clinical-decision-support
Pass
Audited by Gen Agent Trust Hub on Nov 8, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements a local-only execution model. Python scripts are restricted to processing local JSON files and explicitly prohibit network access, environment variable reading, and external API calls.
- [DATA_EXFILTRATION]: No data exfiltration capabilities were found. Scripts contain defensive measures against processing Personally Identifiable Information (PII), such as the PERSON_LEVEL_KEYS filter in scripts/_common.py which rejects identifiers like ssn, mrn, and date_of_birth.
- [REMOTE_CODE_EXECUTION]: No evidence of remote code execution or dynamic code execution was found. The skill relies exclusively on the Python standard library and does not use eval, exec, or subprocess modules.
- [PROMPT_INJECTION]: The skill documentation includes extensive Hard Safety Boundaries and Stop Conditions designed to prevent the AI from providing clinical advice or patient-specific diagnoses.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted local data. While this creates a surface for indirect prompt injection, it is mitigated by structural validation of inputs and explicit instructions to the AI to ignore instructions embedded in data.
Audit Metadata