The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands on the host system, specifically the built-in macOS screencapture utility for acquiring images and rm for file deletion. It also suggests using osascript to interface with System Events to retrieve window identifiers.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes screenshots which represent untrusted visual data from the user's environment. This data could contain malicious instructions, text, or UI elements designed to manipulate the agent's behavior at runtime.\n
Ingestion points: Image files created by the screencapture tool as described in the Core Workflow section of SKILL.md.\n
Boundary markers: None; the agent is expected to interpret the multimodal visual content directly without delimiters.\n
Capability inventory: Access to screencapture, osascript, and rm as documented in SKILL.md.\n
Sanitization: No validation or sanitization of the screen content is performed before processing.

screencapture-mac