talk-to

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly reads and reacts to messages in Oracle threads (oracle_threads() and oracle_thread_read() in the autonomous "loop" workflow), meaning it ingests user-generated/untrusted thread content and uses those responses to drive follow-up actions and decisions, enabling indirect prompt injection.