The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local commands using the orca binary or its derivatives (orca-dev, orca-ide) to manage project state, including worktrees and terminals.
[DATA_EXPOSURE]: The skill reads environment variables ORCA_CLI_COMMAND and ORCA_DEV_REPO_ROOT to resolve the correct executable path.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface. Ingestion points: Instructions are retrieved from the output of ORCA skills get orca-cli as specified in SKILL.md. Boundary markers: None provided to distinguish command output from system instructions. Capability inventory: Full access to orca subcommands across all scripts, including worktree, terminal management, and browser control. Sanitization: No escaping or filtering is performed on the command output before it is interpolated into the agent context.

orca-cli