skills/edenfunf/moltbot/peekaboo/Gen Agent Trust Hub

peekaboo

Pass

Audited by Gen Agent Trust Hub on Jun 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill interfaces with the peekaboo CLI utility to perform extensive system-level automation on macOS. This includes simulating mouse clicks, drags, and scrolls; injecting keyboard sequences and text; reading and writing to the system clipboard; and executing automation scripts stored in JSON files.
  • [EXTERNAL_DOWNLOADS]: The skill metadata specifies the installation of the peekaboo binary from a public Homebrew tap on GitHub (steipete/tap/peekaboo). This installation is required for the skill's functionality.
  • [PRIVILEGE_ESCALATION]: To function, the Peekaboo tool requires the user to grant macOS 'Accessibility' and 'Screen Recording' permissions, which are high-privilege entitlements necessary for UI automation and visual inspection.
  • [DATA_EXPOSURE]: The tool accesses sensitive data by design, including the ability to capture screenshots, record screen regions, and read all content currently stored in the system clipboard.
  • [PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection, as it is designed to read untrusted content from the user's screen or clipboard and act upon it.
  • Ingestion points: Screen content is read via peekaboo see and peekaboo image; clipboard content is read via peekaboo clipboard.
  • Boundary markers: None; the skill does not include instructions to the agent to treat UI-captured text as untrusted or to ignore embedded instructions.
  • Capability inventory: Full UI control capabilities including input simulation, application management, and command execution via the CLI.
  • Sanitization: No sanitization or validation of data captured from the screen or clipboard is mentioned.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 17, 2026, 03:35 AM