om-code-review
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data from repository diffs and issue trackers. It mitigates injection risks by explicitly instructing the agent to treat all repository content as data rather than instructions. It further directs the agent to identify and report any embedded directives (e.g., 'ignore previous instructions') as suspected prompt injection attempts.
- [COMMAND_EXECUTION]: The skill performs shell command execution via a 'validation gate' defined in the repository's configuration. This is the primary function of the skill to mirror CI processes. The instructions minimize risk by restricting execution to operator-vouched configurations and requiring regex-based sanitization of arguments before interpolation.
- [DATA_EXFILTRATION]: The skill establishes strict 'Secrets hygiene' rules. It requires the agent to redact API keys, tokens, and credentials from all reports, plans, and logs, ensuring that sensitive information is not surfaced to the LLM output or external trackers.
- [SAFE]: The instructions demonstrate high security awareness, specifically addressing the risks of path traversal, SQL injection, and unsafe input processing within the code review checklist, while also applying these principles to its own operational logic.
Audit Metadata