preparation density mapped on a 5-tier intensity scale (0 = 0 MCQs, 1 = <20, 2 = <50, 3 = <100, 4 = 100+), rendered via Recharts.

📡 REST API Reference

All protected endpoints require a valid JWT via the HTTP-Only token cookie or the Authorization: Bearer <token> header.

🔐 Authentication (`/api/auth`)

Method	Endpoint	Auth	Purpose
`POST`	`/api/auth/register`	Public (Rate Limited)	Register a new user and issue JWT cookie
`POST`	`/api/auth/login`	Public (Rate Limited)	Authenticate user and issue JWT cookie
`POST`	`/api/auth/logout`	Public	Clear JWT auth cookie
`GET`	`/api/auth/me`	Authenticated	Retrieve authenticated user profile
`PATCH`	`/api/auth/goal`	Authenticated	Update user's daily MCQ target
`PATCH`	`/api/auth/target-colleges`	Authenticated	Update pinned target colleges list

📊 Tracker & Logging (`/api/tracker`)

Method	Endpoint	Auth	Purpose
`POST`	`/api/tracker`	Authenticated	Create or merge a daily study session
`GET`	`/api/tracker`	Authenticated	Get paginated log history (`limit`, `skip`)
`GET`	`/api/tracker/:date`	Authenticated	Get specific log by `YYYY-MM-DD`
`DELETE`	`/api/tracker/:date`	Authenticated	Delete a daily log entry

📈 Analytics & AI Diagnostics (`/api/analytics`)

Method	Endpoint	Auth	Purpose
`GET`	`/api/analytics/summary`	Authenticated	Aggregate career stats, accuracy, weak topics
`GET`	`/api/analytics/weekly`	Authenticated	Daily breakdown for last N days (`?days=7`)
`GET`	`/api/analytics/suggestions`	Authenticated	Cached or fresh Mistral AI / rule-based suggestions
`GET`	`/api/analytics/calendar`	Authenticated	365-day activity heatmap (`?year=YYYY`)

🧪 MCQ Practice & Question Bank (`/api/practice`)

Method	Endpoint	Auth	Purpose
`POST`	`/api/practice/import`	Authenticated	Import custom MCQ test & update QuestionBank
`GET`	`/api/practice/my-tests`	Authenticated	List all tests created by the user
`GET`	`/api/practice/:id`	Authenticated	Retrieve complete test with questions
`GET`	`/api/practice/:id/latest-result`	Authenticated	Retrieve most recent attempt result
`POST`	`/api/practice/:id/submit`	Authenticated	Submit attempt, record score, sync with Tracker

🗺️ Study Planner (`/api/planner`)

Method	Endpoint	Auth	Purpose
`GET`	`/api/planner/list`	Authenticated	List all user roadmaps
`GET`	`/api/planner`	Authenticated	Retrieve active primary roadmap
`GET`	`/api/planner/:id`	Authenticated	Retrieve specific roadmap by ID
`POST`	`/api/planner`	Authenticated	Create a new roadmap
`PUT`	`/api/planner/:id`	Authenticated	Update roadmap schedule
`PATCH`	`/api/planner/:id/active`	Authenticated	Set a roadmap as the active primary
`DELETE`	`/api/planner/:id`	Authenticated	Delete a specific roadmap

📓 Mistakes & Revision (`/api/mistakes`, `/api/revision`)

Method	Endpoint	Auth	Purpose
`POST`	`/api/mistakes`	Authenticated	Log a new concept mistake
`GET`	`/api/mistakes`	Authenticated	Get mistakes (`?subject=OS&page=1&limit=20`)
`GET`	`/api/mistakes/analytics`	Authenticated	Subject-wise error frequency metrics
`PATCH`	`/api/mistakes/:id`	Authenticated	Update mistake or correction
`DELETE`	`/api/mistakes/:id`	Authenticated	Delete a mistake record
`POST`	`/api/revision`	Authenticated	Add topic to revision queue
`GET`	`/api/revision`	Authenticated	List revision queue items
`GET`	`/api/revision/stats`	Authenticated	Priority & completion status counts
`PATCH`	`/api/revision/:id`	Authenticated	Toggle completed status or update details
`DELETE`	`/api/revision/:id`	Authenticated	Remove item from revision queue

🏛️ Colleges & System

Method	Endpoint	Auth	Purpose
`GET`	`/api/colleges`	Public	List all seeded West Bengal MCA colleges
`GET`	`/health`	Public	Server health status, environment, and uptime

🗄️ Database Schema Design

User

Core document storing authentication credentials, gamification state, AI cache, and preferences. Key fields: email (unique, indexed), streak subdocument (current, longest, lastActiveDate), xp, level, dailyMCQGoal, aiCache (suggestion array + timestamp + MCQ count at generation), and targetColleges.

