cohort-study-quality-assessment-nos
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external cohort study text and PDF documents, creating an interface for indirect prompt injection.
- Ingestion points: PDF files via
scripts/extract_pdf.pyand study text articles provided directly by users. - Boundary markers: The skill includes specific instructions to evaluate inputs against strict criteria located in
references/nos_criteria.md. - Capability inventory: Execution of local scripts (
scripts/calculate_nos_score.py), writing toextracted_text.txt, and PDF text extraction via the well-knownPyPDF2library. - Sanitization: The documentation contains specific input validation rules and JSON escaping instructions for command-line arguments to mitigate processing errors.
Audit Metadata