MCP-Compatible Headless CMS

The Model Context Protocol (MCP) lets AI code editors talk to external tools. ElmapiCMS has an official MCP server that gives editors like Cursor and Claude Code 17 tools for managing collections, content entries, and assets. Your AI editor becomes a CMS client.

R
Raşit Apalak
·
·
10 min read

MCP is an open standard that connects AI tools to external systems. If your headless CMS has an MCP server, your AI editor can manage content without you ever opening the admin panel. ElmapiCMS is one of the first headless CMS platforms to ship an official MCP server.

What is the Model Context Protocol?

The Model Context Protocol (MCP) is an open standard created by Anthropic. It defines how AI tools can discover and call functions on external services. Think of it as a bridge: on one side is your AI editor (Cursor, Claude Code), on the other is a service with an API (your CMS). The MCP server sits in between, translating natural-language requests into API calls.

Without MCP, you'd paste API documentation into a chat and ask the AI to write fetch calls. With MCP, the AI connects directly to the service and calls its API through defined tools. It knows what's available, what parameters each tool needs, and what responses look like.

Why Does MCP Matter for a CMS?

A headless CMS is all about structured content: collections, fields, entries, assets. Setting up content models is repetitive work. Creating test data is tedious. Querying content to verify it works is another round trip to the admin panel.

With an MCP server, your AI editor handles all of that:

  • Content modeling – "Create a Products collection with name, price, description, image, and category fields"
  • Test data – "Add 10 sample products with realistic prices and descriptions"
  • Querying – "Show me all products under $50 sorted by price"
  • Frontend code – "Read the Products schema and generate a typed API client"

The AI does the CMS work and the frontend work in one session.

1. NomaCMS

NomaCMS is our managed, AI-native headless CMS. You get structured content, a REST API, a JavaScript SDK, and AI built in for writing, translation, and editing. Ship to Next.js, Nuxt, Astro, or other stacks without running your own CMS server. Connect Cursor, Claude Code, or other MCP-compatible tools using the Noma MCP server. Plans start at $15 per month after a 7-day free trial; see nomacms.com for tiers and limits.

Why consider NomaCMS

  • Managed cloud with global CDN for assets and no server ops on your side
  • AI assistant, inline field actions, and one-click entry translation
  • REST API and SDK with predictable shapes for frontends and automation
  • MCP server for AI editors (npx -y @nomacms/mcp-server)
  • Project auth, webhooks, locales, and team workspaces on paid plans

Best for

Teams that want a hosted CMS with AI-first workflows and delivery infrastructure included, not only a self-hosted install.

Try NomaCMS

Start a free trial or compare plans on the Noma site.

Start free trialRead the docs

ElmapiCMS MCP Server

ElmapiCMS provides an official MCP server published on npm as @elmapicms/mcp-server. It connects any MCP-compatible AI editor to your ElmapiCMS instance.

Key features

  • 17 tools covering the entire ElmapiCMS API
  • 3 built-in resources the AI reads for context (field types, query syntax, general reference)
  • Advanced query support with 13 filter operators, OR groups, relation filtering
  • Multi-project support (add multiple MCP configs for different projects)
  • Open source on GitHub

Install

npm install -g @elmapicms/mcp-server

Supported Editors

Cursor

Add to ~/.cursor/mcp.json:

{
  "mcpServers": {
    "elmapicms": {
      "command": "npx",
      "args": ["@elmapicms/mcp-server"],
      "env": {
        "ELMAPI_API_URL": "https://your-domain.com/api",
        "ELMAPI_API_KEY": "your-api-key",
        "ELMAPI_PROJECT_ID": "your-project-uuid"
      }
    }
  }
}

Verify in Settings → MCP. The server should show a green status.

Claude Code

Add via the CLI:

claude mcp add elmapicms \
  -e ELMAPI_API_URL=https://your-domain.com/api \
  -e ELMAPI_API_KEY=your-api-key \
  -e ELMAPI_PROJECT_ID=your-project-uuid \
  -- npx @elmapicms/mcp-server

Claude Code discovers the tools automatically.

Other MCP clients

Any tool that implements the Model Context Protocol can use the ElmapiCMS MCP server. The server uses stdio transport and follows the MCP specification.

The 17 Tools

The MCP server exposes these tools to the AI editor:

Project

  • get_project – Read project info (name, locales, settings)

Collections & Fields

  • list_collections – List all collections
  • get_collection – Get a collection with its full field schema
  • create_collection – Create a collection (optionally with fields in one request)
  • update_collection – Update collection name and slug
  • reorder_collections – Change collection display order
  • create_field – Add a field to a collection
  • update_field – Update field properties
  • reorder_fields – Change field display order

Content Entries

  • list_entries – List entries with filtering, sorting, pagination
  • get_entry – Get a single entry by UUID
  • create_entry – Create a new entry
  • update_entry – Update an existing entry
  • delete_entry – Soft-delete an entry

Assets

  • list_assets – List assets with pagination
  • get_asset – Get an asset by UUID or filename
  • upload_asset – Upload a local file

Advanced Queries

The list_entries tool supports the full ElmapiCMS query API. You don't need to remember the syntax; just describe what you want and the AI builds the query.

13 filter operators

eq, lt, lte, gt, gte, not, like, in, not_in, null, not_null, between, not_between

OR groups

Combine conditions with OR logic: "Find entries tagged 'clearance' OR related to the 'Summer Sale' campaign."

Relation filters

Filter by related entries: "Find blog posts where the author's name is 'John'."

Pagination and sorting

Paginate results, sort by any field (ascending or descending), count entries, and exclude fields from responses.

Built-in Resources

The MCP server includes 3 resources the AI reads automatically for context:

  • Field Types Reference – All 16 field types with their options, validations, and common patterns
  • Query Reference – Full documentation of filter operators, sorting, pagination, and relation queries
  • General Reference – Overview of ElmapiCMS concepts and API structure

This means the AI already knows how your CMS works before you ask your first question. It reads these resources to understand what field types are available, how to structure queries, and what options each tool accepts.

Do Other CMSes Support MCP?

As of early 2026, most headless CMS platforms do not provide an official MCP server:

  • Strapi – No official MCP server
  • Payload CMS – No official MCP server
  • Directus – No official MCP server
  • Sanity – No official MCP server
  • Contentful – No official MCP server
  • Hygraph – No official MCP server
  • ElmapiCMS – Official MCP server with 17 tools

You can build custom MCP servers for any CMS with a REST or GraphQL API, but that requires development effort. ElmapiCMS ships one out of the box.

Conclusion

An MCP-compatible headless CMS lets AI code editors manage your content directly. ElmapiCMS provides an official MCP server with 17 tools, 3 built-in resources, advanced query support, and multi-project configuration. It works with Cursor, Claude Code, and any MCP-compatible client. Install the npm package, add your credentials, and your CMS becomes part of the AI editing workflow.

Share this post: