The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/get_model_info.py executes system commands via the subprocess module to interface with the Azure CLI (az) and a specialized API tool (fab). These operations are used to retrieve authentication tokens and metadata directly from Power BI workspace endpoints for auditing purposes.
[EXTERNAL_DOWNLOADS]: The skill documentation in references/performance.md and references/ai-readiness.md references official tools and repositories from the Microsoft GitHub organization (microsoft/fabric-toolbox). These are trusted sources for Power BI and Fabric development resources.
[PROMPT_INJECTION]: The skill defines a workflow for analyzing user-supplied metadata, including field descriptions and 'AI instructions' within the semantic model. While this represents a potential surface for indirect prompt injection, the skill's capabilities are focused on reporting and read-only analysis, which maintains a low risk profile. Evidence:
Ingestion points: Reads table and column descriptions and AI-specific instructions from the model metadata (TMDL files).
Boundary markers: Not explicitly defined in the prompts or scripts for the data being analyzed.
Capability inventory: Execution of CLI-based metadata collection via subprocess in scripts/get_model_info.py and report generation.
Sanitization: None explicitly defined for metadata strings before they are processed by the agent.

review-semantic-model