The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The instructions explicitly direct the agent to "re-run it with escalated permissions" if a command fails with a runtime_access_denied error due to sandbox blocks. This is a direct attempt to bypass security boundaries established by the execution environment.
[DYNAMIC_EXECUTION]: The skill functions as a "discovery stub" that does not contain its own logic. Instead, it instructs the agent to load task-specific guides at runtime using the ORCA skills get orca-cli command. This dynamic loading of instructions means the agent's actual behavior is determined by content not present in the static skill definition.
[COMMAND_EXECUTION]: The skill executes local binaries (orca, orca-ide, or orca-dev) and determines which executable to use based on environment variables like ORCA_CLI_COMMAND and ORCA_DEV_REPO_ROOT. It also includes instructions to start services using ORCA open if they are not already running.

orca-cli