The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires the agent to resolve and run local executables, including orca, orca-ide, or a command specified in the ORCA_CLI_COMMAND environment variable.
[INDIRECT_PROMPT_INJECTION]: The skill instructions direct the agent to fetch its primary logic at runtime using ORCA skills get orchestration. This creates a dependency on the output of a local shell command to define subsequent agent behavior and task coordination.
Ingestion points: The output of the ORCA skills get orchestration command (SKILL.md).
Boundary markers: None identified in the provided stub.
Capability inventory: The skill description grants the agent capabilities for shell commands, terminal control, and worktree management (SKILL.md).
Sanitization: No specific sanitization or validation of the fetched instructions is described.
[DYNAMIC_EXECUTION]: The agent is instructed to "load the version-matched guide" before running further commands, which involves executing a tool to generate instructions that the agent must then interpret and follow during the session.

orchestration