The Agent Skills Directory

[SAFE]: The skill is a well-structured engineering tool that incorporates clear safety and privacy guidelines for its execution environment.\n- [COMMAND_EXECUTION]: The PowerShell script scripts/New-ComsolMotorNvhRun.ps1 creates directory scaffolds for simulation projects. It uses standard PowerShell cmdlets and employs ValidatePattern to sanitize user-provided case names, preventing directory traversal or injection.\n- [DYNAMIC_EXECUTION]: The skill workflow utilizes comsolcompile to build Java-based simulation models. This is an intended and documented requirement for the COMSOL Java API and does not involve executing arbitrary remote or untrusted code.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied CSV files for result validation using scripts/validate_nvh_exports.py.\n
Ingestion points: External CSV data (eigenfrequencies.csv, campbell_long.csv) read by the Python script.\n
Boundary markers: The workflow relies on strict code-level data validation rather than instructional delimiters.\n
Capability inventory: Local file-system management and model compilation within the COMSOL ecosystem.\n
Sanitization: The validator script implements robust checks, including finiteness verification for floats and positive-value constraints for frequencies, raising exceptions for malformed data.\n- [DATA_EXFILTRATION]: The instructions in SKILL.md and the validation gates in references/evidence-gates.md mandate the removal of sensitive identifiers, workstation paths, and license-server strings from outputs, demonstrating proactive data privacy management.

comsol-motor-nvh-evidence