cad
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes Python modeling scripts and the
cadgenutility to generate geometry, perform exports, and run validation checks. - [EXTERNAL_DOWNLOADS]: Fetches the
cadgentool (version 0.7.20) from the PyPI registry using theuvxpackage manager. This is a standard dependency for the skill's vendor-specific functionality. - [INDIRECT_PROMPT_INJECTION]: The skill has a potential attack surface through the processing of external CAD documents and user-provided sketches.
- Ingestion points: Files are ingested through
read_sceneandread_stepfunctions inreferences/inspection-and-validation.md; visual references like PNG sketches are used for modeling interpretation. - Boundary markers: The instructions in
references/cad-brief.mddirect the agent to record meaningful assumptions and clarify missing information rather than blindly following embedded data. - Capability inventory: The skill can write and execute Python code, manage files in project directories, and launch a local web server for model viewing via
cadgen viewer. - Sanitization: Parsing of 3D geometry is performed using the OpenCascade (OCP) kernel via the
build123dframework, which naturally validates structural integrity during the modeling process.
Audit Metadata