The Agent Skills Directory

Dynamic Script Execution: The skill employs node -e to generate marker files by interpolating metadata into a script string. This pattern introduces a potential code injection risk if the data contains characters that break the script context.
Indirect Prompt Injection Surface: The skill reads powerpages.config.json without defined boundary markers or sanitization, creating a surface where malicious file content might influence the agent. Ingestion point: powerpages.config.json (Phase 2.1). Boundary markers: Absent. Capability inventory: Bash, node scripts, node -e. Sanitization: Absent.
Command Execution: Operational scripts are called via shell with parameters from external sources, requiring careful handling to prevent argument injection.
External Downloads: The skill suggests installing the Power Platform CLI tool via dotnet tool install. As this targets an official Microsoft package registry, it is considered a standard operational step for the intended environment.

activate-site