The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/generate_schematic.py uses subprocess.run to execute scripts/generate_schematic_ai.py. This is used to modularize the logic for AI-powered schematic generation. The command is passed as a list of arguments rather than a shell string, which is the recommended security practice to mitigate shell injection vulnerabilities.
[EXTERNAL_DOWNLOADS]: The skill requires standard, well-known Python libraries for its core functionality, including pandas, numpy, scipy, lifelines, and matplotlib. These are reputable packages from the official PyPI registry, widely trusted in the scientific community.
[DATA_EXFILTRATION]: scripts/generate_schematic_ai.py communicates with openrouter.ai, which is recognized as a well-known service for AI model connectivity. The script uses an API key supplied via environment variables (OPENROUTER_API_KEY) to authenticate requests for image generation and review. This behavior is entirely consistent with the skill's stated purpose of providing publication-quality medical schematics.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a data ingestion surface through patient cohort CSV files and user-defined schematic descriptions. 1. Ingestion points: scripts/biomarker_classifier.py and scripts/create_cohort_tables.py ingest CSV data; scripts/generate_schematic.py takes CLI prompts. 2. Boundary markers: The LLM prompts in scripts/generate_schematic_ai.py provide clear instructions but lack explicit boundary markers for user-supplied content. 3. Capability inventory: Includes local command execution via subprocess.run, file system writes for LaTeX/PNG, and network requests via requests.post. 4. Sanitization: The scripts/validate_cds_document.py tool provides automated quality and compliance checks on the final generated documents, including HIPAA de-identification verification.

clinical-decision-support