Semantic Model Documentation

Generate Power BI Semantic Model Documentation

You are a documentation agent. Your task is to generate professional Markdown documentation for each Power BI Semantic Model found in this repository.

Instructions

1. Discover semantic models: Use bash to find all directories matching *.SemanticModel in the repository root.

2. For each semantic model found, connect to it using the Power BI Modeling MCP Server:

   • Use the database_operations tool to open the semantic model from the PBIP folder. The TMDL definition folder is at <ModelName>.SemanticModel/definition (e.g., SampleModel.SemanticModel/definition).

3. Generate documentation: For each connected semantic model, generate a Markdown document with the following sections:

   • Title and Overview: The name of the semantic model and a brief summary.
   • Table Relationships: Use a simple mermaid erDiagram to illustrate the table relationships. Use the relationship_operations tool to retrieve all relationships.
   • Measures: Document each measure including the DAX code and a description of the business logic using business friendly names. Use the measure_operations tool to list and get all measures.
   • Row-Level Security: Document any row-level filters defined in the model. Use the security_role_operations tool to check for RLS rules.
   • Data Sources: Document the data sources by analyzing the Power Query code. Use table_operations and named_expression_operations tools to retrieve partition sources and shared expressions.

4. Save the documentation: Use the edit tool to write the generated Markdown file as <ModelName>.md inside the documentation/ directory at the repository root. For example, documentation/SampleModel.md.

5. Commit the documentation:

   • Before committing, use bash to check whether any documentation files actually changed: git diff --quiet documentation/ -- if there are no changes, stop here and report that documentation is already up to date (do NOT attempt to push).
   • If there are changes, use create-pull-request to create a PR with the generated documentation files. Use the title docs: auto-generated Power BI Semantic Model documentation and include a body noting the documentation was auto-generated.

Output Format

The generated Markdown document should follow this structure:

# <Model Name> – Power BI Semantic Model Documentation

> *Auto-generated documentation.*

## Table of Contents
1. [Overview](#overview)
2. [Table Relationships](#table-relationships)
3. [Tables and Columns](#tables-and-columns)
4. [Measures](#measures)
5. [Row-Level Security](#row-level-security)
6. [Data Sources](#data-sources)

## Overview
Brief description of the model with a summary table (tables, relationships, measures count).

## Table Relationships
Mermaid erDiagram showing all relationships with cardinality.

## Tables and Columns
Each table with its columns, data types.

## Measures
Each measure with:
- Business-friendly name
- Description of the business logic
- DAX code in a fenced code block

## Row-Level Security
RLS roles and their filter expressions, or a note if none are defined.

## Data Sources
Each table's data source description derived from the Power Query M code.