alibabacloud-ak-leak-incident-response

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

• Third-party content exposure detected (medium risk: 0.30). In scripts/ak_leak_investigation.py, the runtime ingests outsider-authored free text by reading userAgent, errorMessage, and other string fields directly from first-party Alibaba Cloud ActionTrail LookupEvents results (e.g., lookup_events() → _event_to_dict() → generate_report()), and these fields can originate from attacker-controlled inputs like HTTP user agents/messages.