Directus MCP. Manage your database structure with conversation.
Gemini Works with every AI agent you already use
…and any MCP-compatible client
Just plug in your AI agents and start using Vinkius.
Directus MCP Server connects any AI agent to your open-source data platform. Use it to manage your entire SQL database—read collections, audit schemas, provision records, or wipe data—all through natural conversation.
It gives your agent direct control over your entire headless CMS structure.
What your AI agents can do
Create cms record
Writes a new record into a Directus collection using a structured JSON payload.
Get collection details
Retrieves the structural properties that define an active content collection or table.
Get single item
Fetches a specific record from the database using its unique UUID.
The agent writes or updates entire JSON records in any collection, or deletes them entirely.
The agent reads the schema of collections and tables, listing all fields and structural rules.
The agent finds specific records within a collection or lists all available content types.
The agent lists active users and checks the system's defined access control rules (RBAC).
The agent verifies media storage contents and provides download links for uploaded files.
Ask AI about this MCP
Supported MCP Clients
Gemini Waiting for input…
019d7586create cms record
Writes a new record into a Directus collection using a structured JSON payload.
019d7586get collection details
Retrieves the structural properties that define an active content collection or table.
019d7586get single item
Fetches a specific record from the database using its unique UUID.
019d7586list collection fields
Lists all available fields and metadata for a specified content collection.
019d7586list collection items
Finds and lists all available records within a specific content collection.
019d7586list directus files
Retrieves metadata about all files stored in the Directus media library.
019d7586list directus users
Lists all active user accounts configured in the Directus instance.
019d7586list schema collections
Enumerates the defined structured rules and tables within the underlying PostgreSQL database.
019d7586patch cms record
Modifies an existing record in a collection by providing a unique ID and new JSON data.
019d7586wipe cms record
Permanently deletes a specific content record from the database.
Choose How to Get Started
Build a custom MCP for your own tools, or connect a ready-made integration from our catalog.
Build Your Own
Turn any API into an MCP. Import a spec, define Agent Skills, or deploy with MCPFusion.
- Import from OpenAPI, Swagger, or YAML specs
- Create Agent Skills with progressive disclosure
- Deploy to edge with MCPFusion framework
- Built in DLP, auth, and compliance on every call
- Real time usage dashboard and cost metering
- Publish to catalog or keep private
Make Your AI Do More
Start with Directus, then connect any of our 4,700+ other servers whenever your AI needs more. One click, no limits.
- Use this MCP plus 4,700+ others, all in one place
- Add new capabilities to your AI anytime you want
- Every connection is secured and compliant automatically
- Track usage and costs across all your servers
- Works with Claude, ChatGPT, Cursor, and more
- New servers added to the catalog every week
What you can do with this MCP connector
You hook up your AI agent to your Directus data platform. It gives your agent total control over your open-source CMS, letting it read, write, and delete data just by talking to it. You treat your database like a conversation, not a bunch of API calls.
Write and modify database records
Your agent writes new records into any Directus collection using a structured JSON payload (create_cms_record), modifies existing records with a unique ID and new JSON data (patch_cms_record), or permanently deletes content records (wipe_cms_record).
Inspect database structure
It lets your agent list all available fields and metadata for a specific content collection (list_collection_fields), check the overall structure of collections and tables in the underlying PostgreSQL database (list_schema_collections), and retrieve the structural properties that define an active content collection (get_collection_details).
List and audit content items
Your agent finds specific records using a unique UUID (get_single_item), or it finds and lists all available records within a collection (list_collection_items). You can also list every defined content collection to see what data types exist (list_directus_collections).
Manage user accounts and permissions
It lists all active user accounts configured in the Directus instance (list_directus_users).
Handle digital assets
Your agent gets metadata about every file stored in the Directus media library (list_directus_files).
How Directus MCP Works
- 1 Subscribe to the Directus MCP Server on Vinkius.
- 2 Provide your Directus Base URL and Static Token (located in User Directory > User > Token).
- 3 Use your AI client (Claude, Cursor, etc.) to prompt the agent to perform a data action, like 'List all articles in the 'blog' collection'.
The bottom line is that your AI client talks to the Directus server, and the server uses the available tools to perform the data action directly.
Who Is Directus MCP For?
Database Administrators, Full-stack Developers, Product Managers, and Content Editors. You need this if you spend time writing complex SQL queries, clicking through admin dashboards, or manually verifying content structure. This server lets you talk to your data platform directly via natural language, bypassing the UI entirely.
