accreditation-desk-review

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

• Third-party content exposure detected (medium risk: 0.30). In scripts/deskreview.py, the extract runtime path walks the user-supplied MA evidence folder and LLM-ingests extracted free text from outsider-authored .pdf/.docx/.xlsx/.txt files (and treats unreadable scans as gaps), so an outsider can submit poison via that folder upload path.