legal-ocr
Warn
Audited by Socket on Sep 19, 2026
1 alert found:
AnomalyAnomalyscripts/mineru_ocr.py
LOWAnomalyLOW
scripts/mineru_ocr.py
No clear evidence of intentional malware in this snippet; it is primarily an integration client for MinerU document conversion. However, it creates meaningful security exposure by (1) extracting an untrusted remotely downloaded ZIP with zipfile.ZipFile.extractall() without path traversal validation (ZIP Slip risk) and (2) downloading arbitrary https?:// image URLs found in untrusted Markdown without allowlisting or robust safety controls. These issues warrant mitigation (ZIP member path validation, download domain/size restrictions, content-type checks) before use in adversarial or semi-trusted environments.
Confidence: 66%Severity: 68%
Audit Metadata