Audits schemas, checks user permissions, and manages underlying PostgreSQL tables without writing a single complex SQL statement.
Creates, updates, and deletes content records and digital assets directly from their IDE or chat, treating the database like a simple API endpoint.
Manages headless content and digital assets by simply describing what they want to change or find.
What Changes When You Connect
- Audit Schemas without SQL: Use
list_schema_collectionsto see every PostgreSQL table structure. You don't need to write complex SQL or log into the Studio UI just to verify what fields exist. - Direct Content Manipulation: Write or update records instantly.
create_cms_recordandpatch_cms_recordlet your agent modify data payloads directly from the chat or IDE, treating the database like a simple JSON store. - Audit User Access: The
list_directus_userstool lets you list all admin identities and see who has access. This helps you audit your Role-Based Access Control (RBAC) layer without digging through user panels. - Manage Assets Easily: Don't lose track of media files.
list_directus_filesretrieves structural matching for all stored media, giving you the download routes needed for any frontend. - Inspect Data Types: Use
list_collection_fieldsto inspect a collection's internal arrays. You can validate payloads against the specific column configurations, preventing bad data from entering the system. - Retrieve Specific Content: Need one piece of data?
get_single_itemfetches a record by its UUID. It's faster and cleaner than running a general search query.
Real-World Use Cases
The dev needs to check the article structure.
The developer needs to know if the 'articles' collection has a 'featured_image' field before writing the frontend logic. They can't manually check the Directus UI. Instead, they ask their agent to run list_collection_fields. The agent returns the full metadata, confirming the field exists and what data type it accepts, letting the developer proceed immediately.
The content team needs to correct a user's profile.
A user's profile details are wrong, and the dev needs to update them. Instead of using the admin dashboard, the content editor just tells their agent: 'Update user 123's phone number.' The agent runs patch_cms_record with the new JSON payload, modifying the record without any manual UI interaction.
The administrator needs to wipe old test data.
A large batch of test data from a previous sprint needs to be cleared out of the database. The admin doesn't want to run complex deletion scripts. They prompt the agent to run wipe_cms_record on the target records, instantly clearing the allocated data and keeping the database clean.
The developer needs to list all possible content types.
The developer wants to see every single collection available in the system to plan a new feature. They run list_schema_collections. The agent returns a clean list of all defined tables and their names, giving the developer a complete map of the database structure.
The Tradeoffs
Trying to discover everything via item listing
Running list_collection_items multiple times with varying filters to piece together the schema. This is slow, hits rate limits, and only gives you data, not the structure.
→
Use get_collection_details first. That tool shows the entire structure of the collection. Then, if you need data, use list_collection_items or get_single_item.
Manually writing complex database calls
Writing raw SQL queries into your client application just to check if a field exists or to update a record. This is brittle and doesn't account for RBAC.
→
Let the agent handle it. Use list_collection_fields to check field validity, and use patch_cms_record to update the data. The agent abstracts the SQL layer for you.
Assuming a single data source for everything
Trying to manage both content and media files using only collection listing tools. Media assets are stored separately and require different commands.
→
Always use list_directus_files for media assets. This tool specifically handles the file storage structure, keeping your content management separate from your asset management.
When It Fits, When It Doesn't
Use this if your primary need is managing, auditing, or generating content within a headless CMS environment. If your data layer is based on Directus/PostgreSQL and you need to perform CRUD operations or audit the schema without writing SQL, this is your tool. Don't use it if you need to interact with a completely different backend system (like a dedicated payment processor or an external graph database) — those require specialized API tools. If your goal is just to read a small amount of public data, a basic API call might suffice. But if you need to read the schema, write records, or manage users, this MCP Server gives your agent the necessary depth of control.
Independent Platform Disclaimer: Vinkius is an independent platform and is not affiliated with, endorsed by, sponsored by, verified by, or otherwise authorized by Directus. All third-party trademarks, logos, and brand names are the property of their respective owners. Their use on this website is strictly for informational purposes to identify service compatibility and interoperability.
VINKIUS INFRASTRUCTURE
Cloud Hosted
Managed infra
V8 Isolated
Sandboxed per request
Zero-Trust Proxy
No stored credentials
DLP Enforced
Policy on every call
GDPR Compliant
EU data residency
Token Compression
~60% cost reduction
Works with Claude, ChatGPT, Cursor, and more
The Model Context Protocol standardizes how applications expose capabilities to LLMs. Instead of operating in isolation, your AI gains direct access to external platforms, live data, and real-world actions through secure, standardized connections.
