word-document-processor

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly unpacks and reads arbitrary user-supplied .docx files (see the "Reading/analyzing content" workflows, pandoc text extraction and ooxml/scripts/unpack.py and the Document class in ooxml.md), so the agent will ingest and interpret untrusted, third-party document content as part of its workflow.