MCP KQL Server
Ask your agent to run Kusto (KQL) against Azure Data Explorer clusters using natural language and discovered schema.
Overview
MCP KQL Server is a MCP server for the Operate phase that runs KQL against Azure Data Explorer from agent prompts via NL2KQL and schema discovery.
What is this MCP server?
- Execute KQL from AI prompts with NL2KQL translation
- Schema discovery so agents do not guess table and column names
- Azure Data Explorer (ADX) integration for real cluster queries
- PyPI stdio package mcp-kql-server v2.1.3
- Combines database analytics with agent-driven investigation workflows
- Server version 2.1.3
- PyPI stdio package identifier mcp-kql-server
- MCP schema dated 2025-10-17
What problem does it solve?
You waste cycles translating English questions into KQL and hunting ADX schema while debugging production data.
Who is it for?
Builders who already use Azure Data Explorer for logs or metrics and want NL-assisted KQL from Claude Code or Cursor.
Skip if: Projects with no Azure/ADX footprint or teams that forbid agent-executed queries on production clusters.
What do I get? / Deliverables
Your agent can discover schema and execute vetted KQL in ADX so you answer operational questions faster from the chat.
- Agent-invoked KQL queries against configured ADX resources
- Schema discovery output usable in follow-up prompts
- NL2KQL-assisted operational investigations from the IDE
Recommended MCP Servers
Journey fit
ADX and KQL are primarily used after launch to query logs, metrics, and operational datasets in production. Monitoring is the canonical shelf because NL2KQL and schema discovery support incident triage and telemetry investigation.
How it compares
ADX query MCP bridge, not a general SQL MCP or a standalone dashboard product.
Common Questions / FAQ
Who is MCP KQL Server for?
Solo builders and operators on Azure who query ADX with KQL and want their MCP agent to draft and run queries with schema awareness.
When should I use MCP KQL Server?
Use it during incidents, performance investigations, or routine telemetry checks when you need NL2KQL and live ADX results inside the agent.
How do I add MCP KQL Server to my agent?
Install the PyPI package mcp-kql-server (v2.1.3), configure stdio MCP in your client, and connect ADX cluster credentials per the GitHub repo.