machine-element-sizing
Pass
Audited by Gen Agent Trust Hub on Nov 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides a controlled environment for mechanical engineering calculations. The bundled script
scripts/me_sizing.pyuses only standard Python libraries (math, json, argparse) and contains no network, file-system modification, or dynamic code execution logic. - [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user parameters (e.g., loads, dimensions, materials) which are interpolated into command-line arguments to drive the calculator script.
- Ingestion points: User prompts requesting the sizing or verification of machine elements as specified in
SKILL.md. - Boundary markers: Instructions explicitly direct the agent to collect inputs in SI units and state them clearly before executing the script, which helps structure the data flow.
- Capability inventory: The skill's only significant capability is executing the local
scripts/me_sizing.pyscript. This script outputs JSON data to stdout and does not interact with any sensitive system resources or external networks. - Sanitization: The calculation script utilizes the
argparselibrary to strictly enforce types (float, int) and predefined choices (e.g., specific bolt sizes or material keys), which effectively mitigates the risk of command injection through the parameters.
Audit Metadata