The Agent Skills Directory

[COMMAND_EXECUTION]: The skill employs the Bash tool to execute local Python scripts (e.g., generate_survival_analysis.py, biomarker_classifier.py) and LaTeX compilation commands. These scripts use the argparse module for parameter handling and do not utilize unsafe functions like eval() or os.system() with untrusted input.
[EXTERNAL_DOWNLOADS]: The skill requires standard, well-known scientific libraries including pandas, numpy, scipy, lifelines, matplotlib, and pyyaml. These are standard dependencies for clinical data analysis and are expected to be sourced from official package registries.
[DATA_EXFILTRATION]: No evidence of unauthorized network activity or data exfiltration was found. The inclusion of scripts/validate_cds_document.py, which uses regular expressions to check for HIPAA identifiers (SSNs, MRNs, phone numbers, and names), serves as a robust defense against accidental data exposure.
[PROMPT_INJECTION]: The instructions and templates within SKILL.md and the assets/ directory focus on clinical data interpolation and professional LaTeX formatting. There are no instructions designed to override system safety protocols, extract system prompts, or induce unrestricted AI behavior.

clinical-decision-support