Fabric Real-Time Intelligence MCP

Fabric Real-Time Intelligence MCP

MCP server for Microsoft Fabric RTI enabling LLM-driven access to Eventhouse, Eventstreams, and Activator.

71
Stars
53
Forks
5
Releases

Overview

This MCP server provides a comprehensive Model Context Protocol (MCP) implementation for Microsoft Fabric Real-Time Intelligence (RTI). It acts as a bridge between AI agents and Fabric RTI services by exposing RTI capabilities as MCP tools, enabling data querying, analysis, and streaming workflows through natural language interfaces. Supported services include Eventhouse (Kusto) for executing KQL queries, Eventstreams for real-time data processing, and Activator triggers for real-time alerting. The server uses Azure Identity (DefaultAzureCredential) for secure, seamless authentication and supports both local HTTP-mode testing and remote deployment. The toolset is extensive: 12 Eventhouse (Kusto) tools, 17 Eventstreams tools (with core and builder operations), and 2 Activator tools for artifact listing and trigger creation. NL-to-KQL translation and smart parameter suggestions provide a unified interface for analytics and streaming workloads. Configurability is via environment variables (e.g., KUSTO_SERVICE_URI, KUSTO_SERVICE_DEFAULT_DB, FABRIC_API_BASE_URL) with optional AZ_OPENAI_EMBEDDING_ENDPOINT for semantic search. The README also documents OBO token exchange, HTTP mode, and remote deployment guidance, noting that the project is in public preview and subject to change.

Details

Owner
Microsoft
Language
Python
License
MIT License
Updated
2025-12-07

Features

MCP Protocol bridge

Exposes Fabric RTI capabilities as MCP tools for AI agents to orchestrate queries and workflows.

Natural Language to KQL

Translates natural language prompts into KQL queries and Eventstream configurations for RTI services.

Secure Azure Identity authentication

Uses DefaultAzureCredential to securely access Fabric RTI resources without manual token handling.

Real-time data access

Direct integration with Eventhouse and Eventstreams enabling live analytics and streaming analysis.

Unified analytics and streaming interface

Single MCP interface supports both analytics (Kusto) and streaming (Eventstreams) workloads with intelligent prompts.

Extensive Kusto tooling

12 Eventhouse tools and 17 Eventstreams tools (plus 2 Activator tools) for broad RTI capabilities.

Semantic search support

Optional Azure OpenAI embedding endpoint enables semantic search via kusto_get_shots.

HTTP mode and deployment support

Supports HTTP mode for local testing and provides remote deployment guidance.

Audience

Data ScientistsUse MCP to query Kusto and analyze data with natural language prompts and embeddings.
AI EngineersBuild NL-to-KQL workflows and RTI tooling for live data and streaming scenarios.
BI AnalystsQuery ADX/Eventhouse and monitor streams via MCP for analytics workflows.
Fabric RTI DevelopersExtend and integrate RTI workflows using the MCP server for customized solutions.
DevOpsConfigure secure access, token handling (Azure Identity), and HTTP deployment for testing and production.

Tags

Microsoft Fabric RTIEventhouseEventstreamsActivatorKustoADXMCPLLMAzure IdentityReal-timeHTTP modeSemantic searchOBOAzure OpenAI