marketing-loops

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

• Third-party content exposure detected (medium risk: 0.30). The required reply-triage, social-listening, brand-mention, launch-reaction, voice-of-customer, review-monitoring, community and newsjacking loops all read outsider-authored free text (inbound prospect replies, public posts, reviews, support feedback) from feeds the agent monitors without first selecting a specific item.

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

• Potentially malicious external URL detected (high risk: 0.70). Operational dependency: the skill directs the agent to read and follow behavior-governing procedures (scheduling rules, weekly ads check, reply classes, signal rules, dunning copy, intent-clustering and checklist method) from github.com/Quegenx/agent-skills at runtime, and that repository's ownership and distribution are not established by credible evidence, so the tier is weak/unknown and high.