gh-pr-review-fix

SUSPICIOUS. The stated purpose is coherent for PR remediation, and the scanner's command-injection hits are false positives from markdown references. However, the skill's critical path depends on `codex-dev`, a third-party CLI not verified as official OpenAI or GitHub tooling, and it is entrusted with repo context plus hosted PR closeout actions. The main risk is disproportionate transitive trust in an unverifiable external tool rather than overt malicious behavior in the skill text itself.