The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions direct the agent to resolve and execute a local command-line interface (CLI) tool (e.g., orca, orca-ide, orca-dev). It guides the agent through running subcommands like status, open, skills get, and computer list-apps to manage desktop application state and retrieve further instructions.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external applications via accessibility trees and screenshots. This data is inherently untrusted and could contain hidden instructions intended to override the agent's behavior.
Ingestion points: Accessibility trees, application lists, and UI state read via the orca CLI commands from the local desktop environment (documented in SKILL.md).
Boundary markers: None explicitly defined in this discovery stub; the agent relies on its own instruction boundaries when processing the CLI output.
Capability inventory: The skill allows the agent to list apps, read visible UI, click controls, type, and set values in other desktop applications.
Sanitization: No specific sanitization or filtering is described for the content read from external application windows.

computer-use