Factory Supervisor MCP Server

An enterprise-grade Model Context Protocol (MCP) server built with NitroStack for real-time factory operations management, telemetry analysis, predictive maintenance, material forecasting, order estimation, and AI supervisor copilot capabilities.

---

🏭 Overview

The Factory Supervisor MCP Server connects AI assistants (such as Claude Desktop, Cursor, Antigravity, or custom MCP clients) directly to industrial shop floor telemetry, equipment diagnostics, inventory levels, maintenance schedules, and production orders.

With 10 modular domains and 16 specialized MCP tools, AI assistants can perform diagnostic checks, detect bottlenecks, forecast equipment failures, calculate risk exposure, and answer natural language questions about factory performance.

---

🚀 Implementation Roadmap (Phases 1 – 6 Complete)

• [x] Phase 1: Core System & Infrastructure Setup
  • Integrated NitroStack application structure (@McpApp, @Module).
  • Implemented health check and diagnostic ping tool (ping).
  • Designed mock telemetry dataset schemas for machines, inventory, orders, maintenance, schedule, and KPIs.
• [x] Phase 2: Domain Data Tools
  • Built core tools for Machines, Inventory, Production Orders, Maintenance Logs, Schedules, and KPIs.
  • Implemented schema validation with Zod and graceful fallback reading mechanisms.
• [x] Phase 3: Consolidated Factory Analytics
  • Built analyzeFactory tool consolidating data across all 6 factory domains.
  • Implemented automated machine severity scoring, inventory low-stock detection, numeric health scoring (0-100), and business impact estimation.
• [x] Phase 4: Predictive AI & Machine Intelligence
  • Developed predictive modeling tools for machine failure risk (predictMachineFailure), stockout forecasting (forecastInventory), order completion estimation (estimateOrderCompletion), bottleneck detection (detectProductionBottlenecks), and health score trends (predictFactoryHealth).
• [x] Phase 5: AI Supervisor Copilot & Executive Reporting
  • Implemented natural language routing copilot (factoryCopilot).
  • Created operational shift briefing tool (dailySupervisorBrief) and C-suite summary tool (executiveSummary).
• [x] Phase 6: Full Integration, Multi-Agent Skills & Production Readiness
  • Integrated all 10 modules into root AppModule.
  • Configured multi-agent developer skills (.agents/skills, .gemini/skills, .antigravity/skills, .claude/skills, .cursor/skills, .codex/skills, .copilot/skills, .opencode/skills).
  • Automated build script to sync mock data artifacts to production distributions (dist/mock).

---

🛠️ Modules & Tools Reference

Module	MCP Tool Name	Description	Key Inputs
Health	ping	System diagnostic health check ping.	None
Machines	listMachines	List all factory machines and operational state.	None
	getMachineStatus	Get detailed telemetry and job info for a specific machine.	machineId (string)
Inventory	getInventory	Get current stock levels for all parts.	None
	getLowStockItems	Retrieve inventory items at or below reorder threshold.	None
Orders	getOrders	Retrieve active factory production orders.	None
Maintenance	getMaintenanceLogs	Retrieve historical and active maintenance logs.	None
Schedule	getProductionSchedule	Fetch master production schedule.	None
KPI	getFactoryKPIs	Retrieve key operational performance indicators (OEE, Uptime).	None
Analytics	analyzeFactory	Consolidated factory report with health score, severity levels, risk assessment, and recommendations.	None
Prediction	predictMachineFailure	Predict failure probability, time window, and recommended actions based on vibration/temp telemetry.	machineId (string)
	forecastInventory	Predict stockout dates, days remaining, and recommended reorder quantities.	partId (optional string)
	estimateOrderCompletion	Estimate completion time, delay risk, and confidence score for an order.	orderId (string)
	detectProductionBottlenecks	Identify capacity, downtime, and material starvation bottlenecks.	None
	predictFactoryHealth	Forecast overall factory health score trends (Today vs. Tomorrow).	None
Assistant	factoryCopilot	AI Supervisor assistant routing natural language questions to appropriate analytics and predictive tools.	question (string)
	dailySupervisorBrief	Concise daily shift briefing highlighting top risks and priority actions.	None
	executiveSummary	High-level executive performance summary and revenue exposure risk.	None

---

📂 Project Architecture

my-mcp-server/
├── factory-supervisor-dashboard/  # Enterprise Next.js 14 Web Frontend
│   ├── app/                         # App Router Pages (Dashboard, Copilot, Machines, Inventory, Predictions, Timeline, Executive)
│   ├── components/                  # Modular React UI Components
│   ├── services/                    # Decoupled MCP API Service Layer
│   └── types/                       # TypeScript interfaces
├── src/
│   ├── index.ts                     # Server entry point & NitroStack application bootstrap
│   ├── app.module.ts                # Root AppModule importing all 10 feature modules
│   ├── mock/                        # Mock telemetry data JSON files
│   └── modules/                     # 10 MCP Feature Modules
├── package.json                     # Root build scripts
└── README.md                        # Project documentation

---

⚡ Quick Start

1. Prerequisites

• Node.js: v18.0.0 or higher
• npm: v9.0.0 or higher

2. Installation

Clone the repository and install dependencies:

git clone https://github.com/Heytish-V/factory-supervisor-mcp.git
cd factory-supervisor-mcp
npm install

3. Environment Configuration

Copy .env.example to .env:

cp .env.example .env

4. Running in Development Mode

Start the development server with hot-reloading:

npm run dev

5. Building & Running in Production Mode

Build the project (compiles TypeScript and syncs mock data to dist/mock):

npm run build
npm start

---

🔌 MCP Client Integration

Add this MCP server to your MCP client configuration (e.g. Claude Desktop, Cursor, or Antigravity):

Stdio Transport Config (Local Execution)

{
  "mcpServers": {
    "factory-supervisor": {
      "command": "node",
      "args": ["c:/Users/Heytish V/my-mcp-server/dist/index.js"]
    }
  }
}

---

💬 Example Natural Language Prompts

Once connected to an MCP client, try asking your AI assistant:

• "What is the current health score of the factory?"
• "Are any CNC machines at risk of failing in the next 24 hours?"
• "Will we be able to complete order ORD-2026-001 on time?"
• "Give me a daily supervisor briefing for today's shift."
• "Should we accept a new rush order right now?"

---

📄 License

Private / Enterprise Proprietary