chain-llm-pattern

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

• Third-party content exposure detected (high risk: 0.90). The workflow clearly ingests untrusted user-generated email content via the "Webhook" (customer-email-triage) which is passed as $json.body.email_content into the Stage 1 Information Extractor and then used to drive Stage 2 classification and Stage 3 routing/score decisions, so third-party content can materially influence agent actions.