The Agent Skills Directory

Command Execution via External Tooling: The skill recommends using npx fabric-app-data to execute DAX queries against semantic models. This pattern involves running a command-line utility, which is a standard method for interacting with these data sources.
Indirect Prompt Injection Surface: The skill processes metadata retrieved from external models, such as table and column names. Because this data is incorporated into the agent's context, there is a consideration for how instructions embedded within model metadata might affect agent behavior.
Ingestion points: Metadata is retrieved through INFO.VIEW functions defined in references/discovery-queries.md.
Boundary markers: The skill does not currently include specific instructions for the agent to ignore or delimit instructions that might be present in the metadata.
Capability inventory: The skill uses shell command execution via the fabric-app-data utility as seen in SKILL.md.
Sanitization: The instructions do not specify a validation or sanitization step for the retrieved metadata values.

schema-discovery