jhzhu89/mcp-server-azure-kusto

MCP Server for Azure Data Explorer

An intelligent Azure Data Explorer access layer designed for AI toolchains. Unlike basic API wrappers, this server provides comprehensive schema discovery and context, enabling AI tools to understand your data structure and generate precise, accurate KQL queries.

Features

• Schema-Enhanced Discovery: Comprehensive table and function schema exposure for AI context
• Intelligent Query Support: Rich metadata enables precise KQL query generation
• Function Intelligence: Complete stored function discovery with parameter definitions
• AI-Optimized Integration: Schema-aware responses designed for AI toolchains
• Flexible Authentication: OAuth2 delegation and application modes with automatic caching

Tools

Core Tools

• run-query: Execute KQL queries with intelligent result management
• list-functions: Discover stored functions with metadata
• get-function-schema: Retrieve function signatures and parameters
• list-databases: Browse available databases

Setup

1. Install dependencies:

bun install

2. Configure authentication:

# Quick start with Azure CLI (for local development)
az login

# Set cluster URL in tool calls or environment
export AZURE_AUTH_MODE="application"  # default

3. Build and run:

bun run build
bun run start

Configuration

Authentication Setup

The server uses @jhzhu89/azure-client-pool for intelligent client management with automatic caching and credential fallback.

Application Mode (Default - Recommended)

export AZURE_AUTH_MODE="application"
export AZURE_APPLICATION_AUTH_STRATEGY="chain"  # cli → managed identity fallback

Delegated Mode (For user-context scenarios)

export AZURE_AUTH_MODE="delegated"
export AZURE_CLIENT_ID="your-app-client-id"
export AZURE_TENANT_ID="your-tenant-id"
export AZURE_CLIENT_SECRET="your-client-secret"

Advanced Configuration Sources

# Option A: Azure App Configuration (Centralized)
export AZURE_APPCONFIG_ENDPOINT="https://your-appconfig.azconfig.io"
export AZURE_APPCONFIG_KEY_PREFIX="azure_kusto_mcp:"

# Option B: Environment variables (shown above)

Query Behavior

• QUERY_WARNING_THRESHOLD: Row count warning (default: 1000)
• QUERY_SOFT_LIMIT: Soft truncation limit (default: 5000)
• QUERY_HARD_LIMIT: Hard truncation limit (default: 50000)

Schema-Enhanced Intelligence

The key advantage over basic API wrappers: deep understanding of your Kusto tables through comprehensive schema discovery.

• Rich Schema Context: Provides detailed table schemas, column types, and function signatures to AI tools
• Intelligent Query Generation: AI can write more accurate KQL queries with full knowledge of your data structure
• Function Discovery: Exposes stored functions with parameter definitions for precise execution
• Context-Aware Responses: Results formatted with schema awareness for better AI comprehension

This schema intelligence enables AI tools to understand your data model and generate precise, executable queries rather than generic templates.

Authentication

The server leverages @jhzhu89/azure-client-pool for sophisticated authentication with automatic caching and credential management.

How it works:

• Application Mode: Uses service credentials (CLI/Managed Identity) with automatic fallback
• Delegated Mode: On-behalf-of flow preserving user identity and permissions
• Smart Caching: Clients cached by authentication context and cluster configuration
• Token Management: Automatic refresh and JWT validation for delegated flows

Supported Methods: Azure CLI, Managed Identity, Client Secret, Certificate, Interactive Browser

Comparison with Other MCP Implementations

Unlike basic Kusto MCP implementations, this server provides:

Key Advantages:

• Schema-Enhanced AI: Provides complete data model context for accurate query generation
• Function Intelligence: Discovers and exposes stored functions with full parameter definitions
• Contextual Responses: All outputs include schema context for improved AI understanding

Development

# Watch mode
bun run dev

# Build with watch  
bun run dev:build