resolve_pr_parallel

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly retrieves and processes "all PR comments and review threads" from GitHub (via commands like gh pr status and bin/get-pr-comments PR_NUMBER), which are user-generated, third-party content that the agent is expected to read and could contain indirect prompt-injection payloads.