Enterprise Operations Hub (MCP Server)

An MCP-native enterprise operations platform that unifies team management, product catalog, order processing, knowledge base, location intelligence, analytics, and DevOps monitoring into a single server — all accessible from any MCP-compatible AI agent or client.

With 15 tools, 7 resources (4 data resources + 3 interactive visual widget resources), and 3 prompts, agents can search users, browse product catalogs, manage orders, generate reports, explore points of interest on interactive maps, and monitor system health — without switching contexts or integrating multiple APIs.

Use Cases

🤖 AI-Powered Enterprise Assistant

Equip your AI agent with direct access to company data. Ask natural language questions like:

• "Find all developers in the Engineering team" → search-users tool with role filter and visual user cards

• "Show me shipped orders from this week" → list-orders tool with paginated table and status badges

• "What's our top-selling product?" → get-product-details across the catalog with ratings and inventory data

• "Generate a sales report for Q4" → generate-report with revenue, conversion, and customer metrics

🗺️ Location-Based Decision Making

The get-location-info tool + location-map widget delivers an interactive point-of-interest explorer:

• "Find restaurants near the conference venue" → visual map with colored pins by POI type

• "Are there any parks close to the hotel?" → filtered by park type, with ratings and addresses

• Filter by restaurant, park, museum, café, hotel, shopping, or landmark — adjust radius from 0.01 to 50 km

📋 Workflow Automation with Approvals

The server supports human-in-the-loop elicitation workflows:

• Approval gates — request-approval tool presents confirmation dialogs for sensitive actions (deploy, delete, grant access)

• Feedback collection — collect-feedback tool gathers structured ratings and comments from users

• Text summarization — summarize-text delegates to the client's LLM for on-demand summarization via the sampling protocol

📊 Real-Time Operations Dashboard

Monitor system health through the agent:

• get-server-status — uptime, request volume, active connections, CPU/memory usage

• list-client-capabilities — introspect what the connected client supports (roots, sampling, elicitation, apps)

• get-user-context — retrieve the current user's locale, timezone, and location for personalized responses

📚 Knowledge Management

• search-knowledge — full-text search across the knowledge base (account, API, and general categories)

• create-document — create reports, memos, guides, or specs with tag-based categorization

• Resources provide structured reference data: POI type catalog, server configuration, API docs, dataset statistics, and interactive visual widgets

Related MCP server: MCP Business AI Transformation

Capabilities at a Glance

Tools (15)

Resources (7)

Resources are the backbone of data access in MCP. The server exposes 4 static data resources and 3 interactive widget resources that render rich visual experiences in MCP-compatible clients.

Static Data Resources

Visual Widget Resources

Widgets are a special type of resource following the MCP ui:// URI scheme. Unlike static resources (data://, config://, docs://) which return raw content, widget resources deliver interactive React-based UIs rendered directly in the MCP client. Each widget is also associated with a tool that populates it with dynamic data.

Prompts (3)

Quick Start

npm install
npm run dev        # HTTP mode with hot reload + inspector + widgets

Open http://localhost:8760/inspector to explore all tools, resources, and widgets interactively.

Multi-Worker Deployment

When you need to handle higher concurrency or isolate workloads, launch multiple server instances with a single command:

# Start 3 workers on ports 8760, 8759, 8758
npm run start:multi -- --workers 3

# Shorthand with -w flag
npm run start:multi -- -w 5

Each worker runs as an independent process with its own port, decrementing from 8760. Press Ctrl+C to gracefully shut down all workers.

Use cases:

• Load distribution — spread incoming MCP client connections across multiple instances behind a reverse proxy (nginx, HAProxy)

• Workload isolation — dedicated workers for different agent teams or environments (staging vs. production)

• Graceful rolling restarts — restart workers one at a time without dropping connections

Client Configuration

HTTP mode

{
  "servers": {
    "enterprise-ops-hub": {
      "url": "http://localhost:8760/mcp",
      "type": "http"
    }
  }
}

stdio mode

Widget resources (ui://widget/*) are HTTP-only (require a browser to render). All other capabilities — tools, static resources, prompts, elicitation, sampling, roots, progress, logging — are fully available over stdio.

{
  "servers": {
    "enterprise-ops-hub": {
      "command": "npx",
      "args": ["tsx", "src/stdio.ts"],
      "cwd": "/path/to/mock_mcp"
    }
  }
}

Or with the built output:

{
  "command": "node",
  "args": ["dist/src/stdio.js"],
  "cwd": "/path/to/mock_mcp"
}

Scripts