unlimited-ocr-document-parsing

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

• Third-party content exposure detected (medium risk: 0.30). In scripts/client.py, runtime OCR text is ingested from outsider-supplied file_url (public HTTPS URL) and downloaded at parse_baidu() lines 211-214 and _download() lines 178-190, then the returned markdown_url/parse_result_url are downloaded and decoded into LLM-ingested free text at lines 261-267.