Contentful MCP Server

What is Contentful MCP Server?

It is a community-driven MCP server that integrates with Contentful’s Content Management API (CMA) for comprehensive content management. It provides tools for entries, assets, comments, spaces, environments, content types, bulk operations, and localization. Intended for use with MCP clients like Claude Desktop.

How to use Contentful MCP Server?

Install via Smithery or run with npx: npx -y @ivotoby/contentful-management-mcp-server. Configure a Contentful Management API token as an environment variable (CONTENTFUL_MANAGEMENT_ACCESS_TOKEN) or argument (--management-token). Optionally set SPACE_ID and ENVIRONMENT_ID to scope operations. Use stdio transport by default, or enable StreamableHTTP with --http.

Key features of Contentful MCP Server

• Full CRUD operations for entries and assets
• Comment management with threaded conversations
• Space and environment creation, update, and deletion
• Content type management including publishing
• Localization support for multiple locales
• Content publishing and unpublishing workflow control
• Bulk publish, unpublish, and validate operations
• Smart pagination limited to 3 items per request

Use cases of Contentful MCP Server

• Manage content entries and assets across multiple spaces
• Collaborate on content with threaded comments and reviews
• Batch publish or unpublish large sets of entries and assets
• Automate space and environment scaffolding
• Validate content consistency before publishing

FAQ from Contentful MCP Server

What prerequisites are needed?

You need a Contentful account and a Content Management API token generated from your account settings.

How is data accessed and where does it live?

All operations use the Contentful Content Management API. Data remains in your Contentful spaces; the server does not store any data locally.

What transport modes are supported?

Two modes: stdio (default, for clients like Claude Desktop) and StreamableHTTP (enable with --http for web-based integrations).

Why are list results limited to 3 items?

To prevent context window overflow in LLMs. Pagination metadata is included, and the LLM is prompted to offer retrieving more items.

Can the server be scoped to a specific space and environment?

Yes, by setting SPACE_ID and ENVIRONMENT_ID environment variables or using --space-id and --environment-id arguments. This restricts tools to those scopes.