ToolVault

AI-powered inventory management system built for construction operations.

---

Overview

ToolVault is a full-stack microservices application designed to streamline tool and equipment inventory management for construction companies. Built solo using an AI-first development approach with GitHub CoPilot, as well as Claude Code. ToolVault demonstrates production-pattern architecture across seven independent services — containerized with Docker and backed by persistent MySQL.

Key capabilities:

• JWT authentication with role-based access control
• Real-time asset check-in / check-out tracking
• AI-powered forecasting that predicts shortages and suggests reorder quantities
• Low stock alerting and transfer request management
• Persistent refresh token system with one-time-use rotation
• Swagger/OpenAPI documentation across all services

---

Architecture

React Frontend (Port 3000)
        │
        │ JWT Bearer Token
        ▼
┌─────────────────────────────────────────────┐
│              Spring Boot Services            │
├──────────────┬──────────────┬───────────────┤
│   Identity   │  Depot Ops   │  Warehouse    │
│   :8081      │  :8082       │  Ops :8083    │
│  Auth / JWT  │  Assets /    │  Stock /      │
│              │  Forecasting │  Inventory    │
├──────────────┼──────────────┼───────────────┤
│  Procurement │ Notification │  Reporting    │
│  :8084       │  :8086       │  :8085        │
└──────────────┴──────────────┴───────────────┘
        │
        │ HikariCP Connection Pool
        ▼
   MySQL 8.4 (Docker Volume — persistent)

Tech stack:

• Java 17 — Spring Boot 3.3.5
• Spring Security — JWT (HS256) + BCrypt
• Spring Data JPA — Hibernate 6
• MySQL 8.4 — H2 (local dev fallback)
• Docker + Docker Compose
• Maven multi-module build
• Swagger / OpenAPI 3 (Springdoc)
• React frontend (in progress)

---

Services

Service	Port	Responsibility
identity-service	8081	Auth, JWT issuance, refresh token persistence
depot-ops-service	8082	Asset tracking, checkout history, AI forecasting
warehouse-ops-service	8083	Stock levels, restock, allocation
procurement-service	8084	Purchase request management
notification-service	8086	Email, SMS, event alerting
reporting-service	8085	Analytics and summary reports

---

Getting Started

Prerequisites

• Java 17+
• Maven 3.9+
• Docker Desktop

Run with Docker

# Clone the repository
git clone https://github.com/Z4kM4rm3l/CEIS400-Group7Project.git
cd CEIS400-Group7Project

# Build all services
mvn clean package -DskipTests

# Start the full stack
docker-compose up --build

MySQL databases are created automatically on first run via scripts/init.sql.

Environment Variables

All secrets are injected via Docker Compose environment variables. For local development, services fall back to H2 in-memory databases — no configuration needed.

For production, set the following in your environment or .env file:

SPRING_DATASOURCE_URL=jdbc:mysql://...
SPRING_DATASOURCE_USERNAME=...
SPRING_DATASOURCE_PASSWORD=...
JWT_SECRET=...
JWT_ISSUER=...
JWT_ACCESS_EXP_MIN=15

---

API Highlights

Authentication

# Register
POST /auth/register
{ "firstName": "...", "lastName": "...", "email": "...", "password": "..." }

# Login
POST /auth/login
{ "email": "...", "password": "..." }

# Refresh token (one-time use rotation)
POST /auth/refresh
{ "refreshToken": "uuid-here" }

Forecasting

# Checkout frequency by tool model (last 30 days)
GET /forecasting/frequency

# Tools at risk of shortage
GET /forecasting/risk

# AI-generated reorder suggestions
GET /forecasting/reorder

Inventory

# List all stock
GET /warehouse/stock

# Top 5 SKUs by quantity
GET /warehouse/top-skus

# Check in / check out assets
POST /depot/check-in
POST /depot/check-out

---

Security

• Passwords hashed with BCrypt
• JWTs signed with HMAC-SHA256
• Refresh tokens stored in MySQL with expiry and one-time-use revocation
• Role-based route protection (USER, ADMIN)
• Custom 401/403 JSON error responses
• Stateless session management

---

Development Approach

ToolVault was designed with an AI-first development methodology mindset — treating GitHub CoPilot and Claude Code as primary development partners rather than autocomplete tools. Architectural decisions, service boundaries, security patterns, and business logic were directed by the developer; CoPilot & Code handled syntax generation and boilerplate.

This approach echoes the emerging discipline of agentic software development — where human judgment guides AI execution to ship production-quality systems faster.

---

Author

Zakary Marmel

• BAS Software Design — GPA 3.93, Dean's List
• AAS Information Technology & Networking — With Honors
• LinkedIn
• GitHub

---

Built with Java, Spring Boot, Docker, and AI-first development practices.