extracting-with-ocr

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The doc references a CLI flag --vlm-api-key for a hosted VLM, which encourages passing an API key on the command line and can require embedding the secret verbatim in commands or outputs.

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

• Third-party content exposure detected (medium risk: 0.30). SKILL.md (“Extracting with OCR”) describes OCR extraction from user-supplied scanned PDFs/images (e.g., xberg extract scan.pdf --force-ocr=true), which at runtime is free text authored by an outsider contained in those documents.