drawing-analyzer

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

• Third-party content exposure detected (high risk: 0.85). Outsider free text can enter the LLM context via the runtime-extracted page text from the user-supplied merged construction PDF: split_pdf.py uses page.get_text("text") to write text/sheet_NNN.txt, and Mode B then reads those text/sheet_NNN.txt files into the agent for answering.