Artificial intelligence news.
Straight from the original announcements.

Vultr Launches Read-Only MCP Server in Public Preview for Cloud Oversight

October 23, 2026, 19:54By the AI News Bites Editorial TeamSource: Vultr
Vultr Launches Read-Only MCP Server in Public Preview for Cloud Oversight
Image: Vultr

Vultr has released its MCP Server in public preview, a hosted, read-only tool that lets users ask questions about their cloud infrastructure in plain language. According to Vultr, the server can view resources and retrieve account information but cannot create, modify or delete anything, and it cannot spend money.

Key points
  • Vultr's MCP Server is available in public preview and is read-only.
  • It is hosted by Vultr, with setup done by adding one URL, signing in and approving access.
  • It cannot create, modify or delete resources, spend money, or reveal credentials.
  • Admins control access through Vultr IAM, and sign-in uses OAuth.
  • It works with clients such as Claude, Cursor, VS Code and Codex, with separate endpoints per product area.

What the Vultr MCP Server does

The Vultr MCP Server is built on the open MCP protocol, which lets users connect their chosen applications to external systems, data and tools. Vultr says the server gives agents access to resources such as compute, Kubernetes, networking, storage, databases, billing and the user account.

Vultr gives several examples of what users can ask: "Which instances are running in London, and on what plans?" or "Does my Kubernetes cluster have an upgrade available?" Users can also issue commands, such as asking for a summary of last month's invoice by product with options for cost savings.

Vultr hosts the server, so there is nothing to install. Setup takes three steps: add one URL to the client of your choice, sign in to Vultr, and approve access.

Read-only by design, with IAM controls

Vultr frames the read-only approach as a way to limit agent actions, which reduces side effects and risk, and spares users from building overly complex and granular approval layers. Requests to add servers or take any action beyond retrieving information will be denied. The server also cannot provide credentials or data that could be used to access customer environments, such as authentication tokens or passwords.

Access is governed by Vultr IAM, which includes a global on/off switch and individual permissions. Admins must first enable the server in Vultr IAM and can then grant agent access to specific users.

  • Sign-in uses the OAuth protocol with a user's Vultr account, with no API key to create or copy.
  • The server keeps no credentials of its own, and each sign-in is tied to one organization.
  • Every request acts as the signed-in user, so the server cannot see anything the user cannot.

Client support and flexible scope

Vultr says the server is designed for efficient tool selection and reduced token consumption. Each product area, such as compute, Kubernetes or databases, gets its own MCP endpoint. Users can therefore let the server reach their entire stack, a few products, or a single product area.

The server works with clients that support remote MCP connections through the standard protocol, including Claude, Cursor, VS Code and Codex. Remote and headless sessions, for example over the Secure Shell Protocol, are supported through device authorization. Users who work with two organizations can add the server twice under different names.

What it means for users

For teams already using AI clients, the preview offers a way to inspect infrastructure and billing without changing their architecture, while permissions stay under admin control. Because the server cannot change resources or spend money, it is positioned for visibility and insight rather than automation.

Vultr describes the current preview as "just the first step" and says more capabilities will arrive soon as it expands agent access. Documentation covers getting started, admin controls, tools and example prompts. The source does not state pricing or a general availability date.

Related news

Wiz Green Agent maps fixes for a critical cloud attack path
Safety and ethicsNovember 13, 2026, 18:05

Wiz Green Agent maps fixes for a critical cloud attack path

Wiz has detailed an investigation by Green Agent into an internet-facing Kubernetes container with a live vulnerability and access to sensitive customer data. The…

For developers
For developersNovember 13, 2026, 17:14

Open Agentic Web Challenge offers $13,000 for interoperable AI agents

A global group of technology, research, standards and civil society organizations has launched the Open Agentic Web Challenge, offering $13,000 for open and…

Products and tools
Products and toolsNovember 12, 2026, 19:21

RemotePC Adds MCP Integration for AI-Assisted Computer Management

RemotePC has introduced an integration that lets customers use ChatGPT, Claude and other compatible AI assistants to monitor remote computers and perform supported…