low-level-code-review
Pass
Audited by Gen Agent Trust Hub on Jul 22, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [SAFE]: The skill is designed to perform localized code reviews, looking for mechanical mistakes like numeric overflows, null pointer dereferences, and typos.
- [COMMAND_EXECUTION]: The skill executes
git diffand related git commands to obtain the source code changes for analysis. These are standard developer tools used within the intended scope of the skill. - [PROMPT_INJECTION]: The skill ingests external data (source code diffs) which could potentially contain malicious instructions (Indirect Prompt Injection). This risk is categorized as low and mitigated by the skill's narrow instructions to identify specific bug patterns and avoid high-level reasoning or following ambiguous intent.
Audit Metadata