The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes standard Git commands (git diff, git rev-parse, git log) and the GitHub CLI (gh) to determine the scope of changes and collect run metadata for the review report.
[COMMAND_EXECUTION]: It leverages common infrastructure linting and security tools if they are available on the local system, including tflint, tfsec, checkov, kubeconform, kube-linter, hadolint, and helm. These are used according to their intended purpose to identify misconfigurations.
[EXTERNAL_DOWNLOADS]: The documentation and checklists in reference.md are based on best practices from well-known technology organizations such as HashiCorp, Docker, and the Cloud Native Computing Foundation (CNCF). The skill explicitly prohibits downloading or installing these tools if they are missing, preventing unverified remote code execution.
[PROMPT_INJECTION]: The instructions are strictly instructional and focused on the review workflow. No patterns designed to bypass safety filters or override system instructions were detected.
[DATA_EXFILTRATION]: While the skill accesses repository content and utilizes the GitHub CLI, its network activity is limited to standard developer workflows for fetching PR data. There is no evidence of sensitive data being sent to unauthorized third-party domains.

review-infrastructure