krt-word-illuminator

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

• Third-party content exposure detected (medium risk: 0.30). In the mandatory runtime workflow, the agent reads and extracts free text from user-supplied DOCX contents via scripts/create_docx.py → lib/worddoc.create_from_request()/inspect_docx() (and similarly for edits/render), treating the document’s extracted paragraphs and XML text as untrusted input at runtime.