The Agent Skills Directory

[SAFE]: The skill implements a multi-model code review process. Analysis of the instructions and reference templates shows a standard development workflow focused on software quality and maintainability.
[COMMAND_EXECUTION]: The skill uses git diff to identify code changes. This is a standard and expected operation for a code review tool within a development environment.
[DYNAMIC_EXECUTION]: The skill spawns subagents (Reviewer A, B, C, D) using a provided Task tool. This is the intended architecture of the platform and does not involve executing untrusted code strings or arbitrary remote scripts.
[DATA_EXPOSURE_AND_EXFILTRATION]: While the skill reads local files and diffs, it does so within the agent's execution context to perform reviews. There are no network exfiltration patterns or requests to external, non-whitelisted domains.
[PROMPT_INJECTION]: The reviewer-prompt.md and SKILL.md contain instructions for the AI subagents. These are well-defined role-play instructions ('You are an adversarial code reviewer') intended to shape the review quality rather than bypass safety filters or ignore system instructions.

interrogate