The Agent Skills Directory

[SAFE]: The skill implements a technical framework for code auditing, utilizing read-only tools like Read, Grep, and Glob to examine source files without the risk of unauthorized system modifications.
[SAFE]: External references are limited to documentation and public repositories for attribution, with no evidence of remote code execution or suspicious dependencies.
[PROMPT_INJECTION]: The skill ingests untrusted code data (ingestion point) which could potentially contain malicious instructions. While it lacks explicit boundary markers or sanitization, the risk is minimized by the read-only capability inventory and the requirement to follow a structured output template for review reports.

code-reviewer