The Agent Skills Directory

[COMMAND_EXECUTION]: The skill facilitates the execution of the orca command-line interface and its variants (orca-ide, orca-dev). It provides a resolution strategy for the executable path using environment variables like ORCA_CLI_COMMAND and ORCA_DEV_REPO_ROOT, or system context checks. These commands are intended for managing project state, terminals, and embedded browser controls.
[PROMPT_INJECTION]: The skill introduces a surface for indirect prompt injection by instructing the agent to load its primary instruction set from the output of a shell command (ORCA skills get orca-cli). (1) Ingestion points: The agent is directed to read the output of the orca-cli sub-command in SKILL.md to obtain its version-matched guide. (2) Boundary markers: There are no explicit delimiters or instructions provided to the agent to distinguish between the CLI's legitimate guide content and potentially malicious instructions embedded in that output. (3) Capability inventory: The agent is granted the capability to execute shell commands (via the resolved ORCA binary) based on the instructions it receives. (4) Sanitization: The skill does not specify any sanitization, validation, or filtering of the content returned by the binary before the agent processes it as instructions.

orca-cli