case-control-study-quality-assessment-nos
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted text from user-provided PDF files via scripts/extract_pdf.py, creating a surface for indirect prompt injection where malicious instructions could be embedded in the source documents. 1. Ingestion points: User-provided PDF files processed in scripts/extract_pdf.py. 2. Boundary markers: Explicit instructions to quote original text and provide reasons in Chinese are present to help distinguish source text. 3. Capability inventory: Limited to local script execution for text extraction and JSON formatting. 4. Sanitization: Input validation rules documented in SKILL.md guide the refusal of out-of-scope requests.
- [COMMAND_EXECUTION]: The skill instructs the agent to execute local Python scripts (scripts/extract_pdf.py and scripts/format_nos_table.py) to perform text extraction and data formatting as part of its primary workflow.
- [EXTERNAL_DOWNLOADS]: The skill references an external GitHub repository (github.com/aipoch/medical-research-skills) for medical research source material and task-specific guidance.
Audit Metadata