pyhealth
Pass
Audited by Gen Agent Trust Hub on Jun 27, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: No malicious instruction patterns such as safety bypasses, role-play injections (DAN), or system prompt extraction attempts were found.
- [DATA_EXFILTRATION]: The skill processes clinical data using local file paths. No evidence of sensitive credential harvesting (e.g., SSH keys, API tokens) or unauthorized network exfiltration was detected.
- [REMOTE_CODE_EXECUTION]: The skill uses standard Python libraries and legitimate package installation commands. No patterns of remote script execution or untrusted data piping to shells were found.
- [SAFE]: The skill provides thorough, legitimate documentation for a recognized healthcare AI library (PyHealth). All reviewed code examples follow standard data science practices and do not exhibit malicious behavior.
Audit Metadata