meta-ads

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly instructs fetching and analyzing public third‑party ad content (e.g., SKILL.md’s "Auditing an account" flow that runs scripts/get_active_ads_copy.py to "Pull active ads, classify by OS rules, produce recommendations" and creative-cadence-operating-system.md’s "Meta Ad Library Process" to visit facebook.com/ads/library), meaning the agent ingests untrusted user-generated/public web content and uses it to drive decisions.