code-review
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFENO_CODE
Full Analysis
- NO_CODE (SAFE): The skill consists entirely of markdown and instructions for the AI agent. There are no scripts, binaries, or configuration files that execute commands on the host system.
- PROMPT_INJECTION (SAFE): The instructions do not contain markers intended to bypass safety guardrails or extract system prompts. The language is focused on the intended task of code analysis.
- DATA_EXFILTRATION (SAFE): No network operations (curl, wget, etc.) or sensitive file path access (SSH keys, credentials) are present in the skill.
- INDIRECT_PROMPT_INJECTION (LOW): The skill is designed to process untrusted user code. While it lacks explicit boundary markers to delimit user input from the review instructions, the skill has no dangerous capabilities (like file writing or command execution) to exploit if an injection occurred.
- REMOTE_CODE_EXECUTION (SAFE): There are no patterns involving the download or execution of remote scripts or packages.
Audit Metadata