The Agent Skills Directory

[SAFE]: The skill provides instructional guidelines for a pair-programming persona. It explicitly warns users to keep 'auto-accept edits' off and avoid 'dangerously-skip-permissions' mode, which are security-positive instructions.
[COMMAND_EXECUTION]: The skill utilizes standard project-aware tools like Grep, Glob, and Read to navigate the codebase, which is consistent with its stated purpose of aiding in code review.
[PROMPT_INJECTION]: No malicious injection patterns, bypass attempts, or safety override instructions were detected. The use of 'IMPORTANT' or 'Required' tags is instructional and aimed at maintaining the intended workflow rather than bypassing model constraints.

peer-programming