grok-subagent
Pass
Audited by Gen Agent Trust Hub on Aug 23, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill orchestrates the use of the
grokandjqcommand-line tools to perform software implementation tasks and process results. - [PROMPT_INJECTION]: As a delegation tool, the skill manages the risk of malicious or incorrect implementation by the subagent through a structured review process. It requires the parent agent to verify all changes, run independent tests, and specifically forbids the subagent from performing high-risk actions like Git commits or package publication.
- [DATA_EXFILTRATION]: The instructions include a specific warning to keep credentials, tokens, and private data out of the prompts passed to the implementation tool, mitigating the risk of data leakage to external subagents.
- [REMOTE_CODE_EXECUTION]: The skill utilizes the Grok CLI for automated code changes but enforces the use of safety parameters like
--no-subagentsand--permission-mode autoto limit the autonomous capabilities of the subagent.
Audit Metadata