The Agent Skills Directory

[PROMPT_INJECTION]: The skill possesses a potential surface for indirect prompt injection as it processes external directory data and incorporates it into agent-accessible reports.
Ingestion points: analyze_ad_security.ts (Graph API sign-in and user data), audit_privileged_groups.ps1 (Active Directory objects), and review_delegation.ps1 (AD ACL permissions).
Boundary markers: No specific delimiters or instructions to ignore embedded commands are used when presenting external data to the agent.
Capability inventory: The skill utilizes bash for validation commands and requires write access for reports. It performs file-writing in audit_privileged_groups.ps1 (HTML report via Out-File) and review_delegation.ps1 (HTML report via Out-File). It performs network operations via Microsoft Graph in analyze_ad_security.ts.
Sanitization: No sanitization or escaping of LDAP or Graph API data is performed before interpolation into markdown or HTML outputs.
[COMMAND_EXECUTION]: Executes administrative PowerShell and TypeScript code to audit Active Directory environments, including group membership enumeration and security descriptor analysis.
[EXTERNAL_DOWNLOADS]: Utilizes official and trusted administrative libraries from Microsoft and Azure (e.g., @azure/identity) as well as native Windows features (RSAT-AD-PowerShell). These references are consistent with the skill's stated purpose and target well-known, trusted services.

ad-security-reviewer