Nutanix MCP Server

[!WARNING] Use at your own risk. MCP servers grant AI models the ability to execute actions against your infrastructure. AI-driven management of production environments carries inherent risk — models can misinterpret intent, hallucinate parameters, or trigger destructive operations. This software is provided "as is" without warranty of any kind. The authors accept no liability for data loss, downtime, or any damages arising from use of this tool. Always review AI-proposed actions before execution and maintain proper backups.

An MCP (Model Context Protocol) server that exposes Nutanix Prism Central and Prism Element APIs as tools for AI assistants like GitHub Copilot, Claude, and others.

Features

• 65 tools — Full coverage of Prism Central v4 and Prism Element v2 APIs
• Prism Central (v4 API) — VM lifecycle, snapshots, clusters, hosts, networking, categories, alerts, tasks
• Prism Element (v2 API) — Direct cluster access for storage, disks, data protection, system config, health checks
• AsBuilt Reports — Generate comprehensive HTML reports with interactive TOC, Mermaid topology diagrams, and print-to-PDF support
• API version routing — Prefers v4, falls back to v3/v2 when needed
• Async — Non-blocking HTTP client using httpx

Available Tools (65)

VM Management — Prism Central v4

VM Snapshots — Prism Central v4

Cluster & Host Management — Prism Central v4

Networking & Images — Prism Central v4

Categories & Tagging — Prism Central v4

Alerts & Tasks — Prism Central v4

Prism Element — Cluster & Hosts (v2 direct access)

Prism Element — Storage

Prism Element — VMs, Networks & Images

Prism Element — Data Protection

Prism Element — System Configuration

AsBuilt Reports

AsBuilt reports include 9 sections: overview, system, hosts (with per-host disk inventory), VMs, networks, storage, data protection (with remote sites and unprotected VM detection), alerts, and health checks. Hypervisor names are mapped automatically (kKvm → AHV). The HTML export features an interactive table of contents with scroll-spy that is hidden when printing to PDF.

MCP Resources (URI-based browsing)

The server exposes resources via nutanix:// URIs, allowing LLMs to browse entities without explicit tool calls:

MCP Prompts

Setup

Prerequisites

• Python 3.10+
• Network access to your Prism Central instance (port 9440)
• Nutanix credentials with API access

Install

cd mcp/nutanix-mcp-server
pip install -e .

Or with dev dependencies:

pip install -e ".[dev]"

Configure

Copy .env.example to .env and fill in your credentials:

cp .env.example .env

NUTANIX_HOST=your-prism-central.example.com
NUTANIX_PORT=9440
NUTANIX_USERNAME=your-username
NUTANIX_PASSWORD=your-password
NUTANIX_VERIFY_SSL=true
NUTANIX_TIMEOUT=30

Run

nutanix-mcp

Or directly:

python -m nutanix_mcp

MCP Client Configuration

This server uses stdio transport — it communicates via stdin/stdout. Each client configures a command to launch the server process.

Tip: Store credentials in environment variables or a .env file, never in config files committed to source control.

Claude Code (CLI)

Add the server to your project with the claude mcp add command:

claude mcp add nutanix -- python -m nutanix_mcp

Or manually create/edit .mcp.json in your project root:

{
  "mcpServers": {
    "nutanix": {
      "type": "stdio",
      "command": "python",
      "args": ["-m", "nutanix_mcp"],
      "cwd": "/path/to/mcp/nutanix-mcp-server",
      "env": {
        "NUTANIX_HOST": "your-prism-central.example.com",
        "NUTANIX_USERNAME": "your-username",
        "NUTANIX_PASSWORD": "your-password",
        "NUTANIX_VERIFY_SSL": "true"
      }
    }
  }
}

For user-wide availability (all projects), add to ~/.claude.json instead.

Claude Desktop

Edit the config file at:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json
• Linux: ~/.config/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "nutanix": {
      "type": "stdio",
      "command": "python",
      "args": ["-m", "nutanix_mcp"],
      "cwd": "/path/to/mcp/nutanix-mcp-server",
      "env": {
        "NUTANIX_HOST": "your-prism-central.example.com",
        "NUTANIX_USERNAME": "your-username",
        "NUTANIX_PASSWORD": "your-password",
        "NUTANIX_VERIFY_SSL": "true"
      }
    }
  }
}

Restart Claude Desktop fully after editing.

GitHub Copilot (VS Code)

Add to .vscode/mcp.json in your workspace:

{
  "servers": {
    "nutanix": {
      "type": "stdio",
      "command": "python",
      "args": ["-m", "nutanix_mcp"],
      "cwd": "${workspaceFolder}/mcp/nutanix-mcp-server",
      "env": {
        "NUTANIX_HOST": "your-prism-central.example.com",
        "NUTANIX_USERNAME": "your-username",
        "NUTANIX_PASSWORD": "your-password",
        "NUTANIX_VERIFY_SSL": "true"
      }
    }
  }
}

OpenCode (sst/opencode)

Add to opencode.json (or opencode.jsonc) in your project root:

{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "nutanix": {
      "type": "local",
      "command": ["python", "-m", "nutanix_mcp"],
      "environment": {
        "NUTANIX_HOST": "your-prism-central.example.com",
        "NUTANIX_USERNAME": "your-username",
        "NUTANIX_PASSWORD": "your-password",
        "NUTANIX_VERIFY_SSL": "true"
      },
      "enabled": true
    }
  }
}

Note: OpenCode uses "command" as an array and "environment" instead of "env".

Docker MCP Gateway

The Docker MCP Gateway can proxy this server inside a container. Two approaches:

Option A: Run directly via Docker

Build a container image and reference it in your MCP client config:

FROM python:3.12-slim
WORKDIR /app
COPY mcp/nutanix-mcp-server/ .
RUN pip install --no-cache-dir -e .
CMD ["python", "-m", "nutanix_mcp"]

Then in any MCP client config:

{
  "mcpServers": {
    "nutanix": {
      "type": "stdio",
      "command": "docker",
      "args": [
        "run", "-i", "--rm",
        "-e", "NUTANIX_HOST=your-prism-central.example.com",
        "-e", "NUTANIX_USERNAME=your-username",
        "-e", "NUTANIX_PASSWORD=your-password",
        "-e", "NUTANIX_VERIFY_SSL=true",
        "nutanix-mcp-server"
      ]
    }
  }
}

Option B: Register with Docker MCP Gateway

If you have Docker Desktop with the MCP Toolkit:

docker mcp gateway run

Configure the gateway profile to include the nutanix server. The gateway then exposes all registered MCP servers as a single unified endpoint.

In your AI client, point to the gateway:

{
  "mcpServers": {
    "MCP_DOCKER": {
      "command": "docker",
      "args": ["mcp", "gateway", "run"]
    }
  }
}

The gateway handles routing, lifecycle management, and credential isolation.

API Version Strategy

Discovering Prism Element Hosts

Use list_clusters to find cluster UUIDs, then list_hosts to find CVM IPs. Those CVM IPs can be used as pe_host in the Prism Element tools.

Development

# Lint
ruff check src/

# Type check
mypy src/

# Test
pytest

References

• Nutanix v4 API Documentation
• Nutanix Developer Portal
• Prism Central v3 API
• Prism Element v2 API
• MCP Protocol Specification