qa-loop

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

• Third-party content exposure detected (medium risk: 0.30). The QA loop ingests “incoming issue” evidence as text/screenshots/recordings/logs from the user-supplied queue (“Phase 3: Intake and Queue Issues” and “Phase 4: Reproduce and Diagnose”), and explicitly treats embedded text/page content/console output/network payloads as untrusted evidence that the workflow reads at runtime.