skills/aipoch/medical-research-skills/quapas-quality-assessment-for-prognosis-studies/Gen Agent Trust Hub
quapas-quality-assessment-for-prognosis-studies
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from user-supplied PDF documents, creating a potential vector for instructions embedded within the text to influence agent behavior.
- Ingestion points:
scripts/extract_pdf.pyreads user-provided PDF files for text extraction. - Boundary markers: Prompt templates in
references/quapas_prompts.mduse[Paper]and[Format]tags to structure the input, providing basic delimitation. - Capability inventory: The
scripts/extract_pdf.pyscript performs local file reads and writes. - Sanitization: The extracted text is passed to the analysis prompts without explicit sanitization or safety filtering.
Audit Metadata