pdf-official

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The guide contains examples that embed plaintext passwords and pass them as command-line arguments (e.g., --password=... and writer.encrypt("userpassword", ...)), which are insecure patterns that would require the model to output secret values verbatim if followed.

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

• Third-party content exposure detected (medium risk: 0.30). In the PDF form-filling workflow (forms.md → fill_pdf_form_with_annotations.py / scripts/extract_form_field_info.py), outsider-authored text is read from the user-provided fields.json/PDF inputs at runtime (e.g., text = entry_text["text"] and then embedded into PDF annotations), so the LLM can ingest free text supplied via those inputs.