feedback-analysis

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

• Third-party content exposure detected (high risk: 1.00). The SKILL.md explicitly requires a mandatory "Multi-Channel Data Unification" step that ingests public user-generated sources (e.g., app store reviews, Twitter/X, Reddit, Trustpilot, social mentions) which the agent is expected to read and use to drive triage, churn scoring, and roadmap decisions, creating clear exposure to untrusted third-party content that could carry indirect prompt injection.