clinical-reports
Pass
Audited by Gen Agent Trust Hub on Jun 26, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill processes untrusted clinical data and patient notes provided by the user. This creates an attack surface for indirect prompt injection (Category 8), where malicious instructions could be embedded within clinical reports to manipulate the agent's output or behavior.
- Ingestion points: User-provided clinical reports and patient documentation files processed by the agent.
- Boundary markers: The skill does not explicitly define or mandate the use of delimiters or 'ignore' instructions for the processed clinical content.
- Capability inventory: The skill is configured with 'allowed-tools: Read Write Edit Bash', enabling the agent to execute Python scripts and manage files.
- Sanitization: There is no evidence of specific sanitization or filtering logic intended to neutralize potential prompt injection attacks within the data processed by the scripts or the agent instructions.
- [COMMAND_EXECUTION]: The skill includes a suite of Python scripts in the 'scripts/' directory to automate clinical reporting tasks. These scripts (e.g., 'check_deidentification.py', 'validate_case_report.py', 'extract_clinical_data.py') are intended for execution via the 'Bash' tool. While the scripts utilize standard Python libraries for file parsing and regular expression matching, they allow the agent to perform automated file system operations.
Audit Metadata