This is a big release. Like, really big. We've got native Model Context Protocol support, a completely new content comparison modal, performance improvements across the board, and a bunch of fixes that have been on our list for a while.
We're going to focus on the main highlights here, but there's a lot more in this release than what we're covering in this post.
Breaking Changes
Before we get into the good stuff, let's talk about what might affect your existing setup.
Removed non-relational types from RELATIONAL_TYPES constant: If you're using extensions or external code that references RELATIONAL_TYPES, note that presentation and group types are no longer included. You'll need to update any custom code that depends on this constant.
Filter hook behavior change: The <scope>.delete filter hook now runs regardless of user permissions and properly uses the keys returned by the hook. If you're using this hook, make sure any necessary permission checks happen in your hook implementation before processing data.
Now, onto the features.
Native MCP Support: AI Agents Meet Your Content
If you've been watching the AI space, you've probably heard about the Model Context Protocol. It's Anthropic's open standard for connecting AI assistants directly to data sources, and as of v11.13, Directus supports it natively.
What does this actually mean? Your AI tools (Claude, ChatGPT, Cursor, etc.) can now interact directly with your Directus instance. No middleware, no copying data back and forth, no custom integrations to maintain.
What becomes possible:
Marketers can ask Claude to migrate blog posts from Google Docs with proper field mapping, organize assets with AI-generated metadata, or build complex landing pages using your existing content relationships.
Developers can prototype entire data models through conversation, create collections with proper relationships, and build automation flows without touching the UI.
The key difference from other AI integrations? MCP works through your existing Directus permissions system. Your AI assistant operates with the same access controls as any other user, and every change gets logged in your audit trail.
Getting started is straightforward:
- Enable MCP in Settings → AI
- Generate an access token for your MCP user (we recommend creating a dedicated user)
- Connect your AI client
- Start managing content through conversation
We've included global delete protection (disabled by default) and comprehensive security controls. For production deployments, check out our security guide in the docs.
The native MCP server supports Claude Desktop, Claude Code, ChatGPT, Cursor, and Raycast. Full documentation including real-world examples and use cases is available at directus.io/mcp.
Content Comparison Modal: Actually Useful Version Reviews
We've completely redesigned how you compare and promote content versions. The old way of reviewing changes between versions involved a lot of manual field-by-field checking. The new comparison modal fixes that.
When you're ready to promote a version, you now get a side-by-side view showing exactly what changed. Fields with differences are highlighted with color indicators, making it immediately obvious what will change when you promote.
The workflow is simple:
- Open the version you want to promote
- Select "Promote Version" from the dropdown
- Review the comparison modal showing all changes
- Accept or reject individual changes as needed
- Click "Promote" to make this version the new main item
Once promoted, that version becomes your active content and the previous main item gets preserved in version history. You can choose to keep or delete the version after promoting.
This is particularly useful for teams managing complex content. Instead of hoping you caught all the changes, you can see exactly what's different before committing to the promotion.
Other Improvements
This release includes a bunch of smaller improvements worth mentioning:
Performance boosts: We added a field-level option that lets you control whether a field is searchable, improving search request performance for large collections. We also created indexes on relevant fields in the activity and revisions tables to reduce query times and enhanced support for MSSQL tables with triggers.
Better error reporting: Data import errors are now more detailed and helpful. Error consistency across the API has been improved, making debugging easier.
SDK updates: Added batch comment updates, field updates, and better handling of nested relational items. The SDK also now supports configurable connect timeouts for realtime connections.
Rate limiting for emails: You can now configure rate limits for sending emails to prevent abuse and stay within provider limits.
Bug fixes: Fixed content versioning for singleton collections, improved tag interface behavior, better decimal separator handling, and a bunch of GraphQL filtering improvements.
Check out the full release notes on GitHub for the complete list of changes, fixes, and dependency updates.
Directus v11.13 is Available Now
Update your instances to get native MCP support, the new content comparison modal, and all the performance improvements.
Head over to directus.io/mcp to explore what's possible with AI-powered content management, or check out the GitHub release notes for the complete changelog.
As always, we recommend backing up your database before upgrading.