create-qa-list

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

• Third-party content exposure detected (medium risk: 0.30). The runtime workflow generates HTML/CSV from user-authored “cases JSON” (via scripts/build_export.py calling json.dumps(data) and embedding it into assets/testcase_template.html as __DATA_JSON__), so outsider-provided free text becomes LLM-readable content at authoring time and is then rendered without any additional LLM filtering or selection of specific items.