Content Modeling MCP Server

✖
Copy to clipboard
Advanced license required

Features described on this page require the Xperience by Kentico Advanced license tier.

Content modeling is a crucial yet time-consuming phase of any Xperience by Kentico project. Analyzing the results of a content audit, identifying key content types, and defining presentation components takes time and requires experience with Xperience. Yet these steps are critical for building a solid content model that determines the success of your entire project.

This guide introduces an AI-powered content modeling approach that streamlines content modeling tasks. By leveraging a structured workflow with built-in validation, you can reduce the time spent on prototyping a content model for Xperience application. The content modeling workflow takes away some of the mundane tasks, such as writing instructions for editors, and helps you quickly deliver project prototypes that are easy to iterate on.

For a comprehensive overview of content modeling concepts and strategies, see Content modeling guide. If you’re new to Xperience by Kentico content modeling, start with Content modeling basics.

What you’ll learn:

  • How to install the MCP server
  • How to use orchestration prompts and validation tools for each modeling phase
  • Ways to validate and visualize content structures for stakeholders
  • Best practices for efficient content model design in Xperience by Kentico

This approach helps agencies and implementation partners accelerate project delivery while maintaining high-quality content structures aligned with modern multi-channel needs. It helps you focus on building prototypes, not final outputs, so you can validate and refine your content model early.

About the Content Modeling MCP Server

The Content Modeling MCP Server is part of KentiCopilot, Kentico’s initiative to help developers effectively adopt AI in building Xperience by Kentico projects.

The server provides a deterministic, tools-based approach to content modeling that follows the Orchestration prompt + Tool pattern. Built on the Model Context Protocol (MCP), an open standard that enables AI agents to dynamically discover and use external tools, the server offers reliable content modeling workflows with client-side LLM execution and server-side validation.

The MCP server works directly within your integrated development environment (IDE), allowing AI agents to access specialized content modeling capabilities without switching between tools or platforms.

For querying and retrieving information about existing content types in your Xperience application, see Content type management API.

Key features

The Content Modeling MCP Server provides the following capabilities:

  • Orchestrated content modeling workflow: Complete 5-phase workflow with built-in validation and transition logic.
  • Phase-specific prompts: Individual prompts for each workflow phase enabling granular control.
  • Validation tools: Deterministic tools that validate data structure and logic at each phase.
  • Knowledge resources: Comprehensive reference materials including field types, best practices, and examples.
  • Multi-format output: Generates structured JSON content models and markdown documentation (including Mermaid ERD diagrams).

Install the MCP server

The Content Modeling MCP Server supports installation across multiple IDEs. Choose your preferred client below for setup:

Development environment

Installation

Documentation

VS Code

Install in VS Code

VS Code MCP Official Guide

Visual Studio

Manual configuration required by adding a new tool in Select tools in the Copilot chat window.

Required parameters:
Server ID: kentico-cm-mcp
Type: HTTP/SSE
URL: “https://ai.kentico.com/content-model-mcp”

Visual Studio MCP Official Guide

Claude Desktop

Manual configuration required.

For the URL, use https://ai.kentico.com/content-model-mcp.

Claude Desktop Remote MCP Guide

Cursor IDE

Install in Cursor

Cursor MCP Official Guide

The following table shows which IDE and AI model combinations work best with the Content Modeling MCP Server:

IDE

Status

Notes

Claude Desktop

✅ Fully supported

Recommended for best experience

Cursor

✅ Fully supported

Recommended for best experience

VS Code

⚠️ Inconsistent

May experience inconsistent behavior with Claude Sonnet models. GPT-5 is stable.

Getting started

To begin using the Content Modeling MCP Server:

  1. Install the Content Modeling MCP Server using the installation links for your preferred IDE.
  2. Open your MCP-compatible client (Claude Desktop, Cursor, VS Code with Copilot, etc.).
  3. Verify the Content Modeling MCP Server appears in your available prompts and tools.
  4. Start the content modeling workflow. The method varies by IDE, some might include the MCP server name in front of the /start_content_modeling prompt:

IDE/Platform

How to invoke

Additional steps

Claude Desktop

Click Add from kentico-cm-mcp

Click START HERE from the dropdown menu, then provide your project description into the Enter prompt inputs pop-up

Claude Code

/start_content_modeling

Provide your project description with the prompt

Cursor

/start_content_modeling

Provide your project description in the pop-up that appears

VS Code

/start_content_modeling or via prompts menu

Provide your project description in the pop-up, then select Insert as text

Providing context for better results

The quality and specificity of your generated content model depend on the context you provide. Consider including:

  • Links to design files (for example, Figma designs - you can also add the Figma Developer MCP for enhanced context extraction)
  • Project documentation, requirements, or specifications (attach files or reference them in your prompt)
  • Content audit results or existing content structures
  • Target audience and channel information

How you add context varies by IDE (attaching files, pasting links, referencing documents). The more relevant context you provide, the more tailored your content model will be. With minimal context, the MCP server generates a more general-purpose model.

Prompts and tools reference

In the context of the MCP server, tools and prompts are capabilities that the MCP server exposes to connected clients (for example, GitHub Copilot, Claude, or Cursor). Each tool performs a specific function and can be called programmatically by the client when it needs to perform content modeling tasks.

