review-pr

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 1.00). This skill fetches and ingests user-generated PR and issue content via gh pr view and gh issue view (see "Phase 1: Fetch PR Data" and "Phase 2: Extract Ticket Context") and then passes PR_DIFF, PR_BODY and TICKET_CONTENT inline to a subagent for review and potential GitHub posting, so untrusted third-party content can materially influence agent actions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill uses the GitHub CLI at runtime (gh pr view, gh pr diff, gh issue view) to fetch PR body and diff from the GitHub API (https://api.github.com), and those fetched contents are injected verbatim into the code-reviewer agent prompt—an external, required dependency that can directly control the agent's instructions.