research

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

• Third-party content exposure detected (high risk: 0.80). The skill uses a market index service (createMarketIndexService / indexService.sync() and indexService.search) and returns fields like entry.description, platform, and probabilities from indexed markets — i.e., public/user-generated market listings — which the agent reads and uses to inform base-rate and resolution decisions, exposing it to untrusted third-party content.