clay

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly downloads and ingests public, user-generated models and metadata (e.g., via Output Routing entries "download/fetch model/sketchfab/objaverse" and the required references/api-integration.md which implements Sketchfab, Objaverse, PolyPizza, Smithsonian download/search flows), meaning untrusted third‑party content is fetched and used in the agent’s workflow and can affect provider selection, validation, and follow-up actions.