clinical-trial-logistics
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill contains conceptual Python code for clinical trial management. Analysis of the logic, including randomization and accountability systems, reveals no malicious patterns, obfuscation, or unauthorized access attempts.
- [INDIRECT_PROMPT_INJECTION]: A theoretical ingestion surface exists in SKILL.md via patient and site identifiers processed by the provided Python classes. Evidence chain: 1) Ingestion points: patient_id, site_id, and reason parameters. 2) Boundary markers: Absent. 3) Capability inventory: No network, file, or shell capabilities found across all scripts. 4) Sanitization: Absent. The lack of sensitive capabilities renders this surface non-exploitable.
Audit Metadata