The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The instructions explicitly tell the agent to "re-run it with escalated permissions" if the sandbox blocks a connection (indicated by a runtime_access_denied error). This is a direct attempt to bypass security boundaries and operating system constraints.
[COMMAND_EXECUTION]: The skill's primary operation involves executing local binaries (orca, orca-dev, or orca-ide). It provides logic for selecting which executable to run based on environment variables like ORCA_CLI_COMMAND and ORCA_DEV_REPO_ROOT.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to load its actual operational guide at runtime using the command ORCA skills get orca-cli.
Ingestion points: Output of the orca command in SKILL.md.
Boundary markers: None present; the agent is instructed to load and follow this "version-matched guide" as its primary source of truth.
Capability inventory: The skill description indicates capabilities including terminal access, repo management, worktree operations, and browser control via the orca CLI.
Sanitization: None. The content loaded from the binary is treated as authoritative instructions for the agent.

orca-cli