alibabacloud-opc-advisor

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

• Third-party content exposure detected (medium risk: 0.30). Source: user-provided free text is ingested at runtime through the skill’s required “Entry Probe” and subsequent sizing/inference questions (routing A1/A2 and then asking Q-scale/Q-userdata from the same user message content), e.g., the workflow directly consumes whatever the user types like “我用 Cursor 写了…” to infer tier and PII/AI signals.