Contentful MCP Server – Content Management Tools
Contentful's official MCP servers let AI agents read and write content, manage content models, work with assets, invoke AI Actions, and operate across Contentful spaces and environments. Use them when an agent needs governed, conversational access to Contentful content operations and content modeling workflows.
Overview
Contentful's official MCP servers connect AI assistants to Contentful spaces
and environments. They expose Contentful content-management capabilities as MCP
tools, allowing agents to help create, update, organize, publish, and model
content while staying within configured Contentful permissions.
What the MCP server enables
Contentful provides both a remote hosted MCP server and a local open-source MCP
server. The servers expose the same core toolset, including tools for:
- Initial context and setup guidance.
- Listing, retrieving, creating, updating, publishing, unpublishing, and deleting
content types. - Searching, semantically searching, retrieving, creating, updating, publishing,
unpublishing, and deleting entries. - Uploading and managing assets.
- Working with snapshots, environments, locales, tags, tasks, comments, and
related content-management workflows. - Invoking Contentful AI Actions where enabled.
- Supporting natural-language content operations such as creating launch
content, updating many entries, organizing campaign assets, or modifying a
content model.
When to use it
Use Contentful MCP when an AI agent should assist with content operations inside
Contentful. Practical examples include drafting and creating a blog post,
updating product entries, publishing approved content, uploading campaign
images, reviewing content types before a migration, adding fields to a model, or
running semantic search across entries. The remote server is well suited for
day-to-day team use where administrators need space, environment, and tool
allow-list controls. The local server is useful for development, scripting, CI,
and environments where OAuth is not available.
Connection and authentication
The remote server is hosted by Contentful at https://mcp.contentful.com/mcp
and uses OAuth 2.1. Users sign in to Contentful through the MCP client, select
spaces and environments for the session, and receive MCP-scoped authorization.
Remote access also depends on the Contentful MCP app being installed and
configured in each target space and environment.
The local server runs with npx -y @contentful/mcp-server. It requires
CONTENTFUL_MANAGEMENT_ACCESS_TOKEN and SPACE_ID; ENVIRONMENT_ID defaults
to master when omitted, and CONTENTFUL_HOST defaults to
api.contentful.com.
Key considerations
Contentful's remote MCP server is documented as Beta. Remote access enforces two
permission layers: the user's Contentful permissions and the per-space,
per-environment tool allow-list managed by the Contentful MCP app. The local
server runs as the identity that owns the personal access token, and there is no
additional per-space, per-environment, or per-tool gate beyond that token. Use
PROTECTED_ENVIRONMENTS locally to block write and delete operations in
sensitive environments, but treat it as an MCP-server guard rather than a
replacement for Contentful roles, permissions, and review workflows. Require
human approval before publishing, deleting, bulk updating, or changing content
models in production spaces.
Supported Transports
streamable_http
URL: https://mcp.contentful.com/mcp
stdio
Command: npx
Args:
-y@contentful/mcp-server
Frequently Asked Questions
- When should an AI agent use the Contentful MCP server?
- Use it when an agent needs to assist with Contentful content operations, such as creating or updating entries, managing assets, reviewing content types, publishing approved content, invoking AI Actions, or searching and organizing content across spaces and environments.
- What does the Contentful MCP server add to an AI agent's capabilities?
- It gives the agent live access to Contentful content-management tools, including entries, assets, content types, environments, locales, tags, snapshots, semantic search, publishing workflows, and AI Actions, instead of relying only on static model knowledge or copied content exports.
- What can an AI agent access or manage through Contentful MCP?
- Depending on the selected transport, permissions, and configured tools, the agent can read and write entries, manage content types, work with assets, inspect snapshots, use semantic search, invoke AI Actions, and operate across the spaces and environments available to the user or token.
- How is authentication configured for the Contentful MCP server?
- The remote server uses OAuth 2.1 so each user signs in with their own Contentful identity and selects scoped spaces and environments. The local server uses CONTENTFUL_MANAGEMENT_ACCESS_TOKEN and SPACE_ID, with optional ENVIRONMENT_ID, CONTENTFUL_HOST, and PROTECTED_ENVIRONMENTS environment variables.
- Which transport should be used for the Contentful MCP server?
- Use the remote Streamable HTTP endpoint at https://mcp.contentful.com/mcp for team use, OAuth, and administrator-controlled space, environment, and tool gating. Use local stdio with @contentful/mcp-server for local development, scripting, CI, or clients that cannot use OAuth.