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.
Model Context Protocol (MCP) in Real-Time Intelligence (RTI) enables AI models, AI agents, and applications to interact with Fabric RTI components using natural language.
The Model Context Protocol (MCP) provides a standardized way for AI models, like Azure OpenAI models, to discover and use external tools and data sources. MCP makes it easier to build intelligent applications that can query, reason, and act on real-time data. MCP also makes it easier for AI agents to find, connect to, and use enterprise data.
Fabric's Real-Time Intelligence provides two types of MCP servers: local and remote. Each option has different deployment models, capabilities, and use cases.
Local MCP server for RTI
The local MCP server for Fabric Real-Time Intelligence is an open-source server that you install, host, and manage yourself. It runs on your local machine and provides read-only access to Fabric RTI and Azure Data Explorer (ADX) resources.
Key characteristics:
- Deployment: Self-hosted on your local machine
- Source: Open-source on GitHub
- Access: Read-only queries to Eventhouse, Eventstream, Map, and Azure Data Explorer (ADX) clusters.
- Management: You manage installation, updates, and maintenance
For detailed information, see Get started with the local MCP server.
Remote MCP servers
Microsoft hosts remote MCP servers and exposes them as HTTP endpoints. You configure your MCP client to connect to these servers without installing or managing any software.
| Server | Description | Capabilities |
|---|---|---|
| Eventhouse MCP server | Enables AI agents to query Eventhouse using natural language | Schema discovery, KQL query generation, data sampling, natural language to KQL translation |
| Activator MCP server | Enables AI agents to interact with Fabric Activator | Create monitoring rules, manage alerts, trigger actions |
MCP Host: The environment where the AI model (like GPT-4, Claude, or Gemini) runs.
MCP Client: An intermediary service forwards the AI model's requests to MCP servers, like GitHub Copilot, Cline, or Claude Desktop.
MCP Server: Small applications that make specific features accessible to AI models, such as running database queries. For example, Fabric RTI MCP server can execute KQL queries for real-time data retrieval from KQL databases.
When to use local vs. remote servers
Natural Language Interfaces: Ask questions in plain English or other languages, and the system turns them into optimized queries (NL2KQL- Natural Language to Kusto Query Language).
| Scenario | Recommended option |
|---|---|
| Query Eventhouse or ADX data with full control over the server | Local MCP server |
| Query Eventhouse without managing server infrastructure | Remote Eventhouse MCP |
| Create monitoring rules and alerts in Activator | Remote Activator MCP |
| Use in cloud agent platforms like Copilot Studio or Azure AI Foundry | Remote MCP servers |
| Need offline or air-gapped access | Local MCP server |
| Want automatic updates and maintenance | Remote MCP servers |
Supported AI clients
Both local and remote MCP servers work with popular AI clients:
Supported RTI components
Eventhouse - Run KQL queries against the KQL databases in your Eventhouse backend. This unified interface lets AI agents search your real-time data, analyze patterns, and take actions based on what they find.
Note
You can also use the Fabric RTI MCP Server to run KQL queries against the clusters in your Azure Data Explorer backend.
Considerations and Limitations
Security
MCP as a phenomenon is very novel and cutting-edge. As with all new technology standards, consider doing a security review to ensure any systems that integrate with MCP servers follow all regulations, and standards your system is expected to adhere to. This review includes not only the RTI MCP servers, but any MCP client/agent that you choose to implement down to the model provider.
You should follow Microsoft security guidance for MCP servers, including enabling Entra ID authentication, secure token management, and network isolation. Refer to Microsoft Security Documentation for details.
Permissions and Risk
MCP clients can invoke operations based on the user’s Fabric Role-Based Access Control (RBAC) permission. Autonomous or misconfigured clients may perform destructive actions. You should review and apply least-privilege RBAC roles and implement safeguards before deployment. Certain safeguards, such as flags to prevent destructive operations, are not standardized in the MCP specification and may not be supported by all clients.
Compliance Responsibility
This MCP server may be installed, used and share data with clients and services, such as third party LLMs, AI agents or services that operate outside Fabric’s compliance boundaries. You are responsible for ensuring that any integration complies with applicable organizational, regulatory, and contractual requirements.