AI agents: the content index is at /llms.txt. Every page is also available as Markdown — send the request header Accept: text/markdown, or append .md to the page URL.
Xperience Docs

Cookie consent

We use necessary cookies (opens in a new tab) to run our website and improve your experience while browsing. Additional cookies are only used with your consent. You may revoke your consent on the Cookies Policy (opens in a new tab) page or in your browser at any time.

Configure (opens in a new tab)
  • Documentation
    Administration interface basicsDevelopers and adminsBusiness usersChangelogGlossary
  • Guides
    Welcome to Xperience by KenticoAdministration interface basicsCustomer journeysDigital marketingConfigurationDevelopmentArchitectureUpgrade to Xperience by Kentico
  • Training
    For AdminsFor Business usersFor DevelopersFor Solution architects
  • API Examples
    ConfigurationContent managementDevelopmentDigital marketing
  • Changelog
    Documentation updatesSecurity advisories
  1. Home
  2. Developers and admins
  3. API
  4. Management MCP server

Management MCP server


Was this page helpful?

Send us feedback Go back to top
  • Community Questions & Answers (opens in a new tab)
  • Contact support (opens in a new tab)

The Management MCP server is part of the KentiCopilot initiative, which brings AI-assisted development capabilities to the Xperience platform.

The Management MCP server (formerly known as Content types MCP server) is a Model Context Protocol (MCP)⁠ (opens in a new tab) server that works with any IDE, agent, or CLI tool that supports MCP. It connects to the management API in your local Xperience application and exposes its CRUD operations for Xperience objects as MCP tools that AI agents can discover and call.

This enables AI-assisted management of Xperience objects such as content types, content items, pages, channels, languages, and more. It can be used to automate content modeling workflows, accelerate migration scenarios, and simplify integration with external systems.

You can use the Management MCP server to implement a model created by the Content Modeling MCP Server. See Prototype your model with Content modeling MCP server to learn more.

How it works

  1. You run your Xperience application locally with the Management MCP server configured.
  2. Your IDE or AI client reads the MCP server configuration and starts the Management MCP server.
  3. The MCP server reads the available operations from endpoints created by the management API.
  4. The MCP server exposes those operations as MCP tools.
  5. Your AI agent calls those tools when handling relevant prompts.

The AI agent can then create, read, update, and delete supported objects, such as content types, content items, pages, channels, and more, through natural language prompts. See Reference - Management MCP capabilities for the full list of supported objects and tools, or Use the Management MCP for example prompts, CI guidance, and troubleshooting.

Prerequisites

  • A local Xperience by Kentico instance (see System requirements).
  • Node.js 22⁠ (opens in a new tab) or newer (see Task-specific requirements).
  • An IDE or AI client that supports the Model Context Protocol⁠ (opens in a new tab) (e.g., VS Code with GitHub Copilot, Claude Code, Cursor).

Set up the Management MCP server

See Configure the Management MCP for steps to enable the management API, add the MCP server to your IDE, limit available tool groups and tools, and configure asset uploads.

Security limitations

The management API is only intended for use on local development instances. It currently only provides basic authentication and no authorization options for various types of operations.

Do not enable the management API for production deployments or other publicly available instances.

For the full picture of what access the management API grants an agent, which objects it does not expose, and how the data flows onward to your model provider, see KentiCopilot security.

Update policy and backward compatibility

The Management MCP server and the underlying management API are designed to be operated by AI agents. For this reason, they are not covered by Kentico's standard backward compatibility policy and instead follow a more flexible set of rules:

  • The 12-month backward compatibility period is not guaranteed
  • Tools, API routes, and parameters can be removed or renamed without a transition period
  • Breaking changes can occur without advance notice

Assume that the tool set and the API surface will be reworked and expanded often, including in weekly hotfix releases. All notable modifications are recorded in the product Changelog.

The impact of changes on projects should be low, because agents can discover the current state of the server on their own and adapt. The tool list advertised by the MCP server and the OpenAPI specification⁠ (opens in a new tab) generated by the management API describe exactly what is available in the version you are running, and that description is authoritative. The OpenAPI specification is available under the /kentico-api/management/v1/openapi.json URL path after the Management API is enabled.

Use Reference - Management MCP capabilities as an overview of the covered objects and operations, but rely on your own instance when you need an exact answer.

To limit the impact of updates on your projects:

  • Only access the management API through an AI agent rather than calling it directly (for example, from scripts).
  • Let agents read the available tools and routes from the MCP tool list and the OpenAPI specification instead of working from a fixed list.
  • Do not hardcode tool names in prompts or agent configuration.
  • Do not generate a client SDK from the OpenAPI specification. Such a client becomes outdated as soon as the API surface shifts.

Kentico provides support for the Management MCP server and responds to reported issues. Because this area moves quickly, fixes and updates are delivered on a best-effort basis.

The features are shaped by early feedback, so check the product roadmap⁠ (opens in a new tab) to see what's coming and share your ideas.

On this page

  • How it works
  • Prerequisites
  • Set up the Management MCP server
  • Security limitations
  • Update policy and backward compatibility

Related pages

  • Reference - Management MCP capabilities
  • Configure the Management MCP
  • Use the Management MCP