device-profile-management

Pass

Audited by Gen Agent Trust Hub on Mar 10, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The PowerShell scripts Initialize-DeviceProfile.ps1 and Sync-DeviceProfile.ps1 dynamically execute local command-line utilities such as winget, scoop, npm, choco, git, node, python, and claude to verify installation status and retrieve versions.
  • [COMMAND_EXECUTION]: The Update-DeviceProfile function in SKILL.md accepts a script block as a parameter and executes it dynamically using the PowerShell call operator (&), representing a surface for dynamic code execution.
  • [DATA_EXFILTRATION]: The skill aggregates sensitive system information, such as OS version, CPU/RAM specifications, and full file system paths of installed applications. This data is written to a centralized 'Admin Root' directory intended for cloud synchronization (e.g., Dropbox or OneDrive), which moves local system metadata to external storage providers.
  • [PROMPT_INJECTION]: The skill processes untrusted data from profile.json and tool outputs. Ingestion points: profile.json (potentially synced from other devices) and tool version outputs. Capability inventory: File system writing, command execution via call operator, and system configuration querying via CIM. Sanitization: Absent. This creates a surface for indirect prompt injection where the agent might act upon data synchronized from external sources or captured from command outputs.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 10, 2026, 03:49 AM