The Agent Skills Directory

[SAFE]: The skill defines a structured, evidence-backed code review process that operates primarily as a read-only tool. It utilizes standard local environment tools for git and filesystem operations without engaging in suspicious network or command execution patterns.- [SAFE]: The instructions for external research and skill loading are designed to use platform-provided, configured tools (harness discovery) rather than arbitrary remote code downloads or unverified third-party dependencies.- [SAFE]: The skill actively promotes security best practices by including specific dimensions for injection, credential handling, and privacy in its review logic, which enhances the overall security posture of the development workflow it serves.

ultra-review