cicd-intelligent-recovery

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly instructs agents to fetch and analyze open/public third‑party content—e.g., the "Failure Pattern Researcher 1" task in Step 2 directs researching "GitHub issues for libraries we use" and "Stack Overflow questions" (and Step 1 downloads GitHub workflow logs), and those external findings are used to drive synthesis, root‑cause decisions, and automated fixes, creating a clear vector for indirect prompt injection.