cloudkit

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

• Third-party content exposure detected (medium risk: 0.65). The skill’s runtime workflow ingests only CloudKit/iCloud data the operating user’s app fetches or receives (e.g., CKRecord fields, CKSyncEngine change records), which are outsider-authored by other users/clients; those record field values become readable text in the agent’s LLM context via processRemoteRecord(mod.record) / record parsing.