skills/elizaos/eliza/peekaboo/Gen Agent Trust Hub

peekaboo

Warn

Audited by Gen Agent Trust Hub on Feb 18, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [EXTERNAL_DOWNLOADS] (MEDIUM): The skill initiates the installation of a binary from a third-party Homebrew tap (steipete/tap/peekaboo). This source is external and does not belong to the predefined list of trusted organizations or repositories.
  • [COMMAND_EXECUTION] (MEDIUM): The skill is built around executing a powerful CLI tool that can drive system input, manage applications, and execute local script files (peekaboo run).
  • [DATA_EXFILTRATION] (LOW): The tool has the capability to read sensitive system data, including the clipboard (peekaboo clipboard) and screen contents (peekaboo image, peekaboo see). Although no automated external transmission is defined in the skill, the exposure of this data to an AI agent presents a risk of accidental or malicious data handling.
  • [PRIVILEGE_ESCALATION] (LOW): The tool requires 'Screen Recording' and 'Accessibility' permissions on macOS. While these are necessary for the tool's primary purpose, they grant significant control over the user's operating system.
  • [INDIRECT_PROMPT_INJECTION] (LOW): The skill creates a surface for indirect prompt injection by processing data from the UI that may be controlled by third parties (e.g., website content in a browser window).
  • Ingestion points: Screen captures and UI analysis via peekaboo see and peekaboo image --analyze, as well as clipboard reading.
  • Boundary markers: None identified in the skill instructions to separate UI-derived data from system instructions.
  • Capability inventory: Significant capabilities including click, type, app launch/quit, and run (script execution).
  • Sanitization: No evidence of sanitization or filtering of the captured UI text or clipboard data.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 18, 2026, 06:11 PM