power-bi-business-analysis
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill incorporates strong defensive design patterns by enforcing mandatory 'Approval Gates' (defined in
SKILL.mdandplanning-part-02.md). It explicitly prevents the agent from executing implementation tools, calling Fabric write APIs, or publishing reports until the user has reviewed and approved a locked specification document. - [INDIRECT_PROMPT_INJECTION]: The skill processes external data via Microsoft Learn tools and user-provided business requirements. This attack surface is mitigated by: 1) Ingestion points: Microsoft Docs search and user interviews. 2) Boundary markers: Strict requirement document templates. 3) Capability inventory: Access to Power BI CLIs and Fabric APIs. 4) Sanitization: The 'GREENFIELD APPROVAL BARRIER' ensures all processed data undergoes human review before impacting downstream actions.
- [EXTERNAL_DOWNLOADS]: The skill references official developer tools and services from well-known organizations, specifically the Azure CLI (
az), Microsoft Learn resources, and Node.js-based Power BI CLIs. These references are legitimate and consistent with the skill's professional purpose.
Audit Metadata