tooluniverse-clinical-guidelines
Pass
Audited by Gen Agent Trust Hub on Mar 15, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: No malicious instructions or patterns to bypass safety filters or override system behavior were found. The instructions are focused on guiding the agent through tool-assisted clinical search tasks.
- [DATA_EXFILTRATION]: No unauthorized access to local sensitive files or hardcoded credentials was detected. All network operations are directed towards reputable clinical databases.
- [REMOTE_CODE_EXECUTION]: The skill utilizes standard library calls for the tooluniverse package. It does not include patterns for downloading and executing untrusted code or arbitrary command execution.
- [EXTERNAL_DOWNLOADS]: The skill retrieves information from authoritative medical organizations such as NICE, WHO, and the NCCN. These sources are well-known and trustworthy for the skill's primary purpose.
- [SAFE]: The skill exhibits standard and safe behaviors for a multi-source data retrieval tool, incorporating necessary validation steps and clear output structuring requirements.
Audit Metadata