Deploy Agentic AI with MCP Server for Nutanix Cloud Platform

We are thrilled to announce the technology preview of the MCP server for Nutanix Cloud Platform, a powerful new tool designed to securely and efficiently bridge the gap between AI/LLMs and your Nutanix infrastructure. Model Context Protocol (MCP) is an open-source standard for connecting AI assistants to the point-of-record systems where data lives, including content repositories, business tools, and development environments.

As organizations increasingly adopt AI-powered applications, the MCP server for Nutanix Cloud Platform provides a production-ready way for AI assistants such as GitHub Copilot, the Claude AI assistant by Anthropic, Cursor, Postman agents, and Custom Datacenter Management agents to directly interact with Nutanix's V4 APIs.

Transformative Use Cases

MCP server for Nutanix Cloud Platform allows infrastructure admins to communicate with the infrastructure the way they want while reducing the friction to get to the how.. Here are the key use cases you can unlock with the technology preview of this release:

1. Creation of Automation Scripts with Coding Agents

By integrating the MCP server into Agentic AI systems such as Cursor, Claude Code or VS Code, admin and IT operators can now either directly perform operations, build automation scripts or build custom agents that can orchestrate Nutanix’s V4 APIs. The MCP server exposes the following 4  tools:

These tools help coding agents and LLMs retrieve per-operation code samples and full parameter schemas to generate accurate automation scripts for any V4 API operation in Nutanix SDK-supported languages (Python, Go, JavaScript, curl, PowerShell, C#) as well as any REST/JSON-compatible language or tool including Ansible, Terraform, and others. The V4 API supports OData filtering, sorting, and field selection; idempotency via request IDs; and ETag-based conditional updates all of which are provisioned through the MCP with per-operation field lists and usage constraints. Each operation includes structured annotations covering required parameters, valid filter/select fields, write workflow guidance, and immutable fields, enabling developers to generate correct scripts without consulting external documentation.

2. Simplified Migration to V4 APIs

With legacy Nutanix APIs (v0.8, v1, v2, and v3) planned for deprecation, migrating to the V4 API family is critical. The MCP server for Nutanix API acts as a migration assistant. Users can simply provide an existing script that is based on V3 API and prompt the agent to convert it to V4 API based script. 

3. Configuring Precise IAM Permissions and Custom Roles

You can prompt the coding agent to determine the exact permissions required to run a specific workflow or code block. This allows you to accurately construct your own custom roles and Access Control Policies in Identity and Access Management (IAM), tailoring permissions for your service accounts. 

4. Create your own custom agents to display custom UI to see the status of your datacenter

IT operators/Admins can leverage the MCP server to build their own specialized custom AI agents that can build your datacenter and monitor it by querying their infrastructure's health/capacity, and system performance. By executing operations such as listing virtual machines or retrieving entity-specific statistics, your custom datacenter management agents can dynamically render custom UI dashboards that reflect the real-time status of your Nutanix environment.

Architecture: Secure Execution Path to the V4 API Gateway

Designed to allow autonomous AI agents to operate safely and accountably within your environment, the MCP server acts as a streamlined passthrough. All execution, governance, and security controls are robustly handled by the Nutanix Prism V4 API Gateway.

Agentic AI Ready: Built on strong foundation of Highly available cluster OS, Distributed database platform and Enterprise API Innovation

The MCP Server helps AI agents build a semantic layer on the underlying hybrid cloud datacenters. The semantic layer helps with reasoning. If your underlying platform does not present a consistent state, the MCP server by itself will not add much value. Further, if your underlying platform does not provide a governance layer, then the onus to provide that governance will be on agents which will potentially compromise the security.  This MCP Server extends the reliable, scalable and secure Nutanix Cloud platform that consists of the following key architectural elements:

• Distributed and Reliable State management: One Distributed database which is consistent across the cloud. Single Entity State is stored in a Distributed Database that is consistently stored for the entire cloud, across distributed datacenters. This helps AI agents build that semantic layer and traverse the entire topology of the Hybrid cloud even if it is highly distributed across multiple datacenters comprising robo/edge, core dc, NC2, AI factory deployments and carries both Kubernetes and VM workloads.
• Highly available Cluster OS: This helps the Agents detect errors, fix them and resume/recover.
• V4 API built for Agentic era: Nutanix has spent 4 years building secure, scalable, and performant V4 APIs to create a platform that is fully ready for Agentic AI. The V4API platform is built on a strong entity model and groups operations in logical namespaces. The MCP server is a passthrough that streamlines the agent requests to V4 API Gateway where most of the heavy lifting that is required for  scale, security, performance, session management happens.
• Fine-Grained RBAC at Every API Level: Nutanix Prism features precise Role-Based Access Control (RBAC) at the individual API operation level. Agents can only execute calls to the specific APIs they have been given explicit access to, with the V4 API Gateway strictly enforcing these decisions.
• Service Accounts for Least Privilege: AI agents operate using dedicated service accounts rather than standard user accounts. This grants agents only a minimal set of permissions, enabling tighter access control, operating under the principle of least privilege, and helping to prevent the unintended elevation of privileges to a logged-in admin user.
• Throttling: Safeguards at the API Gateway are designed to prevent agent swarm attacks, helping to prevent overactive AI agents from overwhelming your cluster resources.
• Metering: The API Gateway natively tracks usage to help you understand and measure exactly how much API usage is being driven by your AI agents.
• Auditing: Comprehensive audit trails are maintained at the gateway level, providing full visibility so you can easily trace back exactly which AI agent made a specific API call.

Download and Deploy

You can download and deploy the technology preview of the MCP server from developers.nutanix.com. You can deploy it on your desktop directly from the IDE of your choice: Cursor, Claude code etc. The MCP server for Nutanix Cloud Platform features a low memory footprint of under 10MB and is designed to be hosted wherever it fits best in your workflow.

Nutanix Agentic AI Solution

In addition to the MCP server technology preview, Nutanix has announced Nutanix Agentic AI solution. Nutanix Agentic AI solution is a comprehensive full-stack software offering that establishes a cloud operating model for constructing and governing enterprise AI factories. In collaboration with the NVIDIA accelerated computing ecosystem, this solution streamlines operations while enabling maximized performance and robust security to efficiently optimize GPU resources and overall token expenditures. For more information on Nutanix Agentic AI solution please visit https://www.nutanix.com/solutions/ai

Thanks to the many contributors to this release:

• Nikhil Boba (Lead Developer)
• Keshav Jindal (Lead Developer)
• Vishal Srivastava (Dev Manager)
• Nitish Khune (Dev Manager)
• Akanksha Singh (Dev Manager)
• Devyani Kanada (Architect)
• Hitesh Menghnani (Product Management)
• Kostadis Roussos (Distinguished Engineer)
• Ranjit Sawant (Product Management)