Tracker

Per-user, per-day study log with compound unique index { userId, date }. Each document maps a dynamic subjects object ({ total, correct, accuracy } per subject), totalMCQs, accuracy, timeSpent, weakTopics, and a granular sessions array for intra-day breakdowns.

Practice — PracticeTest, QuestionBank, PracticeResult

PracticeTest — User-imported test with full question array including codeSnippet, options, and answer.
QuestionBank — Platform-wide aggregated repository with compound unique index { subject, question } preventing duplicate ingestion.
PracticeResult — Immutable attempt history including per-question userAnswers, score, accuracy, and timeTaken.

Planner

Multi-roadmap document with compound unique index { userId, title }. Each roadmap contains a plans array ({ date, subject, topics, mcqTarget, status }) and an isActive flag that drives the dashboard mission widget.

Mistake & Revision

Mistake — subject, topic, mistake description, correction concept, tags, repeatCount. Compound index { userId, subject }.
Revision — topic, subject, priority (high | medium | low), status (pending | completed), dueDate, completedAt.

College

Seeded reference collection for 14 West Bengal MCA institutions: name (unique), location, type (Govt | Govt Aided | Private), tier, cutoff (GMR upper threshold), minCutoff (GMR lower threshold).

🚀 Deployment & Infrastructure

PrepZone is deployed across a split-host production architecture:

Layer	Platform	Details
API Backend	Render	Node.js environment, production start command
Frontend	DigitalOcean VPS	Static Vite production build served via Nginx
Reverse Proxy	Nginx	SPA routing rewrite, API traffic forwarding, `X-Forwarded-Proto` headers
Database	MongoDB Atlas	Cloud-hosted, connection-pooled via Mongoose
AI Services	Mistral AI API	External LLM call from backend service layer

The Express API is pre-configured with trust proxy: 1, ensuring accurate IP detection for rate limiting behind Nginx. Nginx handles SPA fallback routing (all non-static paths rewrite to index.html) and forwards API requests to the Express process. The frontend build outputs a fully optimized dist/ bundle via Vite's production pipeline.

✅ Implementation Status

Shipped & Verified

Potential Future Direction

Collaborative Peer Mock Battles — Real-time 1v1 MCQ duels via WebSockets
PDF PYQ Ingestion — Automated parsing of past JECA papers via OCR
Push Notification Reminders — Native Web Push for revision items and streak preservation
Syllabus Mastery Radar Chart — Per-topic mastery visualization against state benchmarks

📄 License & Acknowledgments

PrepZone is distributed under the MIT License.

Designed & Built by: Ramiz
Target Curriculum: West Bengal Joint Entrance Examination Board (WBJEEB) — MCA Entrance (WBJECA)
AI Diagnostics: Mistral AI (mistral-small-latest)
Open-Source Foundations: React, Vite, Node.js, Express, MongoDB / Mongoose communities

Name		Name	Last commit message	Last commit date
Backend		Backend
Frontend		Frontend
README.md		README.md

Latest commit

History

Folders and files

Repository files navigation

🛡️ PrepZone — AI-Powered JECA Preparation & Learning Ecosystem

📑 Table of Contents

🌟 Overview

✨ Core Capabilities

🏗️ Technical Architecture

Frontend

Backend

Database

Project Directory Structure

🛠️ Technology Stack

Frontend

Backend

⚙️ Engineering Highlights

🧠 AI Integration — Mistral Diagnostic Engine

📊 Data & Analytics

🔒 Security & Reliability

🔬 Core Subsystems — Technical Detail

1. The Scholar's Journey — Gamification & XP Engine

2. WBJECA Rank Predictor — Pro V2

3. MCQ Master & Offline Practice Engine

4. Roadmap Architect & Multi-Plan Study Planner

5. Mistake Bank — Active Recall & Revision Queue

6. Analytics & Consistency Heatmap

📡 REST API Reference

🔐 Authentication (/api/auth)

📊 Tracker & Logging (/api/tracker)

📈 Analytics & AI Diagnostics (/api/analytics)

🧪 MCQ Practice & Question Bank (/api/practice)

🗺️ Study Planner (/api/planner)

📓 Mistakes & Revision (/api/mistakes, /api/revision)

🏛️ Colleges & System

🗄️ Database Schema Design

User

Tracker

Practice — PracticeTest, QuestionBank, PracticeResult

Planner

Mistake & Revision

College

🚀 Deployment & Infrastructure

✅ Implementation Status

Shipped & Verified

Potential Future Direction

📄 License & Acknowledgments

About

Topics

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages

🔐 Authentication (`/api/auth`)

📊 Tracker & Logging (`/api/tracker`)

📈 Analytics & AI Diagnostics (`/api/analytics`)

🧪 MCQ Practice & Question Bank (`/api/practice`)

🗺️ Study Planner (`/api/planner`)

📓 Mistakes & Revision (`/api/mistakes`, `/api/revision`)