This server provides 10 capabilities that interface natively with Claude, ChatGPT, Cursor, and any MCP client. No middleware. No custom integration required.
Available Capabilities
Auditing a CMS database used to mean logging into the admin UI and clicking through multiple tabs.
Today, checking the data structure is a multi-step process. You log into the Directus Studio, navigate to the collection, and then click into 'Fields' to see the metadata. If you want to see the underlying PostgreSQL tables, you might need to access a separate tool or run a complex query just to enumerate the schema. It's slow, it's clunky, and it's easy to miss something.
With this MCP Server, you simply ask your agent to 'Show me the schema for the 'orders' collection.' The agent runs the necessary tools—like `list_schema_collections`—and gives you a structured, clean list of every defined table and field. You get immediate, programmatic visibility into your entire data architecture.
Using Directus MCP Server: Control data writes and asset retrieval.
Before, updating a record meant jumping between the content editor and the media manager. If you needed to change a user's profile and swap out their avatar, you had to perform multiple steps: find the user record, then manually upload the new file, and then remember to update the record with the new file's ID. This always broke something.
Now, your agent handles the whole sequence. You tell it to update the user and swap the photo. It uses `patch_cms_record` for the data and `list_directus_files` to verify the asset path, executing the entire workflow in one go. It's a single, reliable operation.
Common Questions About Directus MCP
How do I check if a specific field exists in the 'products' collection using list_collection_fields? +
Yes, you check the collection's schema using list_collection_fields. This tool inspects the internal array definitions and confirms if the field exists, what its data type is, and if it accepts formatting.
What is the difference between get_collection_details and list_schema_collections? +
get_collection_details shows the structure of a single, specific collection. list_schema_collections lists all the defined, high-level tables in the entire database, giving you a full map of the system.
Can I delete an entire record using wipe_cms_record? +
Yes, wipe_cms_record is designed to irreversibly delete a record by its ID. Use this when you need to completely clear out data allocations, making sure you understand that this action cannot be undone.
How do I list all users with list_directus_users? +
Running list_directus_users returns a precise array of all active user accounts in the Directus layer. This is useful for auditing and understanding who has access.
How do I use list_directus_files to check which media assets are available for download? +
It retrieves the structure and paths for all uploaded media. This lets your agent build download links and verify file existence across the entire platform.
What's the difference between get_single_item and list_collection_items? +
Use get_single_item to pull one specific row by its UUID. list_collection_items finds bounded routing spaces, giving you a list of IDs or records instead of a single item.
If I update a record using patch_cms_record, how do I verify that the changes were saved? +
The tool provides a confirmation payload upon successful mutation. Your agent reads this payload to verify the updated database values and the ID used for the change.
Can I use create_cms_record to write a record if I don't know the exact collection name? +
No, you must specify the target collection name in the payload. The tool requires the full collection name to write the record and prevent errors.
Can my agent list all collections and their fields in Directus? +
Yes. Use the 'list_schema_collections' tool to see all tables, and then use 'list_collection_fields' for a specific collection. The agent retrieves the absolute structural details defining precisely which fields accept formatting.
How do I create a new record in a specific Directus collection? +
Use the 'create_cms_record' tool. Provide the collection name and a JSON payload mapping the columns. Your agent will orchestrate the request to drop the formatted endpoint into your database and return the new UUID.
Can I audit authorized users and their roles via the agent? +
Absolutely. The 'list_directus_users' tool retrieves the explicitly mapped profile arrays iterating the exact users authorized within the database layer. This helps you monitor RBAC boundaries and organizational identities.
Use it with your favorite AI tools
Connect this server to Cursor, Claude, VS Code, and more.
More in this category
Conda (Anaconda.org)
Enable your AI agent to search packages, inspect metadata, and explore channels on Anaconda.org via the Conda API.
Dokku
Manage self-hosted apps via Dokku — monitor containers, scale processes, handle environment variables, and stream logs directly from any AI agent.
DataFrame Aggregator Engine
Perform blazingly fast GroupBy and Aggregations on massive CSVs local. Save millions of AI tokens and get mathematically perfect sums, means, and counts.
You might also like
Yousign
Secure electronic signatures for documents and contracts with AI using Yousign V3.
Livestorm
Connect your AI agent to Livestorm to manage webinars, registrations, attendees, and analytics via natural language.
Accelevents
All-in-one event management platform — manage events, attendees, and session registration via AI.