The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The scripts/agent.py script references the standard requests library for HTTP communication with the Microsoft Graph API. This dependency is documented in the accompanying references.
[DATA_EXFILTRATION]: The agent performs network requests to official Microsoft authentication and Graph API endpoints (login.microsoftonline.com and graph.microsoft.com) to retrieve Conditional Access policy data for analysis.
[INDIRECT_PROMPT_INJECTION]: The skill processes data fields from Microsoft Entra ID (such as policy display names) which originate from an external environment. If the generated audit reports are consumed by an agent without proper input handling, they could serve as an injection vector.
Ingestion points: Azure AD policy configurations and named locations are retrieved via the Microsoft Graph API in scripts/agent.py.
Boundary markers: The generated report does not include explicit markers to separate data from instructions.
Capability inventory: The skill is capable of performing network operations and writing output files to the local system.
Sanitization: Content is serialized as JSON, providing structural integrity but not filtering for potential prompt injection patterns in natural language strings.

implementing-conditional-access-policies-azure-ad