macpilot-automation

Warn

Audited by Gen Agent Trust Hub on Mar 5, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill includes functionality to execute arbitrary shell commands via macpilot shell run. This allows the agent to perform any action a user could in the Terminal, including file system modifications or system changes.
  • [DATA_EXFILTRATION]: The skill provides tools to read sensitive system information, specifically the clipboard content using macpilot clipboard get, and network details such as local IP addresses and Wi-Fi names.
  • [PROMPT_INJECTION]: The skill possesses a broad surface for indirect prompt injection as it processes untrusted data from the system environment.
  • Ingestion points: Text read from the clipboard, UI labels and values retrieved via ui find-text, and the output of executed shell commands.
  • Boundary markers: None. The skill instructions do not include delimiters or warnings to treat system-retrieved data as untrusted.
  • Capability inventory: Includes keyboard/mouse emulation, application management, arbitrary command execution, and notification control.
  • Sanitization: There are no instructions for the agent to sanitize or validate data retrieved from the clipboard or UI before processing it or using it to drive subsequent actions.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 5, 2026, 09:45 AM