The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or unauthorized data access were identified. The skill is entirely instructional and operates within a limited, read-only scope for code analysis.
[COMMAND_EXECUTION]: The skill allows the use of narrow, read-only commands such as git inspection and targeted file reads to verify implementation details. This behavior is standard for the stated purpose of code review and does not pose a security threat.
[PROMPT_INJECTION]: The skill's instructions to adopt an 'adversarial' and 'skeptical' posture are task-specific guidelines intended to improve review quality. These do not attempt to bypass the AI agent's core safety filters or override system-level constraints.

review-implementation