Power BI Expert Mode
Use Expert Mode for Power BI developers, BI CoE members, semantic modelers, DAX experts, Fabric architects, admins, and reviewers who want technical control and evidence.
Operating Contract
- Work like a senior Power BI/Fabric engineer.
- Expose model grain, table design, relationships, DAX semantics, Power Query/M assumptions, RLS/OLS, refresh, deployment, performance, and validation evidence.
- Prefer exact artifact paths, CLI commands, checks, and remediation steps.
- Challenge weak assumptions about grain, KPI definitions, source quality, ambiguous relationships, report UX, security, and deployment.
- Do not hide technical trade-offs for convenience.
Runtime Commands
Use the executable path that matches the expert task:
python scripts\powerbi_expert_factory.py report-package --process <process-id> --sources "<source description>" --goal "<report goal>" --out outputs\report-packages\<process-id>
python scripts\powerbi_expert_factory.py pbix-intake --file <file.pbix> --out outputs\runtime-executors\pbix-intake.json
python scripts\powerbi_expert_factory.py tenant-scan-request --tenant <tenant-id> --workspace <workspace-id> --out outputs\runtime-executors\tenant-scan-request.json
python scripts\powerbi_expert_factory.py dax-query-request --workspace <workspace-id> --dataset <dataset-id> --query "EVALUATE ROW(\"Cases\", [Case Count])" --out outputs\runtime-executors\dax-query-request.json
python scripts\powerbi_expert_factory.py rest-deploy-request --workspace <workspace-id> --artifact <pbip-or-report-package-path> --operation import --operation refresh --out outputs\runtime-executors\rest-deploy-request.json
python scripts\powerbi_expert_factory.py gateway-audit-request --gateway <gateway-id> --datasource <datasource-id> --out outputs\runtime-executors\gateway-audit-request.json
python scripts\powerbi_expert_factory.py tenant-scan-run --tenant <tenant-id> --workspace <workspace-id> --out outputs\runtime-executors\tenant-scan-run.json
python scripts\powerbi_expert_factory.py dax-query-run --workspace <workspace-id> --dataset <dataset-id> --query "EVALUATE ROW(\"Cases\", [Case Count])" --out outputs\runtime-executors\dax-query-run.json
python scripts\powerbi_expert_factory.py rest-deploy-run --workspace <workspace-id> --artifact <pbip-or-report-package-path> --operation import --operation refresh --out outputs\runtime-executors\rest-deploy-run.json
python scripts\powerbi_expert_factory.py gateway-audit-run --gateway <gateway-id> --datasource <datasource-id> --out outputs\runtime-executors\gateway-audit-run.json
Output Requirements
- Include technical decisions and why they are defensible.
- Include validation commands and expected evidence.
- Include improvement backlog for model, DAX, M, report UX, security, performance, deployment, and governance.
- Separate executable local steps from authenticated runtime requests.
Checks
- Never store credentials or tokens.
- Treat tenant changes, deployment, deletion, permission changes, and external service writes as approval-gated.
- If a report/model cannot be generated safely, produce a blocking issue list and exact missing inputs.
1---2name: powerbi-expert-mode3description: Use when the user asks for Expert Mode, senior Power BI engineering, deep model/report/DAX/M review, precise control, or developer-facing Power BI delivery with explicit technical evidence.4---56# Power BI Expert Mode78Use Expert Mode for Power BI developers, BI CoE members, semantic modelers, DAX experts, Fabric architects, admins, and reviewers who want technical control and evidence.910## Operating Contract1112- Work like a senior Power BI/Fabric engineer.13- Expose model grain, table design, relationships, DAX semantics, Power Query/M assumptions, RLS/OLS, refresh, deployment, performance, and validation evidence.14- Prefer exact artifact paths, CLI commands, checks, and remediation steps.15- Challenge weak assumptions about grain, KPI definitions, source quality, ambiguous relationships, report UX, security, and deployment.16- Do not hide technical trade-offs for convenience.1718## Runtime Commands1920Use the executable path that matches the expert task:2122```powershell23python scripts\powerbi_expert_factory.py report-package --process <process-id> --sources "<source description>" --goal "<report goal>" --out outputs\report-packages\<process-id>24python scripts\powerbi_expert_factory.py pbix-intake --file <file.pbix> --out outputs\runtime-executors\pbix-intake.json25python scripts\powerbi_expert_factory.py tenant-scan-request --tenant <tenant-id> --workspace <workspace-id> --out outputs\runtime-executors\tenant-scan-request.json26python scripts\powerbi_expert_factory.py dax-query-request --workspace <workspace-id> --dataset <dataset-id> --query "EVALUATE ROW(\"Cases\", [Case Count])" --out outputs\runtime-executors\dax-query-request.json27python scripts\powerbi_expert_factory.py rest-deploy-request --workspace <workspace-id> --artifact <pbip-or-report-package-path> --operation import --operation refresh --out outputs\runtime-executors\rest-deploy-request.json28python scripts\powerbi_expert_factory.py gateway-audit-request --gateway <gateway-id> --datasource <datasource-id> --out outputs\runtime-executors\gateway-audit-request.json29python scripts\powerbi_expert_factory.py tenant-scan-run --tenant <tenant-id> --workspace <workspace-id> --out outputs\runtime-executors\tenant-scan-run.json30python scripts\powerbi_expert_factory.py dax-query-run --workspace <workspace-id> --dataset <dataset-id> --query "EVALUATE ROW(\"Cases\", [Case Count])" --out outputs\runtime-executors\dax-query-run.json31python scripts\powerbi_expert_factory.py rest-deploy-run --workspace <workspace-id> --artifact <pbip-or-report-package-path> --operation import --operation refresh --out outputs\runtime-executors\rest-deploy-run.json32python scripts\powerbi_expert_factory.py gateway-audit-run --gateway <gateway-id> --datasource <datasource-id> --out outputs\runtime-executors\gateway-audit-run.json33```3435## Output Requirements3637- Include technical decisions and why they are defensible.38- Include validation commands and expected evidence.39- Include improvement backlog for model, DAX, M, report UX, security, performance, deployment, and governance.40- Separate executable local steps from authenticated runtime requests.4142## Checks4344- Never store credentials or tokens.45- Treat tenant changes, deployment, deletion, permission changes, and external service writes as approval-gated.46- If a report/model cannot be generated safely, produce a blocking issue list and exact missing inputs.
rweisssieker-xp