The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses local shell commands to interact with the Grok CLI (grok models, grok login, mktemp). These operations are the primary purpose of the skill and include instructions for manual user intervention for sensitive tasks like authentication.
[PROMPT_INJECTION]: The skill acts as a bridge for passing user-provided prompts to an external model (Grok). It implements mitigations by suggesting the use of files (prompt.md) via apply_patch instead of direct shell interpolation to prevent command injection at the system level.
[DATA_EXFILTRATION]: While the skill communicates with an external service (Grok), it explicitly instructs the agent not to pass secrets or sensitive environment data in the prompts.

grok-build-cli