uncertainty-and-units
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: No patterns attempting to bypass safety filters, override system instructions, or extract restricted information were detected.
- [DATA_EXPOSURE_AND_EXFILTRATION]: No network operations, hardcoded credentials, or access to sensitive system paths (like .ssh or .aws) were found. The skill includes explicit checks to reject network URLs and symbolic links for both input and output files.
- [REMOTE_CODE_EXECUTION]: No remote script downloads or piped execution patterns (e.g., curl|bash) were detected. All dependencies (pint, uncertainties, numpy, scipy) are well-known, established scientific packages.
- [DYNAMIC_EXECUTION]: Mathematical expressions are evaluated using a custom AST walker in
scripts/_common.py. This implementation uses a strict whitelist for AST nodes and functions (e.g., math.sin, math.exp), effectively preventing the execution of arbitrary Python code. - [COMMAND_EXECUTION]: The skill does not use subprocesses or system calls to execute external commands. All logic is implemented using the standard library and the declared scientific dependencies.
Audit Metadata