The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection and command injection through the processing of external data files. 1. Ingestion points: The utility scripts scripts/user_management.py, scripts/tenant_setup.py, and scripts/powershell_generator.py ingest user-provided data from external JSON and CSV files. 2. Boundary markers: The instructions do not provide robust boundary markers or safety warnings for the agent to validate external data before generating scripts. 3. Capability inventory: The skill generates scripts for high-privilege administrative actions across Microsoft 365, including user provisioning, security policy enforcement, and mailbox management. 4. Sanitization: Input data is directly interpolated into PowerShell script templates using f-strings without escaping special characters, enabling potential command injection.
[COMMAND_EXECUTION]: The skill centers on generating and executing powerful PowerShell commands that operate with Global Administrator privileges, representing a significant risk if the generated scripts are compromised via malicious input.
[DYNAMIC_EXECUTION]: The skill generates PowerShell (.ps1) scripts at runtime based on Python templates and user input. While script generation from templates is a standard pattern for automation, the lack of input validation in the templates creates a security weakness for administrative workflows.

ms365-tenant-manager