The Agent Skills Directory

[SAFE]: The skill implements a read-only review process using standard tools like git, gh, and grep. It explicitly prohibits editing, staging, or committing changes, ensuring a safe operation environment.
[SAFE]: External references are limited to local files (e.g., ENGINEERING_STANDARDS.md in the plugin root) and a vendored checklist included with the skill. There are no remote downloads or executions from untrusted sources.
[SAFE]: The checklist.md contains security guidance (e.g., flagging SQL injection, shell injection, and race conditions), but these are instructions for the agent to look for in the code being reviewed, not malicious behaviors of the skill itself.
[SAFE]: Network operations are restricted to standard git and gh (GitHub CLI) commands for fetching repository metadata and diffs, which are appropriate for a code review tool.

basic-review