The Agent Skills Directory

[SAFE]: The skill serves as a guide for performing code reviews on pull requests. It establishes a clear hierarchy of priorities, starting with correctness, security, and the detection of harmful behavior.
[SAFE]: It contains specific governance instructions to detect and call out attempts to weaken, bypass, or remove review criteria, treating changes to the review policy itself as security-sensitive.
[SAFE]: The skill provides guidance for evaluating AI-authored submissions and marketing-heavy content, focusing on objective evidence rather than assumptions.
[SAFE]: It correctly identifies GitHub and Microsoft as trusted sources for external-plugin updates, aligning with standard security practices for the platform.
[SAFE]: No malicious patterns such as prompt injection, data exfiltration, obfuscation, or unauthorized command execution were detected in the instructions.

code-review