Skip to main content
Glama
log-wade

datacenter-mcp-server

by log-wade
README.md
# datacenter-mcp-server

Mission-critical data center engineering tools for AI agents. 8 professional-grade calculation engines covering cooling, power, GPU thermal optimization, UPS/battery sizing, tier classification, and commissioning workflows.

Built on the [Model Context Protocol (MCP)](https://modelcontextprotocol.io) standard — works with Claude Desktop, Cursor, Windsurf, Cline, and any MCP-compatible client.

## Quick Start

### Claude Desktop

Add to your `claude_desktop_config.json`:

```json
{
  "mcpServers": {
    "datacenter": {
      "command": "npx",
      "args": ["-y", "datacenter-mcp-server"]
    }
  }
}
```

### Cursor / Windsurf

Add to your MCP settings:

```json
{
  "datacenter": {
    "command": "npx",
    "args": ["-y", "datacenter-mcp-server"]
  }
}
```

### HTTP Mode (Remote)

```bash
TRANSPORT=http PORT=3001 API_KEY=your-secret npx datacenter-mcp-server
```

## Tools

### dc_calculate_cooling_load
Calculate ASHRAE-compliant cooling requirements for any data center facility. Inputs include IT load, redundancy level, climate zone, altitude, and humidity targets. Returns tonnage, airflow, chilled water capacity, and energy estimates.

### dc_analyze_power_redundancy
Analyze electrical distribution from utility feed through UPS, PDU, and rack-level power. Supports N, N+1, 2N, and 2N+1 redundancy configurations with efficiency and cost analysis.

### dc_assess_tier_classification
Evaluate facility design against Uptime Institute Tier I-IV standards. Identifies compliance gaps and provides actionable upgrade recommendations.

### dc_generate_commissioning_plan
Generate L1-L5 commissioning workflows with phase sequencing, test procedures, and documentation requirements per ASHRAE Guideline 0.

### dc_analyze_rack_density
Analyze rack power density configurations from 5 kW to 100+ kW per rack. Covers cooling strategy selection, weight loading, and power distribution.

### dc_gpu_cooling_optimizer
Optimize cooling infrastructure for GPU/AI workloads (H100, A100, H200, B200, GB200). Calculates thermal loads, recommends cooling strategies (air, rear-door, direct liquid, immersion), sizes CDUs, and projects energy savings.

### dc_ups_battery_sizing
Size UPS modules and battery systems with N/N+1/2N/2N+1 redundancy. Compares VRLA vs lithium-ion with 10-year TCO analysis, floor space estimates, and weight calculations.

### dc_reference_lookup
Query ASHRAE standards, Uptime Institute tier requirements, and industry best practices for data center design and operations.

## Engineering Standards

All calculations comply with:

- ASHRAE TC 9.9 thermal guidelines
- Uptime Institute Tier Standard topology
- NFPA 70 / NEC electrical code
- ASHRAE Guideline 0 commissioning

## Development

```bash
# Install dependencies
npm install

# Build
npm run build

# Run tests (264 tests across 7 suites)
npm test

# Start in stdio mode (MCP default)
npm start

# Start in HTTP mode
npm run start:http

# Inspect with MCP Inspector
npm run inspect
```

## Architecture

```
src/
  index.ts          # MCP server entry point, tool registration
  types.ts          # TypeScript interfaces for all tools
  constants.ts      # Engineering constants and reference data
  middleware.ts      # Express middleware (auth, rate limiting, CORS)
  schemas/          # Zod validation schemas per tool
  services/         # Calculation engines per tool
tests/              # Jest test suites (264 tests)
```

## License

MIT - NextGen Mission Critical

TDQS

A4/5.0

Scored across 8 tools

Disambiguation5/5

Each tool addresses a distinct data center engineering concern—cooling, power redundancy, tier assessment, commissioning, rack density, GPU cooling, UPS sizing, and reference lookup. There is no semantic overlap that would cause an agent to select the wrong tool.

Naming Consistency2/5

Most tools follow a verb_noun pattern (dc_calculate_cooling_load, dc_analyze_power_redundancy, dc_generate_commissioning_plan), but three deviate: dc_gpu_cooling_optimizer, dc_ups_battery_sizing, and dc_reference_lookup use noun-based phrasing. This mixed convention undermines predictability.

Tool Count5/5

Eight tools is a well-scoped count for a data center engineering server. Each tool adds meaningful capability, and the set is neither bloated nor too sparse.

Completeness4/5

The server covers core design and analysis tasks including cooling, power, redundancy, tier classification, commissioning, and GPU-specific cooling. Minor gaps exist (e.g., no energy cost or physical security tools), but the domain is well covered.

Maintenance

ActivityStale
ResponsivenessNo issues