The Agent Skills Directory

[SAFE]: The skill provides highly disciplined instructions for performing code reviews, emphasizing the importance of tracing values and confirming reachability to avoid noise and false findings.
[SAFE]: Interactions with external services are limited to well-known developer platforms (GitHub) via official command-line utilities (git, gh) for the specific purpose of code analysis.
[SAFE]: The skill maintains a clear security boundary by operating in a read-only mode by default, requiring explicit user flags (--fix or --comment) to perform environment modifications or network writes.
[SAFE]: No obfuscation, persistence mechanisms, or unauthorized privilege escalation attempts were detected in the skill instructions or supporting documentation.

code-review