The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from Microsoft 365 services, creating a surface for indirect prompt injection attacks.
Ingestion points: The getInboxMessages function in scripts/setup_exchange.ts reads content from user mailboxes.
Boundary markers: No specific delimiters or instructions to ignore embedded content are implemented in the API retrieval logic.
Capability inventory: The skill possesses broad administrative capabilities including user management, Teams configuration, and email operations across all implementation scripts.
Sanitization: Content fetched from external APIs is processed without explicit sanitization or filtering for potential prompt injection patterns.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of standard and well-known libraries for Microsoft 365 integration.
Evidence: The installation instructions in references/m365_quickstart.md reference official packages such as @azure/identity and @microsoft/microsoft-graph-client from the NPM registry.
[COMMAND_EXECUTION]: The skill is designed to generate and facilitate administrative command execution within the Microsoft 365 environment.
Evidence: The skill documentation in SKILL.md specifies the capability to execute PowerShell commands and M365 CLI tools for automation tasks.

m365-admin