Note
Access to this page requires authorization. You can try signing in or changing directories.
Access to this page requires authorization. You can try changing directories.
Fabric Core MCP Server is a remote endpoint that exposes Fabric resource-management operations through the Model Context Protocol (MCP). Your AI agent selects tools in response to your prompts, and the server calls Fabric APIs using the authenticated identity's permissions.
Tip
New to Fabric MCP? Start with the Fabric MCP Servers overview to understand the protocol and compare Core and local servers.
How it works
When you connect your AI agent to Fabric Core MCP Server:
- The agent connects to the remote endpoint at
https://api.fabric.microsoft.com/v1/mcp/core. - You authenticate through a browser-based OAuth 2.0 flow with Microsoft Entra ID.
- The agent discovers tools for search, workspaces, items, permissions, folders, capacities, and OneLake management.
- When you ask a question, the agent selects the right tool and the server calls the Fabric REST API on your behalf.
- Operations use the authenticated identity's Fabric permissions. Fabric records supported operations in audit logs using the identity that executes them.
Supported operations
Fabric Core MCP Server currently supports:
- Catalog search — Search for items across workspaces in the OneLake Catalog by display name, description, or workspace name. Optionally filter by item type.
- Workspace operations — List, create, update, and delete workspaces
- Item management — Create, read, update, and delete Fabric items (lakehouses, semantic models, notebooks, pipelines, and other item types)
- Workspace role management — Grant and revoke workspace roles (Admin, Member, Contributor, Viewer)
- OneLake shortcuts — List and inspect shortcuts, create or update individual shortcuts, create shortcuts in bulk, delete shortcuts, and reset the shortcut cache.
- OneLake data security — List data access roles, create or update a single role, replace an item's complete role set, and delete a role.
- OneLake settings — Read workspace settings and configure the default storage tier, diagnostics, and diagnostic-log immutability.
- Folder organization — Create and manage folders within workspaces
- Capacity information — List available capacities and check operation status
For the complete list of tools, see the tools reference. For OneLake inputs and safeguards, see OneLake management tools. OneLake data access roles control access to an item's data; they are separate from workspace roles.
Usage scenarios
Explore your Fabric environment
- "Search for items described as 'Sales Revenue'"
- "Find lakehouses with 'Customer' in the name"
- "List all my Fabric workspaces"
- "Show me all semantic models in the Sales Analytics workspace"
- "What lakehouses exist in my development environment?"
Create and manage resources
- "Create a workspace called Sales Analytics Dev"
- "Create a lakehouse named CustomerData in the Dev workspace"
Manage permissions
- "Who has Admin access to Sales Analytics?"
- "Add john@contoso.com as a Contributor to my Dev workspace"
Inspect OneLake shortcuts and security
- "List the shortcuts in the Sales lakehouse and show their targets. Don't change them."
- "Show the OneLake data access roles for the Sales lakehouse, including their permissions and members."
- "Read the OneLake diagnostics settings for the Sales Analytics workspace."
Automate workflows
- "Create a new workspace with a lakehouse, then add my team as Contributors"
- "Find all workspaces where I'm an Admin and list their capacity assignments"
Limitations
- Operations use the authenticated identity's Fabric permissions
- Some advanced Fabric features may not have corresponding MCP tools yet
- Resource-management tools aren't general-purpose tools for querying or modifying data in lakehouse tables. For workload-specific data operations, see Choose a Microsoft Fabric MCP server.
Considerations
Security
MCP is a developing standard. Review the security of systems that integrate with MCP servers, including MCP hosts, clients, agents, AI applications, and models. Confirm that they meet your system requirements, standards, and expectations.
Follow Microsoft security guidance for identity and access control, secure token management, and network security. See Security in Microsoft Fabric. For shared authentication and auditability guidance, see Security and compliance.
Permissions and risk
MCP clients can invoke operations based on the authenticated identity's Fabric role-based access control (RBAC) permissions. Autonomous or misconfigured clients may perform destructive actions.
Review and apply least-privilege RBAC roles and implement safeguards before deployment. Certain safeguards, such as flags to prevent destructive operations, aren't standardized in the MCP specification and might not be supported by all clients. Review agent-generated tool calls and your client's approval settings before allowing changes.
For OneLake changes, review the target workspace and item, shortcut conflict policy, and role definitions before approval. Replacing the complete role set can remove existing access. Storage-tier changes can incur charges, and diagnostic-log immutability restricts changes to retained logs. See the OneLake management safeguards.
Configure authentication by following Get started with Fabric Core MCP Server.
Compliance responsibility
This MCP server may interact with clients and services that process data outside Fabric's compliance boundaries. Data is processed in accordance with your chosen client's or service's applicable terms and data handling policies.
You are responsible for ensuring that any integration complies with your applicable organizational, regulatory, and contractual requirements.
Next step
Related content
- Fabric MCP Servers overview — Compare Core and local servers
- Core MCP tools reference — Complete tools list
- Model Context Protocol specification
- Microsoft Graph MCP Server