Orchestration prompts

The Content Modeling MCP Server provides orchestration prompts that guide you through the content modeling workflow:

Prompt name

Description

Parameters

start_content_modeling

START HERE: Primary way to begin content modeling. Guides you through a complete 5-phase workflow from requirements gathering to final deliverables.

projectDescription (string, optional): Project description and context

content_modeling_1_requirements

INTERNAL: Phase 1: Requirements gathering and approach selection

projectContext (string, optional): Optional context about the project

content_modeling_2_architect

INTERNAL: Phase 2: Content type design with fields and metadata

approvedApproach (string, optional): The approved approach from Phase 1

requirementsSummary (string, optional): Summary of requirements from Phase 1

content_modeling_3_relationships

INTERNAL: Phase 3: Relationship design between content types

contentTypes (string, optional): List of content types from Phase 2

content_modeling_4_pagebuilder

INTERNAL: Phase 4: Page Builder templates and sections

approach (string, optional): The approved approach

previousPhasesSummary (string, optional): Content types and relationships from previous phases

content_modeling_5_validator

INTERNAL: Phase 5: Final validation and deliverable generation

completeModel (string, optional): Complete content model from all previous phases

Validation tools

The following validation tools are used to verify the outputs of each content modeling phase. Validation tools are called when the agent completes the corresponding phase-specific work as described in the workflow prompts.

Tool Name

Description

Parameters

content_modeling_validate_requirements

This tool validates outputs of Requirements gathering phase: project requirements structure and basic fields. Call this tool ONLY after completing the Requirements gathering work (requirements gathering and approach selection) described in the phase-specific prompt, when the requirements are ready for validation. AI agent must provide the chosen approach based on decision logic in instructions.

projectDescription (string): Project description and context

approach (string): Chosen approach - “atomic” or “page-builder” (AI agent determines this)

channels (string, optional): Target channels

teamExperience (string, optional): Team experience level

contentVolume (string, optional): Expected content volume

contentReuse (string, optional): Content reuse needs

content_modeling_validate_content_types

This tool validates outputs of Content type design phase: content type JSON structure and field definitions. Call this tool ONLY after completing the Content Type Design work (content type design with fields) described in the phase-specific prompt, when the content types are ready for validation.

contentModelJson (string): Complete content model JSON with contentTypes array to validate

content_modeling_validate_relationships

This tool validates outputs of Relationship design phase: relationship definitions and cardinality rules. Call this tool ONLY after completing the Relationship Design work (relationship design between content types) described in the phase-specific prompt, when the relationships are ready for validation.

contentModelJson (string): Complete content model JSON with relationships array to validate

content_modeling_validate_pagebuilder

This tool validates outputs of Page Builder design phase: Page Builder templates or skips for atomic approach. Call this tool ONLY after completing the Page Builder Design work (Page Builder templates, sections, and widgets design) described in the phase-specific prompt, when the Page Builder components are ready for validation.

contentModelJson (string): Complete content model JSON with pageBuilder object to validate

content_modeling_final_validation

This tool validates outputs of Validation & final output phase: final validation of complete content model. Call this tool ONLY after completing the Validation & Final Output work (final validation and output generation) described in the phase-specific prompt, when the complete model is ready for final validation.

completeModelJson (string): Complete model JSON from all previous phases

Content modeling phases overview

The content modeling workflow in Xperience by Kentico is orchestrated by the start_content_modeling prompt, which guides users through all five phases automatically. Each phase is supported by a dedicated prompt and validation tool to ensure accuracy and completeness before moving to the next step:

1. Requirements gathering & approach selection

  • Gather project requirements, business goals, and editorial needs
  • Select the modeling approach (atomic vs. page-builder) based on requirements
  • Prompt: content_modeling_1_requirements
  • Validation tool: content_modeling_validate_requirements

2. Content type design

  • Define content types, fields, and metadata for each entity
  • Ensure field definitions and structures are consistent and complete
  • Prompt: content_modeling_2_architect
  • Validation tool: content_modeling_validate_content_types

3. Relationship design

  • Map relationships between content types, including cardinality and references
  • Validate relationship definitions and ensure correct linking strategies
  • Prompt: content_modeling_3_relationships
  • Validation tool: content_modeling_validate_relationships

4. Page Builder templates & sections

  • Design Page Builder components, such as page templates, Page Builder sections and widgets
  • Validate template structure or skip if using atomic modeling
  • Prompt: content_modeling_4_pagebuilder
  • Validation tool: content_modeling_validate_pagebuilder

5. Final validation & deliverables

  • Perform comprehensive validation of the complete content model
  • Generate deliverables for implementation
  • Prompt: content_modeling_5_validator
  • Validation tool: content_modeling_final_validation

Next steps: Implement your content model

After designing and validating your content model, you can implement it in your Xperience application using the Content type management API. The Management API MCP server allows AI agents to create content types, add fields, and configure reusable field schemas based on your designed model.

Example prompts

Using the start_content_modeling prompt

The only way to begin content modeling is using the start_content_modeling prompt or the alternatives provided in the Getting started table:

  • “Use the start_content_modeling prompt to help me build a content model for a car dealership website”
  • “Guide me through the complete content modeling workflow for an commerce platform using start_content_modeling”