| name | powerbi-ai-model-graph-validator |
| description | Use when a Power BI semantic model relationship graph must be checked for ambiguous paths, fact-to-fact chains, wrong filter direction, unsupported topology, DirectQuery/OLAP limitations, or certification risk. |
Power BI AI Model Graph Validator
Use this skill before publishing or opening generated TMDL models.
Validation Areas
- Star-schema topology: facts filter to dimensions through many-to-one relationships.
- Ambiguous paths between facts and dimensions.
- Fact-to-fact relationships and accidental bridge paths.
- Many-to-many risk and bridge-table requirements.
- Active/inactive relationship strategy and role-playing dates.
- Cross-filter direction and bidirectional filter risks.
- Unsupported DirectQuery, composite model, or live/OLAP edits.
Output Requirements
- Include relationship inventory, invalid paths, severity, and remediation.
- Mark models as
pass, pass-with-risk, or fail.
Executable Support
- Use
scripts/powerbi_expert_factory.py validate --project <pbip-folder> for the current local static model-graph gate.
- Treat any
model_graph error as release blocking.
