powerskills-workiq

Pass

Audited by Gen Agent Trust Hub on Nov 11, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The install action in workiq.ps1 executes npm install -g @microsoft/workiq. This downloads and installs an external tool from a well-known organization's official registry.
  • [COMMAND_EXECUTION]: The Action-Raw function in workiq.ps1 provides an 'escape hatch' that allows the agent to execute arbitrary subcommands of the workiq CLI by passing a string of arguments. While execution is restricted to the Work IQ binary, this provides a broad interface to Microsoft Graph capabilities beyond the specifically wrapped actions.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data that can be used to influence agent behavior.
  • Ingestion points: The ask and fetch actions ingest content from Outlook emails, Teams messages, and SharePoint documents (files: SKILL.md, workiq.ps1).
  • Boundary markers: The skill does not implement boundary markers or instructions to ignore embedded commands within the retrieved data.
  • Capability inventory: The skill has significant write and send capabilities, including do-action (send mail), create (create entities), and delete (delete entities) (file: workiq.ps1).
  • Sanitization: The skill strips ANSI escape sequences from CLI output but does not sanitize or escape the natural language content of messages or documents before returning them to the agent.
  • [DATA_EXFILTRATION]: By design, the skill accesses highly sensitive organizational data, including private messages and calendar events. While no malicious exfiltration code is present, the skill's purpose is to move this data into the agent's context where it could be processed or potentially forwarded by other skills.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 11, 2026, 06:29 AM