clinical-decision-support
Pass
Audited by Gen Agent Trust Hub on Jun 26, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill ingests clinical data from CSV and JSON files and interpolates it into LaTeX and TikZ templates.
- Ingestion points: Untrusted clinical data is loaded via CSV in biomarker_classifier.py and create_cohort_tables.py, and via JSON/Text in build_decision_tree.py.
- Boundary markers: The skill does not utilize delimiters or boundary instructions to isolate untrusted data from the document rendering logic.
- Capability inventory: The agent has access to the Bash, Write, and Edit tools to generate and modify local document files.
- Sanitization: The provided validate_cds_document.py script is designed to detect HIPAA identifiers but does not perform character escaping to prevent LaTeX command injection.
- [COMMAND_EXECUTION]: The skill uses the Bash tool to execute local Python scripts for statistical analysis and report generation.
- [SAFE]: The skill follows security best practices by including a dedicated validation script to identify and prevent the accidental inclusion of PII or HIPAA identifiers in generated research documents.
Audit Metadata