Kontent.ai MCP server

Understand the MCP server

The Model Context Protocol (MCP) is an open standard that defines how applications give large language models (LLMs) access to tools and data. An MCP server offers a set of operations that an AI tool can run on your behalf. Kontent.ai hosts the MCP server at https://manage-mcp.kontent.ai/mcp. You don't install or run anything. Add the URL to the AI tool of your choice and authenticate with a Management API key. With the Kontent.ai MCP server, you can:

• Read your Kontent.ai content model structure and create new content types, snippets, and taxonomies
• Query, create, and update content items and their language variants
• Move content items through the workflow steps, publish, and unpublish content as needed
• Build a content model from diagrams or descriptions that the AI interprets and implements

For more information on the protocol specification, go to modelcontextprotocol.io.

Before you connect

You'll need:

• A Management API key with the permissions the AI needs
• The ID of your Kontent.ai environment
• An AI tool that supports remote MCP servers over streamable HTTP and custom request headers

The hosted server accepts Management API keys only and rejects personal API keys.

Connect your AI tool

Use these values in any AI tool that supports remote MCP servers:

Claude Code

In your terminal, run the following command: Then, run claude mcp list to verify the connection.

Claude Desktop

Claude on the web and in Claude Desktop can connect through a custom connector. For detailed instructions, see how to add a connector that isn't in the directory in the Claude documentation.

1. In MCP server URL, enter https://manage-mcp.kontent.ai/mcp.
2. In Authentication, select No sign-in.
3. In Request headers, select the authorization header and enter Bearer <management-api-key> as the value.
4. Click Add.

The connector doesn't use OAuth, because the MCP server currently supports API keys only. Alternatively, Claude Desktop can connect through its JSON configuration file. The mcp-remote package passes your key to the server.

Cursor

1. Open ~/.cursor/mcp.json to configure all projects, or .cursor/mcp.json to configure one project.
2. Copy the JSON below to add the server.
3. Set the KONTENT_MANAGEMENT_API_KEY environment variable to your Management API key in your shell profile or system environment.
4. Restart Cursor.

VS Code

1. In your workspace, create the .vscode/mcp.json file.
2. Copy the JSON below to add the server.
3. Start the server.
4. When VS Code prompts for the key, enter the Management API key.

Provide the environment ID

The environment ID is an input to the server's tools, not part of the server URL. You can work with several environments in one prompt, such as comparing staging and production, if the API key has access to each of them. Tell the AI which environment to work in. To avoid repeating the ID, add it to your AI tool's instructions file, such as CLAUDE.md in Claude Code. If you work with several environments, list their names and IDs there. If the ID is missing, or the API key can't access the environment, the server returns an error that guides the AI.

Supported operations

The server exposes tools that let your AI assistant manage:

• Content types and content type snippets
• Taxonomy groups and terms
• Content items and language variants
• Assets and asset folders
• Collections and spaces
• Languages
• Workflows and roles

The server doesn't cover asset upload and deletion, users and invitations, webhooks, previews, custom elements, or environment management.

Security and permissions

The AI acts within the permissions of the key you provide. We recommend limiting the key to only the permissions and environments AI needs. Many AI tools ask for confirmation before they run a tool. Treat these prompts as a safeguard, not a guarantee. Content that the AI reads can contain instructions that manipulate it (prompt injection). Kontent.ai runs only the latest version of the server, so tool names and behavior can change between versions.

Common use cases

The examples below show how the MCP server tools empower AI assistants to provide meaningful, context-aware interactions, helping teams design, scale, and maintain content efficiently.

Understand and extend a content model

These actions help AI assistants understand the structure of the project and suggest or create new types based on your needs.

• “What content types are already defined in this environment?”
• “Add a new 'Event' content type with elements for name, date, location, and description.”
• “Update the 'Article' content type to include a taxonomy element for industries.”
• “Delete the 'Press release' content type.”

Reuse elements across content types with snippets

Content type snippets are ideal for maintaining consistency across content types while avoiding duplication. Assistants can define snippets once and insert them wherever needed, ensuring that updates are propagated automatically.

• “Create an 'SEO metadata' content type snippet with title, description, and image elements.”
• “List all content type snippets in this environment.”

Categorize and filter content with taxonomies

Taxonomies provide structure for organizing and filtering content. With the MCP server, AI assistants can create and update taxonomy groups and terms programmatically.

• “Add a taxonomy group called 'Industries' with terms for Healthcare, Finance, and Education.”
• “List all taxonomy groups defined in this environment.”

Manage content items

Draft, update, and delete content using AI. The AI assistants scan what elements your items contain, and the languages you have in your project. They can then generate content based on a content brief or instructions you provide, and create the content item with its language variants.

• “Create a blog post called 'Treatments after heart surgery' based on this content brief in these languages: English, Spanish, and French.”
• “Update the title and summary of the English variant of the item with this ID {item_ID}.”
• “Delete the Spanish variant of the content item 'Mid-century modern décor at home'.”
• “Search all French variants of articles containing the keyword 'avocado'.”
• “Create a new draft version of the English variant of item {item_ID}.”

Work with multiple languages

Understanding language settings helps the assistant make localization-aware modeling decisions, adapt to localized workflows, and offer content model recommendations based on supported languages.

• “What languages are available in this environment?”
• “Create missing Spanish variants for all English blog posts.”

Move content through workflow steps

Workflows define the lifecycle of content from draft to review to published. AI assistants can automatically move content items between workflow steps, helping teams enforce editorial processes and keep content production on track.

• “Move all draft blog posts to the Review step.”
• “Schedule the Spanish variant of item {item_ID} to publish on 2026-12-24 at 09:00 Madrid time.”
• “Unpublish the German variant of item {item_ID} tomorrow at midnight.”

Fetch what’s in your asset library

Assets like images and documents can add structure and enhance your content. AI assistants can fetch or include assets in your content items.

• “List all images modified in the past week.”
• “Get the details of asset {asset_ID}.”
• “Insert the image of the forest trail {asset_ID} into the Autumn Hiking Tips article.”

Find and filter what you’re looking for

Filters help refine what content items you‘re looking for using specific keywords and criteria. You can also search for items based on what you mean in natural language. With the MCP server, AI assistants can help locate items and language variants based on your queries.

• “Find unfinished articles that have ‘cardiology‘ in the name“
• “Show recent Spanish blog posts that are assigned to Sofia Peletier in ascending order“
• “I’m looking for items about plants that thrive in dry environments.“