The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is designed to execute shell commands using the orca binary and its variants (orca-ide, orca-dev) based on the environment configuration.
[DYNAMIC_EXECUTION]: The skill instructs the agent to dynamically load its operational guide by running ORCA skills get orca-cli. This allows the agent's logic and available subcommands to be defined at runtime by the output of a local binary.
[INDIRECT_PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection by ingesting and following instructions provided by the CLI output without validation or boundary enforcement.
Ingestion points: The output generated by the ORCA skills get orca-cli command (SKILL.md).
Boundary markers: None are present; the agent is directed to read the output and proceed with further commands based on that content.
Capability inventory: Access to terminals, worktree management, internal browser control, and public artifact sharing.
Sanitization: No sanitization or filtering is applied to the dynamically fetched instructions.
[DATA_EXFILTRATION]: The skill includes functionality for creating "public artifact links" and sharing HTML or Markdown content, which presents a potential risk of sensitive data exposure if used to share internal or protected information.

orca-cli