What problem does it solve? Changing a Power BI semantic model involves many failure points: authoring DAX against columns that don't exist, breaking Desktop's file-open with malformed TMDL, and shipping measures that return wrong numbers. This Skill sequences the full pipeline so each change is confirmed, authored, landed, validated offline, and proven against real data. ## Core Features & Use Cases - Five-step change pipeline: Confirm the model shape via a live Desktop connection, author with the right sub-skill (DAX, Power Query, relationships, date tables), land edits in TMDL, validate offline with TmdlSerializer, then prove correctness with DEFINE MEASURE queries against real data. - Desktop-breaking gotcha prevention: Enforces rules like requiring discourageImplicitMeasures for calculation groups and never leaving blank lines after /// description comments. - Use Case: When a user says "this measure is wrong" or "add a calculation group", the Skill coordinates connect-pbid, dax-standard, tmdl, and refresh-semantic-model so the .tmdl change opens cleanly in Desktop and returns correct numbers. ## Quick Start Change the model by adding a new measure to the Sales table in my local PBIP project and verify it returns the correct value.