Skip to content

v2.1.4 - NL2KQL Accuracy, Schema Type Normalization & Log Redaction

Choose a tag to compare

View all tags
@4R9UN 4R9UN released this 11 Jun 08:22
· 4 commits to main since this release
v2.1.4
d014cb8
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

MCP KQL Server v2.1.4

Sharpens NL2KQL generation accuracy and hardens logging. No changes to the public tool contract - drop-in upgrade from v2.1.3.

Highlights

Canonical schema type normalization (NL2KQL accuracy)
Schema discovery now collapses every column type to a single canonical KQL vocabulary (datetime, string, bool, int, long, real, decimal, guid, timespan, dynamic) via a new normalize_kql_type() helper, applied across all ingestion paths. Previously, different paths stored inconsistent spellings (e.g. Int64, DateTime, System.String), so data-type-aware generation (time filters, aggregations, grouping) silently missed columns. CSL ColumnType/CslType is now preferred over the .NET DataType.

Credential redaction in logs
New redact_secrets() helper masks user:pass@host, password=/token: pairs, and Bearer <jwt> tokens before any raw query or connection string is logged.

Leaner dependencies & single-source version

  • Removed unused scikit-learn and authlib; pinned pandas>=1.3.0,<3.0.0.
  • Package version now has a single literal source (constants.py), re-exported by __init__.py and enforced for parity with pyproject.toml and server.json in CI.

Tooling

  • Hardened the CI version-consistency check to read the literal source and cross-check server.json.
  • Standardized author/email headers across modules.

Install / Upgrade

pip install --upgrade mcp-kql-server==2.1.4

Authentication unchanged: both query execution and schema discovery use your az login token via Azure CLI authentication.

Assets 2
Loading