Power BI
SemanticOps MCP v3.9.119: Release Notes
Maxim Anatsko
#MCP#SemanticOps#Power BI#DAX#Semantic Model#Release Notes

SemanticOps MCP v3.9.119 is out, with HTML model reports, fully configurable DAX formatting, deeper storage diagnostics, flaky-test classification, combined Power Query commits, and new skills for DAX debugging and performance tuning.
New in v3.9.119
- HTML model reports — generate HTML alongside DOCX and PDF, including multiple formats from one request.
- Configurable DAX formatting — indentation, spacing, line breaks, casing, and other DAX formatting options are now fully configurable in one place.
- Deeper storage diagnostics — model, table, partition, and column detail are reported individually, so unavailable values are no longer shown as zero.
- Test stability classification — tests are now classified as stable, flaky, degraded, or lacking enough history, using persisted retry evidence.
- Combined Power Query commits — related table, partition, parameter, and query-group changes can be staged and committed together in one Desktop model commit.
- New skills — DAX debugging, DAX performance tuning, and safe semantic-model refactoring, with more precise guidance and evidence checks.

Improvements
- DAX dependency checks now use Power BI's calculation dependency metadata when available, with text analysis as a fallback.
- DOCX and PDF model reports use a refreshed visual design with vector relationship and lineage diagrams; A4 or Letter page size is selectable, A4 by default.
- PII and numeric masking recognizes more Chinese, Japanese, and Thai model terminology.
- Choose browser or device-code authentication for each Power BI Service connection; protected sessions are restored before asking you to sign in again.
- Policy previews and commits carry generation and plan details, preventing a stale preview from overwriting a newer policy state.

- Preview mode now works for individual semantic, schema, security, localization, query-group, and partition-refresh changes.
- Model discovery applies exact-name and substring filters consistently before pagination and includes calculated columns in column results.
- Model changes and refreshes distinguish changes that were not applied, applied, or left in an unknown state.
- New hash and aggregate baselines ignore incidental row-order changes and reject truncated captures.
- Tests, baselines, and run history stay bound to their model identity across migrations.

Coming soon in 4.0
A DAX measure quality scorecard: correctness, performance, maintainability, and style, scored together with the findings behind them.

Get it
➡️ Download: GitHub Releases
➡️ Website: semanticops.dev