mcp-kql

The MCP KQL Server provides an AI-assisted bridge between natural language queries and Azure Data Explorer (KQL). It accepts plain-English questions and converts them into optimized KQL queries using schema discovery, AI-powered caching, and live schema validation. With features like execute_kql_query and schema_memory, you can generate, validate, and run KQL against your clusters, retrieve results in JSON, CSV, or tabular formats, and obtain context-aware results and visualizations. The server also supports on-demand schema discovery to keep its memory in sync with your data sources, improving accuracy over time.

Prerequisites:

• Python 3.10 or higher
• Access to an Azure Data Explorer cluster (for live query execution)
• Python package manager (pip)

One-command installation (recommended):

pip install mcp-kql-server

From source (optional):

git clone https://github.com/4R9UN/mcp-kql-server.git && cd mcp-kql-server
pip install -e .

Usage (example):

# Run the MCP server (as a background or foreground process depending on your setup)
python -m mcp_kql_server

Notes:

• The server automatically sets up memory directories and sane defaults for production use.
• No additional environment variables are strictly required, but you can configure Azure connections and PyPI behaviors via standard Python environment variables if needed.

Tips and tips:

• Ensure your Python environment has network access to PyPI and your Azure Data Explorer cluster.
• If you encounter authentication or connectivity issues, verify Azure CLI login and cluster accessibility.
• The server uses on-demand schema discovery to cache table schemas; this improves query accuracy over time.
• Check RELEASE_NOTES.md for details on v2.1.0 improvements like schema-only NL2KQL and auto-update detection.
• For debugging, run with verbose logs or adjust logging level in your environment to capture execution traces.