The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses git log and git blame commands to differentiate between new changes and pre-existing code. These are standard read-only operations for a code review tool.
[DATA_EXPOSURE]: The skill generates an HTML report and saves it to /tmp/code-review-report.html. This is a local file system write for user consumption and does not involve network exfiltration.
[REMOTE_CODE_EXECUTION]: The skill includes guidelines for checking 'Dependency risk' in PRs, but it does not perform any package installations or remote script executions itself.
[INDIRECT_PROMPT_INJECTION]: The skill ingests PR descriptions and code diffs which are untrusted data. However, it implements a confidence filtering phase and structured reporting which acts as a logical boundary. The risk is assessed as safe given the specific use case of local code auditing.

review-pr